Two years ago I moved to Japan, working from home and hacking full time on what I would have used to call side projects in Finland (such as Facebook/iOS apps and recently a subscription site called Candy Japan). Many people have been curious about what it is like to live here, some having the dream of living in Japan themselves one day. A common question I get is "what does it cost?". This is an updated post based on criticisms and comments from a year ago, particularly adjusting such that this would reflect what a person living alone here might pay.

Current estimate for all personal expenses is about 1800 eur or $2300 / month. I live on a spouse visa in a small city called Tokushima, so things tend to be a lot cheaper here compared to Tokyo, where some parking spaces cost close to our current rent. Before moving here I was hoping to save a bit compared to Finland, but as yen got more expensive the cost ended up being similar. The change in exchange rate has really been crazy, one euro being 100 yen instead of 160 yen only a few years ago.

Our place

The whole rent is 714 eur ($480) / month. The rent we are paying here gets us a 70 square meter (750 square feet) place with shiny floors and a tatami mat room plus a shower that talks (it announces with a voice and melody when it is done filling the small furo-bath). For the properties on offer here, this was on the nicer side and is quite spacious for us. An apartment usually does not come with appliances initially. Initially it was a surprise to me to see people carrying away their fridges only to have the next tenant can carry in theirs, but of course where the line goes between what is part of the apartment is somewhat arbitrary.

Back as an exchange student living 30 minutes away from central Tokyo the rent was about the same, but for a cockroach-infested place with half as much space. We also had a singing neighbor and air conditioning that didn't really work. 

Other housing related expenses are gas for cooking, electricity mostly for running air conditioning, water and home insurance. These together are about 73 euro ($93). We did have to pay several months of extra "key money" to even get this apartment, which is thank-you money you traditionally must give to the landlord except in rare cases such as housing specifically aimed at foreigners. I recall we paid something in the order of 2000 eur initially to move in.

Food

The supermarket bill tends to come to about 1000 yen per day, so that would be about 265 euro ($350) per month. Depending on whether you prefer rice or imported rye bread (yum) for your breakfast this can vary. Müesli / cereal also costs a lot more here.

For eating out I spend about 150 eur ($191). Lunch food like ramen noodles or curry tend to cost around 6 eur ($8) a meal. Most splurgy meal I have ever had here was 40 eur ($51). It was a multi-course dinner in the most central hotel on the top floor. Cheapest thing is Udon noodles, 3.5 eur ($4.5) for a delicious (but not very filling) lunch, pictured below.

Communications

For our Internet connection we have fiber to the home. The speed fluctuates and can be double that shown on the screenshot. The monthly cost 46 eur ($59).

To open a 48 month binding contract at 42 eur / month ($54) for an iPhone5 recently cost 100 euro to start, but my own phone is an older model with a cheaper base fee. Be careful though with the expensive voice minutes if you decide to go with these seemingly cheaper plans. I thought I was being clever by using a local to international transfer service (Rebtel) before realizing that just the local part was costing more than the international part, and have since switched to only using Skype for longer calls. Another mobile option for foreigners is a prepaid b-mobile data plan.

Health

Travel

For the longest time we tried to use only public transport, but many of the interesting places just cannot be reached without a car here. Eventually we relented and shelled some 6000 euro ($7655) on a "kei"-type small car. If a car meets the kei definition, you get certain tax breaks. Basically the engine has to be 660cc and dimensions need to be under 1.48 x 3.4 meters (4.9 x 11.2 ft). 

Fuel costs 1.36 eur / liter which comes to $6.56 / gallon. Since I work from home, the only transport I end up needing is for weekend trips. This saves a huge amount. If you take longer trips road tolls can be shockingly expensive, one time we took a 2 hour trip to Kobe and ended up paying some 50 euro ($63.80) worth of tolls on it.

Many streets are extremely narrow here and so far we have not regretted having a small car of this type. 

It's been a while since I last posted an update on Candy Japan. Last time I got tons of useful feedback, so I thought it would be only fair to post another update to see how I am taking it into account.

Scalable content creation

Koichi suggested I start writing and making videos about Japanese candy. Ryanwaggoner suggested that I look into what patio11 has written about Bingo Card Creator, which in a way is also a suggestion to make more content.

In response to these suggestions I started individual pages for each candy. I am still missing the "scalable" part here, just testing things out at this point to get a feel for what that content might look like. It took Googlebot about a week to index my test pages. In the past week I received 13 visits directly attributable to those article pages. 

Found some people on oDesk who were interested in helping, some even native Japanese speakers. I am held back by the feeling that writing about each individual candy might be the wrong approach to take. Not many people search for specific odd candies and those that do probably want that particular one, not a subscription to random candies.

I feel more confident about videos, although haven't had time to make them recently. The top videos (not by me) on some specific candies are getting millions of views on YouTube. Having such hits might bring in customers just from the sheer volume of views.

Shipping suggestions

Giberson suggested I start distribution brances. I would like FBA to be my distribution system, and Moswald was helpful in responding to my questions about Fulfillment By Amazon (FBA), which is Amazon's dropshipping service.

I started looking into it and have understood that normally in the US when you receive an item which has customs duties associated with it, you just get a customs bill (unlike the insane system in Finland where you have to go clear each item in person) attached. Amazon fulfillment center will not accept a shipment with such an attached bill.

I have begun looking into how I could send a package to the Amazon fulfillment center so that the customs would already be paid for. One company that does something like this is Shapiro and they were recommended on the Fulfillment By Amazon site. I contacted them and received an epic application form as response (which seemed to be paper scanned into a PDF).

From the documents I received, I understood that with the US customs systems a "continuous surety bond" is needed. I had to google the meaning of that, but basically it means you pay some money in advance and if you fail to pay customs duties, then those duties are taken from that. Seems that I would need to put at least $550 into that bond, which at least is a reasonably low sum to give this a try.

Thanks also to Rain Republic for some helpful suggestions via email.

Buying suggestions

Dmerfield suggested I buy candy directly from manufacturers instead of from the supermarket. At this point I still feel it makes more sense to use the supermarket, as the time saving is more important than the perhaps 20% price saving I could get. If I start shipping in bulk to US, then of course this will become necessary as otherwise the numbers probably would not work out. Karl11 suggested I contact kitchen staff at tech companies to offer them a bunch of subscriptions. This would also require that I can import larger amounts to US.

Accepting credit cards

This was not particularly suggested by anyone, but I had a strong feeling that accepting credit cards directly on the site without redirecting to PayPal would increase conversions. For a non-US merchant like myself the options to accept cards is quite limited. PayPal has a product to do this, but it is not supported for non-US merchants. I was eventually able to find a company calledWireCard that together with Recurly could make this possible.

Thanks to everyone who made suggestions, keep them coming.

I've had to refuse money from people, because my system simply isn't up to the task yet.

It's kind of amazing how many small details there are even in a simple subscription site. First of all, you naturally need a way to accept a subscription payment. The easy way to start is to slap a PayPal payments button on your site.

Then after you have a few subscribers this way and are ready to graduate to the next step, you find that it is impossible. PayPal subscriptions created through the payment buttons are not compatible with the more advanced API they have. In other words, you cannot manipulate subscriptions through the API which you originally created through the easy buttons.

To avoid creating two different classes of subscriptions, one manipulatable through the API and others not, you keep the simple buttons and just try to get by with those. They are not quite as flexible, you have less control over the checkout steps and cannot easily ask things about the subscriptions later.

