I've been talking to many founders lately. I've noticed a common bias that many of us have: we tend to fully trust data that we collect online. However, if you start building a product based on conclusions you've made solely on studying online groups you might end up pursuing a totally wrong path. Here's how it happened to me.
I have actually 2 absolutely different examples to share.
First case study
First, an audience pretty far from tech. About 4 months ago I stumbled on this group on Facebook — parents of picky eating kids. It has >50K members and I found this to be a fascinating business opportunity. Why?
— I have a strong affinity with this audience (both my 7yo and 3 yo are extremely picky when it comes down to food).
— I've read a lot when I was looking for ways to solve this problem.
— I saw a community that had a very strong pain and it looked like they were looking for solutions.
I talked to my partner about this opportunity and he was genuinely inspired. Being a tech person, he immediately jumped into offering technical solutions that we might create to cater to this audience: a mobile app that will calculate daily consumption of nutritions for a child, compare it to the standards and let a parent know should they panic or not. It might also offer some ways out (ex, make your child eat 1 slice of apple).
I stopped him cold in his tracks explaining that we're going to do a deeper research on this topic first. That's exactly what I did.
1. I located 4 more Facebook groups that were dealing with the same topic. I managed to find a couple of subreddits too. It was not easy, because simple search doesn't yield any relevant results, but luckily there are already tools that help with searching subs (I used Gummy search)
2. I went over Twitter — nothing there.
3. Next step: I manually went through all these groups and took randomly 200 posts (50 from each). I tagged every post, what was it about.
4. And placed it on the graph to see what were the most pressing pain points people were talking about (I used the same approach when launching new items on Amazon several years ago, and it worked wonders for me).
This is what I've got:
As you can see, parents did not need nutrition calculations😂 They needed to share their wins and get some emotional support.
But then I started thinking: what does this data tell me actually? Does it tell us that I should start thinking of a solution that would provide mental support and offer a sense of community to the parents? A community on Circle, perhaps. Or does the data tell me that people come to Facebook and Reddit to rant and rave?
My next step was to run customer interviews. That showed me a totally different picture: by talking life to 20 parents I found out that the problem of having a picky eating kid was IMPORTANT but NOT URGENT, according to the Eisenhower Matrix. Therefore, building a tech solution on it was a very risky idea. Plausible, but low chance of becoming profitable fast.
Second case study
Second example that I want to share is more tech-related. During the search on Reddit and Facebook I discovered a rather impressive group of aspiring entrepreneurs complaining that they can't find a tech co-founder to launch a startup. Again, there are so many matching tools already out there but there's still a segment who is very unhappy with existing options. Again, I did the same exercise.
Step 1. Collect data through online research.
I took 2 subreddit groups on startups and manually went through 200 posts. Then I tagged them by topic "find a tech cofounder", "ways to get funded" etc. Then I build a simple graph in Excel to see where I stand.
What would you do if you saw a data viz like this? You'd probably decide to focus on early hires. That's what I thought too 😅
I have some experience in this subject — I used to hire A LOT while running a content marketing agency, and then retail shops several years ago. Sometimes it went good, sometimes bad, finally I came up with a framework that worked for me. Then I was hired for my first failed startup — and it turned out to be a disaster. Apparently, the principles that were very well applicable in "headhunting" in traditional businesses were misfiring in startups. Then I managed to create a remote team of freelancers, was hired numerous times as a Bubble developer, a product manager and a marketer on several early stage projects. I've seen it from both sides, so to speak. And I felt like I had something to share.
Step 2. This time, I decided to go with a very basic MVP first. I created a very simple landing page and a product on Gumroad open for pre-sales. A book about secrets of hiring better talent for less, being an underfunded startup. I ran several rounds of ad experiments on Reddit and Facebook. I waited. And waited. And waited. And nothing! No pre-sales!
Step 3. Customer interviews coming next. 10x40 min interviews with non-techie founders and I had a slightly different data:
It showed that founders on the idea stage and pre-launch stage mostly had concerns on how to find a tech co-founder. While early hires is a question that startups on the revenue stage solve and, perhaps, they are not so much interested in hiring for less. They want to hire better.
Obviously, I had to pivot the idea of the product. Luckily, I had a huuge personal experience in this area too, because I failed to find one when I decided to launch my first startup. I made every possible mistake, my expectations were way too high, and I was hoping beyond all reasons that a talented and smart person would work for me for free while I could not even articulate properly why my product was for and how I was supposed to find my first 100 customers. So, I just packed up everything that I learned and what worked for me, and created a course How to Find a Tech Cofounder. So far, it shows much better results😂 Now, we're getting feedback and try to figure out what part of this course could be productised.
The takeaways:
1. Collecting data online and analysing it in order to build a better product is a plausible way to do user research. But nothing can be as important as life interviews. Obviously, you have to do them right (for instance, if you ask a person "What is your main pain" or "Would you be interested in buying something like this" — you'd better not do them at all, that's not interviews).
2. Surveys will not work on this stage — I've heard someone offered to replace interviews with them. Surveys are useless for discovery of the problem. They are not so bad for fine-tuning some features and confirming a hypothesis.