23 Feb 2014 By Iris Shoor
Coding Marketing: How We Coded Our Way to 100K Unique Visitors
Can a team of hard-core backend developers do marketing?
One of the main marketing challenges I faced over and over again was how to do effective marketing with a minimal or nonexistent budget. I always relied on the people I worked with, getting them to write, design and help me build creative campaigns. This didn’t prepare me for my next marketing challenge, though: no marketing budget and 5 out of 6 team members being developers––everyone but me. I decided to build our marketing strategy around the thing my team does the best: coding.
Here are some of the techniques we used in order to get from a few hundreds to 100K unique visitors in three months (to our website, blog and mini sites). It’s a proof the new marketer is techier than ever.
You don’t need great writing skills (or any at all) if you have exclusive, fascinating data. One of the first companies to use this technique for content marketing was OKCupid, which mined its database to answer important questions like “Do taller guys have more sex?” You can find the answer here. The first time we published content that relied solely on developers’ work was pretty random. When we had to choose in which Amazon region we wanted to store our data, we ran some tests and found out there was a big difference between the regions. Sounded like an interesting story. We spent an extra day improving our script and making sure we got the data right. The output was super interesting––the AWS Olympics. The results were featured on VentureBeat and published on our blog, bringing massive traffic of our target audience––over 15,000 unique visitors.
That post might have happened by mistake, but when we saw its impact we decided that about 10% of the developers’ time would be dedicated to writing apps and scripts which would result in data for content marketing. The next “coding marketing” project was completely planned and “engineered” to get high traffic. My first decision was to build something around GitHub. Just because it’s a hot topic? Well, yes. We wrote a small app that relied on the GitHub API and downloaded the 30,000 most popular projects. Then we wrote a script that analyzed all the projects and produced a list of all the libraries used in these projects. The outcome was fascinating: we identified the most popular libraries today and what the leading trends were. Writing the code and analyzing the results took us about three days. The result: over 40,000 unique visitors to our blog, most of them developers.
Here are a few other examples of content our developers created:
– The Anatomy of a Great Stack Overflow Question (After Analyzing 10,000) – link
– The Logging Olympics – A Race Between Today’s Top 5 Java Logging Frameworks – link
Our target audience is developers, so all of the examples above are pretty technical. However, I truly believe this method can work for other markets. Most of the large websites/ services offer some kind of API, and using it wisely can uncover amazing data.
Mechanizing market research – what happens when you replace an analyst with a developer?
When I decided content marketing would be our main channel, I started researching other sites and blogs. My goal was to learn what content tends to be more popular, which posts gain more engagement, and what the hottest topics concern. It took a while. So,
We wrote a script that sorted blog posts according to the number of shares, using Twitter, Facebook, Google+ and LinkedIn APIs.
We ran it on a few dozen blogs (technical blogs, blogs of other companies).
The result was a list of all the popular posts in different blogs. That was a great starting point from which to choose the focus of our blog. By easily comparing content which spread well vs. less popular content, I was able to recognize trends I wasn’t aware of. Here are some examples from this research on content that is likely to be popular: comparing different products, any case study that involves an interesting company (one paragraph about Twitter’s backend can make a post fly through the roof), data and benchmarks.
Another interesting “extension” we wrote was a tool that helped me analyze the titles of successful posts. I believe a great title does 30% of the work. I was able to see which words appear more often, whether it is better to use short or long titles, and what the effect of numbers in the title is. You can read more about it here: The dark science of naming your post, based on studying 100 posts (link).
Slice and dice – cut an existing part of the product and launch it as a new one.
This is a very interesting strategy, as it enables creating great content with minimal investment. The idea behind this approach is taking one feature from the existing product and launching it as a completely new product (I’m not talking about a core feature that might compete with your main product.) Here are some advantages of recycling parts of your code:
You’re able to put up a pretty nice app within a day or two.
This app doesn’t need to generate revenue, so it can be very light and even silly.
A brand new product is great marketing material, and it’s relatively easy to get some attention to it. Here is a list of all the websites that are dedicated to new products; just by putting a service in a few of these places, you can gain thousands of visits. Places to post your startup
Here’s an example of how a minor feature got to play a main role in marketing. One of our UI elements is presenting the flow of the code. Instead of using text we have this cute element:
In no way is this Takipi’s core technology. In two days we built the “Stackifier,” an app that allows developers to paste their stack trace (a file with the input of the app) and see it in a nice and organized way. It’s not a “wow” product, nor one we can charge for, but it’s a practical tool. We got about 10,000 visitors in the month of the launch, and we still have some new users coming to the “Stackifier.” We mention that this tool was built by Takipi, and put it under our domain. About 10% of the visitors continue on to check out what Takipi is.
Of all the different marketing channels this one has the best conversion rate of installations; that is, a user coming to Takipi from the Stackifier is more likely to install Takipi than a user coming from our blog. I realized that users are more likely to convert to the same action: if they’re reading, they’re more likely to read something else, if they’re playing, they’re more likely to play a different game, and if they’re testing a new product–you know the answer.
Some other interesting examples:
Visual.ly used its broad set of infographics tools and created a super-focused product to reach out to a specific type of users: marketing managers. They launched the “Google analytics report” which automatically creates a weekly report based on Google analytics, presented as an infographic.
HubSpot used parts of their blogging software and recently launched the “HubSpot Blog Topic Generator.” It’s more of a fun game than a product, but it’s a great and cheap marketing tool.
Using existing APIs to create amazing content in a few hours
This great post made me think of using existing APIs to create a new site which would drive traffic to Takipi: Stop Writing Blog Posts: Ideas for Interactive Content (Link). I decided to go with a calendar. We created java2014.org and scala2014.org. It’s a simple calendar, with all Java and Scala events for this year. We outsourced collecting all the events, and used ODesk to build the events list. It’s useful, has high SEO potential and drives nice traffic of very targeted users.
In the three weeks after launching it we had about 4000 Java developers visiting the site. It might sound cheesy, but in a way I feel it’s very honest marketing: you really give value to the prospects, even if they don’t visit your website. Check out : MapBox, Arbor.js and TimelineJS.
Thanks Chen, Dor, Niv, Tzofia and Moshe for welcoming and cooperating with my crazy ideas.