One thing you can do is query PayPal about the individual payments that are coming to your PayPal account. You can ask things like "list all the payments I received today" and "show me details about this payment". But you cannot ask things like "when is the next payment coming for this subscription".

But you can survive, as almost all of the same information can be reasoned from the individual transactions. There are some services that promise to make the task of running a subscription site easier, services like Recurly and Chargify. Not being a US citizen and using the simple buttons, those don't work for me however.

So now I'm stuck writing code to deal with the buttons created through payments standard. Here are some things that I need to do:

• Continuously poll PayPal so that I have a fast local copy of all transactions
• Additionally poll each transaction, because the default list doesn't contain information such as the buyer's paypal email address
• Also react to PayPal Instant Payment Notifications, because otherwise would not update fast enough and customers would have to wait around.
• Notice if payments stopped coming, so that product doesn't get sent to people who no longer pay. In some cases this is unknowable (product shipping out today, payment due tomorrow, cannot know if payment will actually come, but send anyway).
• Track which shipping address is related to which subscription. This means need user accounts -> account creation page, password reminder emails
• etc etc etc

Is there no service which would make life easier for us poor folks stuck using PayPal payments standard?

Candy Japan has scaled from sending just a few envelopes a month, to a few hundred and now nearly a thousand envelopes a month. As the volume has increased, some pain points became obvious. One of the most laborous is a silly little sticker you have to fill and attach on each envelope you send. The devil called the CN22 form.

On each CN22 form you have to fill the contents of the package, then sign it and write the date for the sending day. Not a big deal when dealing with just a few, but when you get 400 of them in front of you and have to sit down and spend the next two hours writing by hand the same information over and over again, it starts to feel like a huge waste of time that would better be spent... well doing almost anything else.

The first thought was of course to just print them out. I asked the post office if it would be cool to just print this information directly on the envelope. Not cool. Instead I got extra hands to help me write them out. That didn't continue long before I was again feeling very silly about paying someone to be a human printer.

So I went to another post office and asked the same question again. "Yeah sure go ahead, doesn't matter if it's printed" was the reply from a busy postal office worker. This answer totally contradicting what I heard before, I wanted to make sure I wouldn't get my packages rejected, I started reading the Japanese regulations concerning CN22 forms and discovered that there is a special application form you have to fill out to apply for the privilege of printing these forms.

I filled that out and went to a post office to submit my application. It was pretty clear this was the first time they had dealt with anyone applying for it. Phone calls ensued as they started checking what to do about it. Eventually they refused to accept my application, with the excuse that only big cities like Osaka or Tokyo accept CN22 printed directly on the envelope. I see.

By this time volume had gotten large enough that I was getting postal office workers coming to my door to collect packages. These were people that I hadn't met before, so I decided to test if they would agree to accept my request. The answer was no.

In addition to filling these form stickers, another problem was getting enough of them. Even the central post office would often have only a few dozen of them available at the time. Ordering them by phone I was sometimes able to get a few hundred at a time, but never any reasonable number that would last me more than a few weeks. This would lead to situations where I had to send out packages, but wouldn't have enough of the forms.

My wife volunteered to go on an errand to collect more of them from the Tokushima central post office. While there she again asked if it would be alright to just print them on the envelope. This time she got one of those rarer "yes" answers, totally contradicting again what I had heard the last few times I asked about it.

The next time I had a CN22 shortage I decided to just go for it and see what happens. I prepared 300 packages with a CN22 replica authored by yours truly in Gimp printed on them. As I requested them to be collected, I was expecting a slight chance my whole shipment would get rejected. The result?

Silence. Nothing. No complaints, no rejected packages.

Having product images increases sales, but post-processing is a chore which involves clicking around in Photoshop/Gimp to isolate the product and changing levels such that the background becomes pure white. As I've been selling Japanese candy online, taking shots like these has become a repeating task.

You could pay someone to do that, but wouldn't it be nice if it could happen automatically? Today I challenged myself to see if I could manage it. After a few hours of hacking it works quite well, actually the above image is output from the script.

One side effect is that because touching up the product picture is now free, you could even do it hundreds of times just for fun. For example for no particular reason, you could take a hundred pictures at different angles and then isolate the product on each frame to get a nice animation. Not that anyone would be crazy enough to do such a thing.

Tech used is just Python with the Image and ImageFilter modules. To put together the gif animation I used ffmpeg. How to do the isolating part then? The solution actually turns out to be very straightforward, just finding adjacent pixels based on color. Read on if you care for the details.

Step 1:

Convert the image to black and white such that the background becomes solid color that surrounds the object. Now the thing we want to isolate is the blob in the middle. If we had an outline for that, then that can be used to pick out the object from the original photo.

Step 2:

Find one pixel which is known to belong to the object that we want to isolate. I pick the first darkish pixel I can find near the center of the image. Then starting from that pixel, recursively explore neighboring pixels. If you get unlucky the algorithm will fail here if it happens to start from an unconnected island, so in reality I had to use multiple start points. Below is what you get after counting all neighboring pixels to be part of the object. Note we have now isolated it from the background, but there's still gaps inside.

Last July I posted on Hacker News about an experiment to start a Japanese candy subscription service. I live in Japan, so the idea was to send surprise candy stuffed into envelopes twice a month to subscribers directly from here. How did the experiment turn out?

How things started

I am from Finland, but this idea started in Thailand. I was on holiday with a friend, partly inspired by the idea of working online while traveling. Things were going really well for me, I was running a MySpace app (yes, they have a platform) that was generating some revenue and working on that for a few hours each day and enjoying the beach and sun the rest of the time.

We didn't really have the time to start something new, but just for fun we were throwing around site / service ideas. There were many, but one thing I wanted to experiment with was some kind of a subscription service. Originally I thought about traveling around the world and keeping a blog of the places I go, then try to support that by offering to send subscribers souvenirs from each place I visit.

First steps

I never started doing exactly that, but that idea morphed into something new after I settled in Japan. As MySpace died I suddenly had free time and a need to start something new. So now was the time to finally try my hand at a subscription service. I needed to find something local here that I could keep sending. Something consumable, so that unnecessary items wouldn't pile up for the recipient. I enjoy trying out local soft drinks and food, but as I couldn't really send those, I chose candy instead.

To get the service started and to be sure someone would actually pay for it, I just emailed a few people asking if they would pay for such a service. A few said they would, so I created PayPal subscription buttons and sent it to them. I did no API integration, just used the "website payments standard" button generators to get a link that would let the interested people start a subscription by just clicking over from the email they received.

For the first few shipments I had only a handful of subscribers. After the subscriptions seemed to be working properly, I posted about the service on Hacker News. That really kickstarted things and resulted in great initial growth. Some bloggers discovered the service through HN, mentioned it on their sites and that fueled the growth even more. Since those were Japan-related blogs, the traffic from them was very targeted and lead to many subscribers.

Really getting going

The subscriber count was now in the hundreds and we were really scrambling to fulfill all of the orders. I enlisted my wife and some friends to help with the packing efforts and together we managed to do it. Preparing a shipment was an undertaking that took all day from morning to late at night. There is a lot of manual labor involved.

After each shipment things were getting better however. I bought some equipment to print out labels automatically, then later figured out further ways to cut down the time spent packing by hours. The amount of packages was felt even at the local post office, which hardly had any international packages sent, then suddenly they were seeing hundreds. We started to clog up their service, so at one point the postal office boss introduced me to a service which picks up the packages directly from home.

