Apple gave me a really nice birthday present on December 2nd, when they approved my iPhone application: iSimplifiedChinese. For those who don’t know it, it allows you to practice your Chinese pronunciation by sending your pronunciations to a real Chinese person for feedback.
I expected waking up filthy rich the next morning, but unfortunately it turns out I need to do a bit more work for that. There could be many reasons why only one copy was sold in the first week, but I figured that the most import reason was that nobody knew about it. Time to reread the 4HWW chapters on Marketing.
My goal for January is to sell 120 copies or the equivalent in other revenue. If that fails I want to at least obtain enough data and insight to know what to do next. My remaining budget for this is about €50 and 50 hours.
There are roughly two forms of marketing. The first is semi automatic and fairly broad like Google Adwords. The idea is to expose as many people as possible to your offer, like with a physical billboard on a road. The trick is to find billboards on the smallest streets with the most potential customers and the least competition. The second is manual and highly specific; find individual people or organisations willing to spread the word. This post focusses on the first option.
Trademark issues
I created a Google Adwords campaign with the following ad, but that’s where I ran into the first obstacle:
Like dogs that urinate on anything they run into, companies register trademarks and expect full control over any use of them. And because companies don’t have leaches, other companies are so afraid of them they don’t take any chances. The result is that Google forbids using the word “iPhone” in an advertisement text. Imagine being a pizza courier and not being allowed to use the word “pizza” in your advertising because some restaurant trademarked it.
At this point both companies start acting like a government. Google has a well hidden procedure that allows trademark owners to give permission and Apple seems to have a well hidden email address where you can ask their Trademark Team for such permission. They even added an auto responder that basically says they might choose to ignore your email. I wrote them on December 19th and I’ll update this post if they choose to reply.
There are a few ways to get around this problem. I could use the word “smartphone” and end up paying for clicks on Android and other smart phones. I could use alternative names like “apple phone” (apple without caps is a normal noun), “lPhone” (L in stead of i) or some other way to come accross as a Russian phishing site or Chinese karaoke Top 40 cover CD.
A better option is to only show the ad on iPhones. That way I can be sure that the prospect has an iPhone, without using the i-word. A major disadvantage is that iPhone owners looking for ways to study Chinese are probably using their laptop, because they expect to find online course material with large PDF files, etc.. But at least this would get me some data.
Google Adwords and Facebook campaigns
These are the keywords I used.
I immediately ran into the next problem: the suggested bidding price for these keywords is much too high. A little math: the application sells for €11.99 which leaves €7.30 after tax and Apple’s share. So let’s say I’m willing to spend about €1 to find a customer. Of those who click on the ad, perhaps one in a hundred decide to buy it (so far it’s even less). That means I want to spend about €0.01 per click and certainly not more than €0.05. For many of these keywords Google suggests bidding about €1 per click.
My theory is that people who bid €1 per click are selling something worth at least €100. So I need to either make the product much more expensive or find cheaper keywords. But the search terms shouldn’t be too rare either; so far about 4 out of 1000 people click on the ad, so an obscure search phrase like “gimme a cool iPhone application to learn Chinese pronunciation with real teacher feedback” might be useless, because you need about 1 / 0.004 / 0.01 = 25.000 people using it in order to sell one copy. Unless of course conversion rates improve dramatically with those kind of keywords. But then how do you come up with such search terms?
This meant that my ad was almost never shown in search. However Google also displayed the ad on other websites that show Google Ads. That resulted in about 400 clicks this week of which four people went to the App Store. Unfortunately it did not result in any sales. It does however tell you from which websites your clicks are coming, which is very useful information for a more personalized form of marketing.
I also tried a Facebook campaign. They allow you to show the ad based on people’s nationality, age, interests, etc. Similar suggested price per click, but I tried it anyway to get some data. 26 clicks have not lead to any sales as far as I can tell, but that number is probably to low to measure conversion.
I paused the Google campaign until I know how to dramatically improve this situation and I’m leaving the Facebook campaign open for a while longer to see what happens.
Now what?
I did sell a small number of copies over the past month. As far as I know, Apple does not provide any means to figure out where those sales came form, but from what I can see they did not result from the ad campaign. Most likely people found it directly through the App Store.
What I would really like to do is to rule out marketing as the problem for the low sales. I need to be reasonably sure that enough of the right people know about my product and still choose not to buy it. That way I can move on to test other factors such as the price, number of lessons, price per pronunciation, functionality, total market size, description. I’m not sure if the current data allows me to draw any conclusions.
Another intersting observation is that most visitors come from China (51), but they only spend 9 seconds on the site (gobal average 13 seconds). 22 people from Indonesia and 12 from Hong Kong spent one minute on the site and 20 people from Thailand spent 30 seconds. Who are they? Travelers or locals? The Indonesian visitors have their operating system language set to English, but that doesn’t really mean much. Also, these averages could be caused by one or two people leaving their browser open for an hour.
Update Jan 14th: last week I decided on a number of improvements that I want to make to the application and course material and resume the marketing experiment in February. In the mean time I lowered the price and included 60 in stead of 10 feedbacks to make it a more complete package. I also changed the wording in the App Store and on the website. I’m not sure how many people will find my application without the ad campaigns, so these changes may not have any effect on sales until February.