Lessons learned on the road to product-market-fit: the CivikOwl story
We’ve been working on CivikOwl for 15 months and may finally have that elusive product-market-fit. Getting here was painful as the many pivots tested the team’s confidence and mine. Many nights I’d ask myself “was today worth the $1000 expense?” Sleep afterwards was not particularly restful.
The aim of this post is two-fold:
- To share CivikOwl’s lessons in the hunt for product-market-fit so others may benefit.
- Thank our users who steered us to this point. You are by far the most important element of finding product-market-fit.
Too long, didn’t read
The short-version of our story: we picked a problem and solution we were passionate about and matched user feedback to fit this story. But those users had more pressing worries on their mind and discovering what was really top of mind for them helped us find product-market-fit.
Solving the right problem, first
My co-founder, Ajoy, and I initially observed that we sometimes read news that got us fired up about an issue but left us feeling helpless when we couldn’t do anything about it. An example in 2016 that was vivid to both of us was the Brock Turner rape case at Stanford.
We thought that giving readers of news a quick way to take action — particularly contacting their political representatives — would be very helpful. And when we interviewed dozens of people they validated our feelings above.
Failure #1: Give A Damn
After 7 months of development (3–4 months too long) we launched the Give A Damn button on Feb. 13th 2017 to positive feedback but abysmal usage.
We weren’t sure if the problem was our product complexity (adding the Give A Damn button to a story required creating a short-link to it using our widget) or distribution/awareness.
We had a perfect test on April 6th when a well known author, Scott Santens, retweeted his article with our button attached and we had a few hundred people see our product live. A flurry of activity resulted in some usage but no repeat usage.
Now we knew the problem wasn’t awareness. But we incorrectly diagnosed the problem first as a UI/UX problem and later as a collective action problem. i.e. nobody wants to take action first but will if they see others doing it. So we set out to build a forum where people could discuss news they cared about and take collective action.
Failure #2: The forum
In designing the forum we studied Facebook, Twitter, Reddit and Hacker News to identify deficiencies so that we would have a reason for people to join.
We identified the lack of anonymity in Facebook to be a huge impediment to honest conversation. But we also saw that anonymity could easily swing the other way into vitriol.
Our reading of Clay Shirky’s essays led us to design a token based system, called Respects, to encourage healthy discourse. And to this we added our Give A Damn button to make it easy for people to take action within the forum.
We launched the forum on June 12th to a closed beta group of 50 users. This time usage was better because we had lined up people to participate. And our Respects system may indeed have contributed to the generally higher quality of discourse though selection bias of the group is hard to separate out.
But we didn’t have enough critical mass of people in any geography to see collective actions take off. Worse, it was starting to become clear that these actions — our primary differentiator — were a flop.
Fundamentally, people didn’t believe that legislative actions would result in much, even with a group of folks reaching out. And no matter how much we said otherwise they were right. It wasn’t that the action was too hard to take. It was that the result of that action was usually not worth the effort.
Beyond that, one of our users, Alan Berkson, astutely pointed out that no one knew when to come to the forum. We were neither breaking news, nor an effective wrap-up of important news. It was just random news, with a random (but thoughtful) audience, and not fresh enough.
The result was sparse usage and little differentiation.
Third time’s a charm
On July 15th we were mulling over Alan’s feedback and thought about a Chrome extension that could bring people to the site when they were reading something that was being discussed on the forum.
When we started to develop this we were, for the first time, really listening to what people were saying about the sentiments they had about the news. And this was when we finally saw that users like Sam Mallikarjunan had been talking about the same problem for a long time: “I don’t know if I can trust the news anymore.”
The trust problem is what’s top of mind for people. Or, put another way, the problem of trust is higher up in the funnel than taking action on a story. Only when people can once again trust the news, and feel like they understand an issue, are they most likely to take action.
This is what led to our algorithm StoryRank and the CivikOwl chrome extension. The extension helps people evaluate the credibility and bias of a a news story based on its sources and then suggests the best-rated stories from across the political spectrum.
Feel the love of product-market-fit
In our many pivots to find product-market-fit I began to wonder if I’d even know what product-market-fit would look like. I felt like we were discovering pieces of a giant puzzle in the dark and would run out of money before we saw the whole picture.
Well when we finally achieved product-market-fit the three signs below made it obvious:
- Users did not uninstall the product quickly.
- Users reached out voluntarily to suggest features.
- Users told their friends about our product who also installed and used it.
The simplest indicator was the count of times a user said “love” in a user test when talking about our product.
- Zero “loves” = fail. Also, “like” may as well be the same as “hate”, i.e. fail.
- 1 or more “loves” means we’ve got something of value.
As we started to see people use the word “love” about our product we started to feel that we had something of value. We still have more to do but feels like we’re closer to product-market-fit than ever before.
If I could do it over
The story of CivikOwl seems so familiar when I read stories from other entrepreneurs.
Boy meets product. Boy falls in love with product. Boy tries to force-fit parents, err I mean problem, to product.
They say you should not fall in love with your product but rather fall in love with solving a problem. True. The challenge is that it often takes a while to identify the right problem to solve. And this period is frustrating because it feels like you have picked a good problem to address only to realize at each turn that you didn’t dig deep enough.
If we could do it over we’d have spent more time thinking through the steps our user takes with the news on the way to the outcome we wanted. Perhaps we would’ve identified where in the sequence most people were stuck and focused there instead of the bottom of the funnel at the outset.
Another lesson learned is to not let the limitations of technology dissuade you from solving this highest priority problem. When you immerse yourself in a problem you will likely come up with solutions that were not obvious on first glance, i.e. machine learning can solve problems in ways you didn’t know were possible :)
Good luck fellow entrepreneurs!