So now we no longer had to lug the packages to the post office, which saved a ton of effort and waiting in line. But still we were moving boxes of candy from the supermarket to our 3rd floor apartment without an elevator. Since I had been buying a lot of candy very regularly from the supermarket (I represent about 50% of their candy sales), my relationship with the manager had become good enough that I could ask him for help. He agreed to try if they could just do the packaging inside the supermarket, so that we wouldn't need to move the boxes home.

Since I now knew very well the effort involved in making a package, I knew how much time it would take them to do it for me. We agreed on a flat per package packing fee. The initial trial run went well, and they have been making packages for me ever since. I just show up twice a month there with the address labels and other materials I have prepared and they take care of the rest. Post office picks up the packages directly from the supermarket. This has almost turned into a drop-shipping operation.

Decline and uncertain future

The subscriber count peaked at 300, but started to go down after that. The reason? There is always churn in a subscriber business, and unless you can continue to bring in traffic to make up for the churn, there will be a decline.

In my case I enjoyed an immense boost of high quality initial traffic. Those initial subscribers are now churning out and I haven't yet found a way to make up for the lost customers. The only thing helping is great Google search result positioning for "japanese candy" and "candy japan" that is still bringing in a trickle of good quality traffic. 

That alone has not been enough though. The subscriber count is now at 277 and still slowly declining. I briefly experimented with paid advertising, trying Reddit ads etc., but traffic seems to be just too expensive to buy for a service with such a low profit margin as this (it cost about $60 to get a conversion through the channels I tried).

I really hope to find new sources of traffic and to find ways to improve the experience for my existing customers to reduce churn and reverse the trend. The service seems to be something that many people enjoy, and the process of sending the candy is now quite well thought out, so I hope to be able to continue doing this for a long time. 

Thanks for reading. Comments are welcome here or on Hacker News. The website for the service is http://www.candyjapan.com. Email me bemmu@candyjapan.com if you have any questions. 

While doing split testing for Candy Japan I started thinking if there could be a better way to decide which page variation to show for each visit. The way A/B testing is most commonly done is to run a test for a while, showing one variation to half of the users and another variation to the rest, don't look at the results while the test is running, after some pre-decided time stop the test and then start always using the best variation.

After I started buying advertising and actually paying for each visit, it started to interest me whether this is really the best way to do things. The way I started thinking about it was imagining being at the races with two horses, one of which I could enter in the race and my only decision being which to pick for the next round, with a lot of money at stake. In that situation I would very carefully think each time which one to use. I wouldn't just arbitrarily give each a chance at a thousand races and then pick the better one.

This part is about horses

This imaginary horse racing is one where both horses have a predetermined winning ratio, but I can only know it by experimenting. In the beginning I would have no reason to prefer one horse over the other, so I would let both race once and might then have the following situation.

Horse A: Win

Horse B: Win

Hmm... they both won and have a 100% winning ratio. I didn't really learn anything there. Let's let both race once more and add that result to the end.

Horse A: Win, Lose (50% wins)

Horse B: Win, Win (100% wins)

Now even though I don't have a lot of data, surely this means there is a slightly better chance that Horse B would actually be the better choice for all my future horsing needs. I should still give Horse A the benefit of the doubt, it might have just happened to be unlucky. To reflect that, perhaps from now on I will still use Horse A as well, but will start using Horse B a bit more often. 

If I did start using only Horse B from now on, what would need to happen for me to change my mind? Well, if Horse B lost twice in a row, then its winning ratio would also drop to 50% and I would no longer prefer it over Horse A. Now comes the meat of this post.

What is the chance that Horse B would lose twice in a row? If I knew that, then I could start using that to indicate how often I should use that horse. I would need to know what the winning ratios for these horses are to know that, which of course at this point I can't really know, but I can guess based on the horses I have owned in the past. They have all been superstars and have won 50% of the time, so I will suppose that 50% is a good guess for a winning ratio.

For Horse B the chance that it would lose once is then 50% and the chance that it would lose twice is 50% * 50% = 25%. So there is a quarter chance that just with random fluctuation Horse A would catch up. My wild guess: maybe it would make sense to start using Horse A just 25% of the time from now on to reflect the chance it has for catching up? I could then continue adjusting that as more wins and losses come in, always using the losing horse as often as its chance of catching up indicates.

Would this work? Not wanting to think too hard, I did what any programmer would do. I wrote a simulator. 

Split testing split testing engines

At this point I ran out of my allotment for the word "horse" and returned to reality. My simulator isn't about horses, it's about an imaginary website for a startup company. The website gets visits and has a roughly known conversion ratio. The website owner has thought of some split tests, but has to decide when to turn each on or off. For example the site owner could show customer testimonials on the site, which would boost or hurt the conversion ratio by x%. 

In my simulation the initial conversion ratio is 5% and the value of each conversion is $39.90. The time span for my simulation is always 10000 visits, so you might expect the total revenue to be around $19950.

The simulator has modules for each split testing algorithm, which are used to decide which split tests to use for each visits. They don't have access to any information that I wouldn't have in reality, they just get a visit and return which split tests to use. The main loop of the sim then decides whether this time there happened to be a conversion or not.

At first I wanted to just try the simplest possible case as a sanity check: not using any of the split tests at all. I ran the sim 5000 times and took the average revenue over all of the runs. The result was $19939.70, so things seemed to be working as expected. I also tried just serving with random split tests turned on or off, and that also performed similarly.

Next I wanted to try the algo I had been using up to this point. Run the split test for a while, then decide the winner and only use that one in the future. Running the split tests for 500 visits and then choosing the better one resulted in avg. revenue of $23358.48 (I also tried running the tests for 1000 or 2000 visits before deciding, but those did not perform as well).

Horse.

I saved the most interesting test for last. How would my horse algo do? The average after 5000 runs was... $24383.06. Because the bad A/B tests were abandoned earlier and the good ones were used more often earlier, in this little simulated world this way of running tests made an avg. of $1024.58 more revenue.

If you would like to try running it yourself, you can find the Python code for the sim here.

Three months ago I started a Japanese candy subscription service called Candy Japan. It is growing well from word of mouth alone, but I was curious to see if it could grow even more through ads. My friend Nick who runs the site Meme Stickers suggested experimenting with Reddit as a possible source of traffic that could convert for me. In this post I will show how I set up ads there and what the results were.

Reddit Self-Serve Ads

The Reddit self-serve ad system is a bit different from Google AdSense. Instead of bidding for clicks or conversions, you decide on a bid for a certain dayspan in the future. If there happen to be multiple bidders on those days, you get your share of ad views based on how much you bid. There is a minimum bid which depends on whether you target all of Reddit ($20/day) or a certain subreddit ($30/day) to show your ads to people with specific interests.

I wanted to target the "snackexchange" and "japan" subreddits. The minimum bid is $30 per day for subreddit targeting, which seemed very high to me considering that the two subreddits I wanted to experiment with were quite small. Still I decided to go ahead as discovering a profitable ad channel could be worth risking some money on (another part of me fears that the whole of online advertising relies on naive people like me).

Setting up

I went ahead with setting up the ads. After confirming my email address it was straightforward. Here is a screenshot of basically the only screen you need to interact with when setting up the ads.

I did run into a problem though, which was that I am a Finnish guy living in Japan and Reddit only allows paying with a US / UK / Canadian credit card. I went to complain about this on the IRC channel #startups on Freenode. User EvRide graciously stepped up to help me. I sent him the money over PayPal and he then set up the ads for me. Thanks a lot EvRide!

Results

The ads cost me $240 to run. I had 63148 pageviews, 419 clicks and 4 subscriptions = $60 / gained customer.

The cost per conversion was too high to continue advertising on Reddit for the current service. I need to improve the ad or get my landing page to convert better before another attempt. Still, it was encouraging that some customers did find the service through Reddit and it would be an interesting channel if the minimum bids were lifted and international credit cards were accepted.

More to read about Reddit ads

http://www.winningtheweb.com/reddit-homepage-advertising-results.php

http://unethicalblogger.com/2010/11/08/experimenting-with-reddits-self-serve-ads.html

http://swappa.blogspot.com/2011/01/reddit-advertising-results-first.html

http://blog.kissmetrics.com/reddit-marketing-guide/

http://www.gabrielweinberg.com/blog/2010/03/my-duck-duck-go-reddit-ad-by-the-numbers.html

As you may already know, Candy Japan is a subscription service for random Japanese sweets that we run from our living room in Tokushima, Japan. This is a brief report and numbers for August 2011.

Growth

The real growth line is the middle one, it shows the number of subscriptions that were active each day during the month. The top blue line is how many subscriptions had been started and the red line is the number of cancelled subscriptions.

Although it makes this chart a bit harder to look at, I decided to include the cancellations here to make sure that I have an eye on members churning out. If the red and blue lines ever meet, it would mean I had lost all of my subscribers. So far things are looking OK, as the majority of cancellations are because people just wanted to try the service for a month, not because they were disappointed with it.

Unlike Kreci's reports, I'm not mentioning straight revenue numbers here. I feel that would be quite misleading, as it would look much higher than reality because many members have already paid for things that I haven't sent them yet. But my estimated profit per subscriber is 3 euro / month, so you can figure it out from there.

Now onto stuff I learned this month.

Canadians are super nice, but their customs are strict

When you ship a package and there is something wrong with the address or labeling, how serious is it? That package will need to be re-shipped. If your margin is say 20%, that means you need to sell 4 items to get back to where you were supposed to be in terms of profit, just because you made one mistake.

And that brings me to another thing I learned, that when you ship things to Canada, the return address cannot be only in Japanese or they just send them back. There are now members from all around the world, including Saudi Arabia, but no other country has been this strict about the sender address. Positively, another thing I learned is that Canadians are incredibly forgiving and understanding. Everyone I emailed about it forgave me for my mistake.

To make sure everyone got what they paid for, I reshipped the past two shipments to all 18 Canadian members.

Paperwork can be automated

But only if you live in a big city. Every single package needs a customs declaration sticker attached to it, along with my signature, date and description of contents. It gets a bit dull after you have written the same thing 300 times. I applied for permission to print the form directly on the envelope. Denied. Apparently they can only accept such pre-printed forms in Osaka or Tokyo.

Editing videos is easy

But doing voiceovers is hard when you are not a native speaker. We shipped candy which resembled the Japanese food "takoyaki", so we made a video to explain what takoyaki is.

Making the linked video only took about 3 hours to shoot and edit, which is a bit more work but comparable to writing a blog post like this. Still these videos are a nice way to explain to the members what the shipped candy is about. Since some assembly is required in many of them, without a video it would just be too hard to explain.

Although the videos are getting some views, those do not seem to be bringing new converting traffic to the website yet.

Advertising is hard

Getting mentioned on blogs is the best source of traffic I know of, but not very dependable. In order to find more predictable sources of traffic, I looked into advertising. Currently I am trying StumbleUpon, Reddit Self-Serve Ads, AdWords, Facebook Ads and Project Wonderful. So far none of them are turning visitors into customers well enough to make it clear that they would be worth continuing. 

Reddit looks promising and will be the topic of my next post. See you then and as always I welcome all comments.

I recently listed my first item just to see how difficult the process is. About 2 hours of learning about the service and 2 weeks of waiting for shipping & receiving later, I have my test product up.

You probably know about Amazon Marketplace, which is a place for individuals to list their used or new copies of Amazon items for sale. But did you know that listing an item for sale on the "main side" of Amazon is not that difficult either, even if that product is not even listed by Amazon yet? This way when a user presses the big shiny "add to cart" button on Amazon, you could to be the default provider of that item.

Two paths to Amazon

What I learned is that you have two options. You can either let Amazon accept the orders and then notify you with the buyer's address for you to ship to when a sale is made, or you can let Amazon handle the shipping as well, using a service is called "Fulfillment by Amazon" or FBA.

If you have ever handled shipping small amounts of items, you can imagine how liberating FBA could be. You could send thousands of products for Amazon to store and be lying on a beach while your items are bought and shipped, without you having to do anything more. From your point of view, this virtualizes your physical inventory. They even have an API so you can dispatch items programmatically. It might not even be that different in cost, as Amazon has massive scale benefits. Lured by this idea, I decided to investigate using FBA.

Signing up

To sign up I only needed to provide my credit card number (there are some handling and storage fees), address and confirm my account by entering a pin code over an automated phone system. After that I was surprised to already find myself in the "seller central" where all the action involving listing products and shipping happens. To actually receive my proceeds, I will need a US bank account number, but I decided to worry about that later if I actually start making some sales.

Barcodes

One big hurdle for many might be that the products need to have a specific kind of barcode called an EAN / UPC code. If you make small amounts of items yourself, for example I have a friend who makes Meme Stickers, this might be a show stopper for him as getting the code might cost more than the profit he can make. Such a code seems to cost $750 + §150 / year to get several batches of your own codes, or $100 to buy just a single code for one product.

Of course you would need a way to print and attach these barcodes too. However if you are reselling a product made by a bigger manufacturer like I am doing, they should already have such a barcode. Note I am not absolutely 100% sure that no item could be listed without such a code, but did not investigate fully as my item happened to have that lucky code already printed on it.

Sending your items to Amazon

To get your items into FBA, you create a shipment in seller central. You need to list each item that will be inside the shipment and then print a shipping label and a packing slip for the boxes you send so that Amazon can identify the items when they receive them. Additionally is some cases even each item can require a label, unless it belongs to the mysterious class of items called stickerless commingled, in which case they can be put into the box without any labels.

Completing these steps actually did not take long, and soon I had my item ready for sending to Amazon. I sent my item on Aug 10th. As I was sending from Japan, the shipping itself took a bit over a week. After that I started getting notification emails from Amazon. The first one to break the tension was August 18th at 8:36 AM and had the subject "FBA Inbound Shipment Checked-in". I thought that now the item would be up for sale, but it took a few more days of waiting before that actually happened. The next event was same day at 3:32 PM "FBA Inbound Shipment Receiving".

On Aug 19th 1:32 AM I got the message "FBA Inbound Shipment Received In-Full". I thought this would be the end, but checking in seller central I could see my item was now in a "reserved" state. Reading their help I found this can mean many things, such as a customer holding on to the item as their order is about to go through, but in my case it probably meant that Amazon was doing further checks on my item.

Arrival at promised land

Then finally, on Aug 22nd I could see that my item was now up for sale and listed as taking 0.02 cubic feet of space in their shipping center. Now I am waiting to see what happens when someone actually purchases it and already making plans to start using this as my way to export more things from Japan.

Thanks for reading and please comment if you have any of your own experiences to share!

 

As you might already know, I run a service called Candy Japan which mails subscribers Japanese sweets twice a month. It has been mentioned in several blogs and traffic from those blogs has converted into paying customers at roughly 1% conversion rate. I suspected that blog mentions would probably work better than ads, but wanted to find out for sure.

Project Wonderful

Wanting to try something new instead of the same old AdSense, I gave Project Wonderful a try. It is an ad network popular especially among web comic authors and is more transparent in pricing than most. Instead of paying per click or per conversion, you pay "per day". What this means is that you can buy an ad slot on a site and no other ads are shown there while you hold it.

There is an auction system in place to decide the price of each slot. The prices vary from zero to almost a hundred dollars per day. You can search for slots based on several criteria. I decided to go with a mix of sites that had "japan" in the tags and then just some really popular ones to see if my service would have any appeal to a wider audience.

To get started I transferred money to the ad network and created several ads since almost all sites required a differently sized box. 

Tracking conversions by channel

Initially I wanted to use Google Analytics conversion tracking, but found that many people do not click the "return to merchant" link after buying from PayPal, causing many conversions to go unreported. I discovered that rolling your own system is not at all difficult.

To know which ad resulted in a conversion, I set a cookie when a user comes in after clicking an ad. If the user just came in from an ad or has such a cookie, I pass it to PayPal so I can associate the conversion with that ad when I get an instant payment notification.

I learned from a talk given by Daniel James that when buying ads, you should not just track them as one lump but instead split by source of traffic. Each such source is called a "channel". In hindsight this is very obvious. Without knowing which channels work and which do not, you would not know what to keep doing and what to stop.

The results

Stop everything.

The rightmost column is named "visits", but really means landing page views. The "convs" column is the conversions. Leftmost column indicates the source of the traffic. For example c_mspaint refers to MS Paint Adventures.

I realized these sites would probably not convert well, but over 1000 visits and 0 conversions is still disappointing. Granted MS Paint Adventures and Pokéfarm are probably terribly converting sites, but still over 3 million ad impressions with 1000 people deciding that the service is interesting enough to click on, I would expect a conversion.

Conclusion

These ads only cost me $50 to run and might have been profitable if I had just two people sign up and remain signed up for a full year. There are thousands of possible traffic channels. This experiment really only tells me that these particular sites do not convert well for me.

My Japanese candy subscription service Candy Japan has been steadily growing. While the club is still small and sending the envelopes to subscribers does not take more than 2 days per month, it is taking a lot of mental space. I have a bit of anxiety about how to take it forward.

I had been thinking about starting a gift club already for some time. My first spreadsheet with "back of the envelope" -style calculations is from 2008. Back then the idea was "A gift-of-the-month subscription service delivering neatly packaged items from Asia, with a clearly written description of the significance of each month's item".

Back then I was coasting on my MySpace app which was doing well, so I didn't actually start working on the club until the app platform completely collapsed around end of last year, right after I had moved from Finland to Japan again. It was the right time to try something with Japanese products, as I have easy access to them now. 

I had some doubts about going into physical goods again.

There are many repetitive tasks and annoying details such as VAT / customs paperwork to deal with that you cannot really assign to anyone else until you get bigger volume. With my previous online store dealing with Japanese books I was never able to achieve that volume. Plateauing at a just below life sustaining profit was very frustrating. It took a long time to recognize that the bookstore was going nowhere and shutting it down, which turned out to be a great decision as it freed up my time to write the MySpace app, which did vastly better.

This time things look brighter. I already have almost the same amount of customers in two months as I had for the bookstore after two years.

What has been happening lately with the club then, is it all roses? Not completely, here are some things that are causing me anxiety.

1) I worry about getting shut down by PayPal because of the lack of package tracking

PayPal requires you to keep documentation of every package sent. At some random point they can shut you down and then demand to see this documentation. Am I collecting this tracking information in anticipation then? No.

Why not?

To get this tracking information, I would need to pay Japan Post $5 / package. Since my profit per package is only $2, this is impossible. That $5 does not even buy tracking that works in all countries. For assured tracking, the next level is sending by EMS which costs about $20, completely impossible with the current size of envelopes.

2) Shipping directly from Japan feels wrong

I have some curiosity about setting up shipping better to make sure that items arrive as they should. Currently I have only the feedback of subscribers to go by to discover what really happened with the packages I sent. So far I have not received any negative comments and the feedback I did receive were confirmations that packages had arrived safely and intact.

What I dream of is having local hubs where I could send larger packages, with customs cleared.

At this point in time the extra costs from this will almost certainly not be worth it, but out of curiosity I have been looking into it. It seems like a useful thing to know in case I want to start exporting something else too.

So the idea would be to first export the candy to US, then send it forward from there. Amazon now provides a service called Fulfillment by Amazon. You send Amazon your items, they put it in their warehouse and you can then later dispatch it using API calls. Seems great.

The downside is that package branding will be Amazon branding and customers might feel odd about subscribing to an exclusive Japanese club, only to get packages from Amazon instead. But really that would be better for everyone and seems like "the right way" to do things. As an additional bonus, I could list my items for sale on Amazon.com, too. 

I have sent FBA my first test item to see if I can get it to appear in their system.

3) Worry about trouble with FDA because of lack of direct manufacturer connection

While USA is less than half of my subscribers, playing nice with Food & Drug Administration still concerns me. As it is I am sending the envelopes as "gift candy assortments". There is an exception in FDA rules which says that an assorted gift package from an individual to another individual should be let into the country without problems.

If I do not indicate that it is a "gift", then I have to show that the food is safe for consumption in the US. Of course it is, as it is factory sealed and made by one of the largest candy manufacturers in Japan, but I still need to prove that to be true.

I can agree with the idea of checks like these, as they keep food safe and give consumers confidence that they can trust the food they buy. All I need to comply is to send a prior notice to the FDA upon importing food to the US and include the "food facility code" of the factory that made the food. But this is where I am stuck. These facility codes are not available online. To start with I wanted to send candy manufactured by a big company called Meiji. I know this food is safe, as Amazon is already selling Meiji products in the US.

But where can I find the facility code? There is no online directory of food factory FDA codes. I have no personal contacts at Meiji. I have twice attempted to contact them through their web form. Both times I did get a reply, which was "sorry, we can't help you" (「アメリカ食品医薬品局の登録番号につきましてはご回答できませんことをお詫び申し上げます」). 

On the FDA form it seems that this code is optional, but reading other parts of their site it seems that not having this code is a no-go. I have sent email to them to ask if they have any suggestions on what to do. One thing I could do to easily solve this is to only send candy which has already been imported to the US by someone else, if I could find suitable partners. My volume might be too low now to get anyone seriously interested, though.

Until next time

Well, these are the things I have been thinking a lot about lately. To end on a positive note, I was able to negotiate around a 7% discount with the local supermarket for buying candy in bulk. I also managed to find someone to print envelopes for me that already have my address on them, which saves me time and money on address labels. Thanks for reading and please do share if you have ideas on how to solve my anxiety points for Candy Japan.

This is just a small update to show that Candy Japan is alive and well. There are now 180 subscribers. My ramen-profitability line of 300 subscribers is fast approaching.

In my previous posts I revealed the initial price I came up with, 16.50 EUR. That is about 23.95 USD / month for this club. For this price the subscribers get two envelopes of candy each month. Why this price?

Of course I realize that my costs have almost nothing at all to do with what the right price is, but just to test the viability of the idea I settled on starting with cost plus pricing.

It is based on the following estimated costs:

• 6.84 EUR for shipping, 400 yen twice a month.
• 3.42 EUR for the actual candy. Yes, it disappoints me that this is not a larger part. But on the other hand more spent on this would mean it could not fit in the envelopes.
• 0.91 EUR for PayPal fee.
• 0.80 EUR for packaging materials, including some leeway to allow for fancier custom packages in the future.
• 0.74 EUR for hiring part-time help to make the packages. I guessed it would take 2 minutes to create each bi-monthly package. It would cost about this much to hire a starving student at a typical 1000 JPY / hour to do it.
• 0.40 EUR similarly for the time spent responding to customer support emails.

Total estimated costs 13.11 EUR / month. This leaves a pre-tax profit of 3.39 EUR / month per subscriber if I calculate it Groupon-style, totally ignoring marketing costs. Later on I plan on starting to test different price points. I already started A/B testing the landing page for layout modifications, the price will be a similar test.

The marketing costs are nebulous. They might turn out to eat up all the profit.

I am not tracking this accurately yet, but with 180 conversions from 15800 visits the conversion rate is about 1.13%. All of this traffic is from various blogs that already explain the idea of the club, so probably most of the people visiting were already considering to buy. This means some other channels might do a lot worse.

As a starting point with 88 visits required for a conversion, at 0.20 cents per click I could get a customer for 17.60 euro. If they subscribe for at least 6 months I would be at break-even. This is pure fiction of course since I have no idea what the conversion rate for AdWords or other paid channels would be, what a click would end up costing or even what my average subscription length will be.

It might be that reaching out to bloggers to cover the service is a better way to grow, but can that be sustained over a long time?

Candy Japan is a subscription service for Japanese sweets. In this blog I have been writing openly about the business side of it.

I thought the subscriber growth was nice the last time I posted, but now it is exceeding my expectations. Some bloggers graciously mentioned the service and the story is spreading from blog to blog. Here is the current chart of subscribers.

Where are all these subscribers coming from?

That big spike in the end is mostly thanks to crackajack.de and The Daily What. More and more blogs are mentioning the site and there are also a lot of people sharing it on Facebook. In my previous post I annotated the mentions I got before. Once you get the process going, it is not necessary to contact each author separately, they will read each other's publications and report the same story. I have had this experience before and have heard others like James Hong of HOTorNOT report on this spreading pattern.

As the story about the club is moving directly between blogs, it has even mutated a bit. Some stories are saying that I have moved from Japan, which is not true, I still live in Tokushima. Not that it matters, but it is interesting to think what happens to a message as it is passed on and slightly modified each time.

So far things have proceeded pretty smoothly. I have reports from several subscribers that their packages have arrived intact and without problems. I did have a small problem with an address being printed wrong, because my script was ignoring the second address line. I made that mistake up to the subscriber by sending them a bigger package of candy using a tracked quick shipping method (EMS) as an apology.

Will I be able to handle these orders? Yes, I believe I would be able to handle almost any amount, as I have reserved a sufficient margin to pay for packing help if necessary. It took only 2 hours to mail the ~50 orders I sent out last time. This time it looks like I will be sending out perhaps 150, which should be possible to do in one day even just by myself. To make the task a bit easier, I ordered 1000 pre-printed envelopes from a company in Osaka.

I am incredibly grateful to everyone who has mentioned the service on their sites and at the same time curious to see what happens next. Will I get unlucky and the story will die and be forgotten before being passed on to the next place, or will it continue to spread even further?

Two months ago I started a club which sends random surprise candy directly from Japan to subscribers paying on a monthly basis. The club is called Candy Japan and on this blog I try to describe the business side of it as openly as possible.

Here are my subscriber numbers from the start to today.

This chart only includes people who are paying me via PayPal, which includes almost all. In the beginning the chart goes from 0 subscribers to 3 subscribers after I contacted some customers I knew from before. At that point the club did not even have a home page.

(1) I finally created a home page for the club and posted about it on Hacker News on July 9th in a post titled "Experiment: Japanese candy subscription service". That directly lead to 5 people subscribing, and a few more in the following days.

(2) The Japan blogs Tofugu and Japanator decide to make my day by mentioning the club on July 14th. 8 people immediately subscribe on the day of mentioning and many more in the following days.

(3) My second post on Hacker News, "Getting my first subscribers". 4 people subscribe right away, more in the following days.

(4) A bunch of people including @superamit and @letmebegreat mention the club on Twitter, bringing the subscriber count to the current 52. I believe some of these tweets were because I emailed the subscribers to let them know their candy was in the mail, and in the end I included a short plea to mention the club on their blogs.

Thanks to everyone for your support and encouragement.

Candy Japan is a subscription service delivering bi-monthly envelopes of random Japanese sweets. In this post I am going into detail on the costs of running it. My information is still limited, having run the club for about a month, but I will attempt to break it down as well as I can at this point. 

Why would I reveal these things? I want to contribute something back after enjoying the posts of Kalzumeus on his blog about Bingo Card Creator and KreCi's monthly revenue posts. Revealing my numbers might also get me valuable feedback that will help me improve the club. It also turns out that when you write interesting content, other bloggers read them and mention you on their blogs. Getting the message out there is useful for getting members for the club.

Variable costs

These are the costs for each additional subscriber. Monthly subscription cost for the club 16.50 EUR or 23.95 USD. At the current exchange rates that averages out to 16.73 EUR. This gets the subscriber two envelopes per month. I wish I could write that most of the subscription price is my profit, but unfortunately almost all of it I am expecting to go into expenses. Some of these are guesstimates at this point.

#1 Shipping 6.54 EUR

Shipping has so far been averaging 6.54 EUR / subscriber / month and is my biggest cost. For the previous envelopes I sent, shipping was 330 JPY by Airmail, taking 4 days to arrive, depending on country. Next cheapest option is "SAL" at 280 JPY and takes "2-3 weeks". The absolute cheapest one takes 2-3 months. A small hack I have considered to get cheaper postage and minimize trips to the post office would be to combine the use of Airmail and SAL. Instead of going to the post office every 2 weeks, I could just go once and still get the packages to arrive roughly 2 weeks apart. Send the first package using airmail, but then immediately also send the next package using SAL. Airmail is delivered in 4 days, while economy airmail is "2-3 weeks", so the average difference would then be about the required 2 weeks. 

This would save a lot of time and about 7% on shipping, but I am not doing this yet. I worry though that the post office might actually deliver the economy packages faster than they say, causing deliveries to get clumped together.

#2 Product 3.50 EUR

This might not seem enough to fill two envelopes with candy per month, but so far it seems that it is. As an example, in the next shipment I plan on sending hamburger candy sets. I think it is a really fun and tasty product, but regardless the cost here is only 148 JPY (1.33 EUR). At first I thought I would need to continue buying these items from the supermarket like everyone else until I reach some massive scale, but already when ordering 50 pieces I was able to use a provider I found on Rakuten that could ship them to me for only 120 JPY / piece. I am looking forward to reaching a bigger scale to save even more.

I think this is a great product to include, and I hope you can agree I am not cutting any corners here on the product side, even if the cost is cheap. If I had a larger budget, I would probably look into getting regional product gift packs, but those would not fit in a normal envelope and would be more complicated for the customers to receive, probably requiring trips to the post office to receive in some countries.

#3 Handling 1.10 EUR

This is the cost of labor for attaching address labels, filling in the customs note for each envelope, sealing the envelope with tape, attaching the correct amount of stamps and then dropping the package in the mailbox or lining up at the post office to get it sent.

Even though I did not count my own blogging / coding labor as a cost, this one I will as it is much easier to calculate and could easily be done by someone else. Every student in Japan tends to have a part-time job or "baito" .The typical rates for part-timers are between 800-1000 JPY / hour (7.20 - 9.00 EUR). I assume there will be some side costs, so I am guessing in reality it would be closer to 11 EUR. Allowing 5 minutes per subscriber comes to about 1.10 EUR.

#4 Paypal fee 0.90 EUR

This is what Paypal is charging me to handle each payment. Might be reduced in the far future by using some other provider, but not being US based limits my options a bit. 

#5 Customer support 0.80 EUR

This is a really wild guess, based on imagining a future scenario where I would have 3000 subscribers. At that point someone might need to spend all their time supporting them. It might seem like that would not be possible for one person to handle, but even if every subscriber sent a mail every month, that would still average to 100 per day and seems possible. It does not seem now that the amount of mail would be that high.

#6 Packing materials 0.56 EUR

An envelope costs about 0.05 EUR and sealing tape is almost free at around 0.01 EUR per envelope. Customs declaration stickers are free. I use two address labels, one for my return address and one for the recipient address, so four per month total. These cost about 0.11 EUR each, expensive as I am currently getting cartridges for a Brother label printer. Each cartridge costs about 8 euro and can make about 70 labels. If I did more comparison shopping on the cartridges, I could probably halve the price, so a lot of room here to improve.

- Profit 3.33 EUR

Total costs added up to 13.40 EUR., comparing with the subscription revenue of 16.73 EUR, we get 3.33 EUR as result. The fun question then is, where should this 3.33 EUR go. This is actually the very point at which a lifestyle business and a swing for the fences -style ideology would start differing. At first I will not mind putting everything back into the business. In fact I bought the label printer using money I had gotten from the previous envelopes where I had written the addresses by hand. The profit would go completely back into the business at first, when clear ways to do that would be available. I would like to use more effort to do guest blogging, since I feel it is out of the options related to marketing, least bad. When it works well, it can even be a win win situation. 

In my next posts I'm planning on talking about my attempts to support Bitcoin. I will also write about the possibility of open sourcing this whole effort so that others with similar projects could find some snippets of code that could be useful for them. I also would talk more about my ideas on churn, ARPU and those things and how they relate to this candy business. If you would like to see me focus on something else, please suggest a topic in the comments.

Until next time, thanks for reading.

Candy Japan is a subscription service where for a monthly fee you get a surprise envelope filled with random Japanese candy every two weeks. In these blog posts I am hoping to share what I have learned from starting this small business. In my previous post I described how I got the idea for the service, but left it as a cliffhanger if anyone actually subscribed to it.

At first Candy Japan was without a name or a homepage, it was simply a suggestion to my opt-in mailing list of about 500 people interested in Japan. My post to the list was something like this, except it was in Finnish:

Hi there, greetings from Japan. You are receiving this message because you joined my mailing list. I've noticed that there are many Japan fans like to enjoy Japanese snacks such as Pocky chocolates. That made me think that perhaps a club for trying out random Japanese candies / snacks might be a good idea. I would send random candy periodically by mail. Sometimes it might be chocolate, sometimes something different. It would be a subscription service, kind of like a magazine subscription is. I would like to get some feedback on this idea, please respond to these questions and feel free to give any other feedback you might have.

- Are you generally interested in Japanese candies / snacks?

If you are interested then...

- Would you rather get a surprise random selection or pick the candy yourself?
- How often would you prefer to get an envelope of candy?
- What would you expect such a service to cost if you received a package twice a month?
- Do you have access to a credit card?

If you would like to unsubscribe from this mailing list, just reply so.

From the responses I could see that there was some interest, so I continued messaging with those people to see if they would actually go ahead and subscribe. The price range people thought was reasonable intersected the range which would still make a small profit. For the frequency of packages I think there was not that much preference, so I decided to send the packages every other week. In the future I might want to offer more options for delivery frequency, but I wanted to start as simple as possible. One regret with this system is that a "Japanese candy of the month" club would have been much easier to explain and would have had a more friendly price point than this strange bi-monthly arrangement. But I will run with this system for the time being.

Out of the people who saw my message on the list, 5 subscribed. I thought this was pretty good, as many of the people in the list might not have access to credit cards and the service did not even have a home page at this point. On the other hand some of these people had already dealt with me in my previous ventures and might have some extra trust towards me because of that. In Finland I had ran a Japanese online comics store, a Tokyo tour service and worked as a translator for some Japanese comics companies, so some people may have figured that they could trust their money with me again. So I was worried that this 1% conversion rate would not be repeatable because I would not enjoy similar trust in other circles.

After posting to Hacker News and then subsequently getting picked up by Japan blog Tofugu and shared around on social media sites, I was happy to see that even people who did not know me previously were making orders. Eventually in the first week I managed to get 21 subscribers. I will post about my profit margin and expenses in more detail later on. After I have sent a few more shipments I should understand all the costs better. In my future posts I would also like to share some more technical details such as how I set things up with PayPal to actually receive the payments.

Instead of a single up-front fee many services are now based on subscriptions, since the web makes it easier to offer services on a monthly basis. A somewhat predictable monthly source of revenue does sound appealing, so I wanted to get in on the action too with some service of my own. But what can I offer? The brainstorming began.

While completing my computer science degree I spent 2 years as an exchange student in Tokyo, Japanese being my minor subject. After my exchange period was over and I was back to my home country of Finland, I was constantly thinking about going back to Japan. In 2010 I finally fulfilled that dream and moved back to Japan, this time to a small city called Tokushima on the island of Shikoku, about 500km / 300 miles from Tokyo. Somehow I wanted to take advantage of my location, so I started brainstorming what kind of monthly subscription service I could start from here.

When you are in a foreign culture, even simple things can be very interesting. Take grocery shopping. When I go to the supermarket here, many things are not at all familiar to me, and old familiar things are not available. I miss rye bread for example. But the things that are available more than make up for the missing things. Many strange fruits and vegetables, spices and sea creatures up for sale. I liked Kaiware sprouts so much that I even started growing them on my balcony.

One fun thing is the huge variety of different sweets.  Some really strange, such as pieces of wasabi-flavored seaweed or DIY kits for making sushi-shaped sweets yourself. I figured everyone likes sweets and snacks, so it seemed like a good place to start experimenting with a subscription service. And so the idea for Candy Japan was born. 

I have some experience of running a small online shop before, so this time I wanted to selfishly forget about inventory control and just offer a single product for sale. Maintaining some inventory is not that difficult, but can add to costs and be a bit of a mental burden when you occasionally happen to stock something that never sells. To combat this, this time the product will not be any specific candy, but rather a random changing bi-monthly candy that the subscribers would receive. I thought this might be interesting for the subscribers too, since each package they receive would be a surprise. Subscribers might even tell their friends about the candy they happen to get or perhaps even taste it together.

To get the project underway, the first thing I wanted to be sure about was that this is something some people might actually want. As a small test, I calculated a price at which the service would make a modest profit and e-mailed my opt-in list of people from my previous online shop to see if there would be interest in subscribing at that price. I wanted to be sure that people would be actually willing to whip out their credit cards for this, so I immediately started charging for the club, even before I had a home page for it.

Did I get anyone to pay for this? I'll write about it the next time!

Platform launches are exciting times. A lot of small players turn into big ones as their apps take off and their studios expand from bedroom operations into multi-employee companies. Facebook, iOS and Android have been the biggest ones, but there have been some successes on lesser known platforms such as my own apps on MySpace and BuddyPoke on Orkut and multiple other platforms as well.

I would say that currently one of the most underappreciated platforms is Facebook apps for pages. What are they? Well, nowadays every business is starting to have their own Facebook page. It makes sense to them. Their customers become their fans, and then they have an easy way to promote their products and services to those fans in the future by posting page updates. In case you are confused, pages are distinct from the old Facebook groups, which I are basically being deprecated now in pages' favor. Groups also have a 5000 fan limit, so they are not suitable for serious businesses even if they do stick around.

Each Facebook page has multiple tabs. Tabs that all pages have in common are the Wall and Info tabs. On the Wall the company or brand can post things that might interest their fans, and those posts may get federated to appear on their fans' news feeds depending on the Facebook secret sauce EdgeRank formula. The Info tab is simply a few pieces of information about the business or brand, such as a description of what that brand is or where their homepage and physical addresses are.

So where is the opportunity? Well, you can create custom tabs. These are businesses we are talking about here, so unlike apps for users, it is much more likely that a freemium model would work for Facebook page apps. As for many businesses Facebook pages start to be their central presence online for their business, it is quite easy to imagine that there are some business-specific things they might want to offer on their pages such as reservations, e-mail newsletter sign-ups and the like.

These apps are quite straightforward to create as well. The procedure is quite similar to creating normal Facebook apps, the major difference being that you are not just dealing with users, but "pages" as well. In some cases you are acting as the page, such as when you are presenting content on that page's tab, but in other cases you might represent the user, such as when querying which pages that user has access to administrating. Once you get this permissions flow working, creating these page apps is quite easy.

Better yet, there is not as much competition here as you might expect from such a nice opportunity. The biggest player in the field is Involver with several apps under their belt. But looking at the app listings it does not seem that there is that much competition here yet, at least nothing compared to people making gaming apps for the general public that is much less willing to part with their cash! By asking for the right permissions, you can even have the privilege of posting to page feeds and being able to reach hundreds of fans at once with your app's messages, assuming they are relevant to the page. Seems like a great distribution channel.

I have done two such apps, AirBnB Listings and Fan of the Week. Each took me about one week to create and the process was quite painless. My major stopping point was figuring out how to get the add & permissions flow to work smoothly, but now that I have that sorted out I feel that my next page app will be even less effort than before If you are a developer considering what to develop for, I take a serious look at the Facebook page apps opportunity. There are businesses that could use your apps and are probably willing to pay for them too.

Here's my own limited perspective on how Google+ apps will offer new opportunities for budding entrepreneurs. The release of such a platform is just a matter of time, and will very likely be based on OpenSocial. When a new platform gets launched, the bar will be very low for the kinds of apps that will succeed. Of course Google will have learned from the experiences of MySpace and Facebook and will limit the viral channels more than they did even initially, but still success in the beginning should be a bit easier.

Why do I feel qualified to write about this? I am probably not the most qualified person on the planet, but I did have some successful apps on both MySpace and Facebook when their platforms launched. On Facebook I was a few months late from the initial launch, but still managed to whip together something that got nearly 10 million installs. On MySpace I had an app that still remains in the top 20 with over 5 million installs, although that does not mean much as the platform is all but dead now.

The kinds of things I expect to succeed on the inevitable Google+ platform include anything that has some sort of viral component that is effective enough to ensure that the app will spread widely, but not blatant enough to get banned or secretly de-emphasized by Google. At first the userbase will be very affluent and geeky, so things like friend graph visualizers might be popular. As more and more "normal" people join the network, things like receiving points for recruiting friends to the app will gain in popularity. Existing games will reuse their brands and assets, FarmVille will launch on Google+ and getting points for recruits will be less obvious and will be branded as "neighboring" your friends instead of just plainly inviting them.

There will be Q&A and quiz apps that will become immensely popular, unless these are features that will get integrated into Google+ before third parties have a chance to have their own takes on them. If there is an images API, then many apps that do some sort of transformations on user profile pictures will be popular. Causes will probably launch on Google+ and top the charts, and remain there as everyone loves to be charitable. They will remain uncontested, as making such an app will require connections with charities that most app developers will not have, unless there is a Google.org app in there.

Most of the money will be again made with social games. Some of these will not require a massive up-front investment. There will be some games that are more text-based and will profit from users paying with virtual currency. Google+ users will be more professional and less casual, so these will not be as popular as similar games on Facebook, but there will be some developers that will profit enough to make a living off of them.

There will be an app that allows you to anonymously answer questions about your friends. This app will be quite popular and make a lot of money on AdSense. How to profit from Google+? Read up on OpenSocial, refresh Hacker News until you see the platform release announcement and then code like mad.

Most people would agree that cold calling, unsolicited email and advertising are annoying things that the world would be better off without having. Imagining for a moment that we live in such an ideal world, how will people find out about new products or services?

Tech entrepreneurs are often criticized for only making services for other Internet-savvy techy people. Why aren't we making the lives of people in other industries easier? Well, suppose you take this advice and start making software that will really improve the lives of people who work at bakeries. You complete your software after 5 years of hard work. It's a true masterpiece and will double the profitability of a bakery, so you can't wait to get people to use it. But in this world where you aren't allowed to call, send unsolicited e-mails or advertise, how will you ever get a bakery to hear about your product? Mostly they bake stuff, maybe using POS software or excel spreadsheets to adjust recipes or keep books. They don't actively search online for things that might improve their bakery, and they certainly don't hang out at Hacker News, so they will never hear about your great new product that way.

You might say an industry event is the solution, but in a world where you are not allowed to initiate contact, an industry event will never get organized. After all, the event organizers would somehow need to contact the bakeries to inform them about their event, but that is not allowed in this world. They cannot even put an ad in the paper, because in this imagined society no advertising is possible.

The purpose of this post is to invite comments on how to resolve this situation. How would the ideal world work such that useful products could be discovered by everyone that could benefit from them, but without resorting to aggressive or annoying means of delivering the message?

Thanks to elkclone and wahnfrieden on #startups for suggestions to this post.

Inspired by patio11's comment on Hacker News, here are the results of my latest A/B test that I ran on Coolest Friends. I separated new users into three groups, the control group, the "no display ads" group and the "fix IE margin bug" group. The last group had a change that is actually a bugfix, and was included as a sanity test. 

The "no display ads" group was never exposed to the ad unit highlighted in green in the screenshot. I wanted to know if having this ad unit would lower conversions, and was happy to discover it did not.

"Conversion" here is "visited again the next day", or more accurately "generated a pageview 24-48 hours since installed app". The conversion rate of the control group was 11.00%, conversion rate of "no display ads" group was 11.29%. This is not a significant difference. There were 2348 users in "no display ads" group and 2271 users in the control group. If I am not mistaken, this amount of users would have been enough to make a statistically significant finding of a change in conversion rate of about 2 percentage points.

The sanity test group had 2322 users in it and a conversion rate of 12.53%. Most users use IE and the margins in the UI were messed up on it, so correcting the bug should raise the conversion rate, which it did. One interesting finding was that the "spin the wheel" link highlighted in red in the screenshot received a lot more attention when the ads were replaced by whitespace. The link lets users gamble with the points they have earned in the app. With ads replaced by whitespace 10.09% used the wheel vs. 4.88% normally.