Despite many contrary hints in books and places like Indie Hackers, I thought that idea validation would get easier (and less important) as a startup takes shape. At least in the case of colabel, I must correct myself: It does not end, and it does not get easier. But we found a way how to conduct interviews with people to do just that effectively.
Why idea validation is hard
When you are starting, nobody knows you (except when you already have an audience). Also, compared to the number of people you should speak to, you don’t know anyone. Most people don’t really care about the product you are building. Not because they are bad people or your product sucks but because it’s simply not a match.
On top of that, contacting people and opening up about the beliefs that your life depends on is a scary thing on its own. But since you already know all that, let’s move on.
What we tried
Personal network. Worked. You should always try that first if possible. However, not everyone is looking for an automation platform for processing unstructured data, and acquaintances get annoyed by everything that remotely looks like sales very quickly.
Content marketing. Didn’t work for evaluation. Content takes about six months to become relevant in search engines, and many businesses don’t have that much time when trying to get quick feedback. Having said that:
“Publishing great content is almost never a bad idea.“— @thilohuellmann
LinkedIn campaign. Didn’t work. LinkedIn is probably the only registry in the world where 80% of users maintain their biographies to perfection – at least the ones that might be interested in what we have to sell. But it was a complete bust with 0 (that’s a zero) signups that cost us a fortune.
Twitter. Didn’t work. At least the bubble we operate in is a tough crowd, and it is certainly not useful to attract interview partners by the dozens, let alone hundreds for every hypothesis you want to test.
Reddit. Didn’t work for us. We didn’t make a big effort either, admittedly.
Communities. Didn’t work for us. People are generally friendly and supportive when it comes to shooting down a landing page, but we didn’t manage to find the people who were genuinely trying to solve a problem we could fix.
LinkedIn is a weird place. Many spooky people post and repost weird stuff to their feeds and send random requests without following up. We also did that for a while, but it didn’t work until we found Phantombuster and changed our attitude from bad to less bad.
By checking out their website, you’ll quickly get what their product does, but we didn’t understand how powerful it could be. Phantombuster is one of several web automation platforms that allows you to speed up certain things: Search people, scrape profiles, send invites and messages.
We initially played around with it to boost our Twitter following a bit. But since we had written off the platform as a sales channel months ago, that was only for fake pride. LinkedIn was a much tougher nut to crack as the platform is highly restrictive regarding company activity. But that was the wrong objective anyway.
The whole thing got wild when we started using one of their pre-built workflows. They allow you to chain things together using their templates, thereby getting proper guidance, next-level social media automation, and the necessary metrics to evaluate.
We ended up running a search for “operational excellence productivity” in Germany and Austria. This led to only about 700 results, but that was far more than we needed to speak with and even left us with the opportunity to speak about such delicate issues in our mother tongue.
Once we knew how to make the whole thing sing and dance, we ran into another problem.
Changing our attitude
Initially, contacting people by the dozens felt weird. We currently don’t have any sales pressure, but we knew that the ultimate validation would be a customer commitment of some form in the back of our minds. And a hard sale on a contact request is probably the worst conversation starter and only leads to one thing: LinkedIn recognizing that you are trying to avoid their ad wallet as people mark you as spam. And rightfully so.
The things we ended up trying worked almost as bad: Sending blank requests (works to connect but annoys people really bad once you make your pitch), refer to being a podcast host, and thereby implying to consider them for the show (terrible idea and I still feel bad for it), or ask them if they have a specific problem (works sometimes but people are primed from the very first touchpoint, thus making the outcome “yay” or “nay”).
Neither of the versions worked as intended, so we discussed our approach again.
What are our true intentions? Validate a feature.
What do we need the most? Open discussions.
What is the consequence if, after that discussion, they didn’t grant us their official signup? Exactly what we needed: Proof that they weren’t all interested in our service, allowing us to move on.
So we changed our message to the following:
You might guess that this changed things, and it did: All of a sudden, people were accepting my requests and even contacting me proactively. After submitting 20 requests and within half a day, I already had 3 appointments scheduled and 5 more friendly chat conversations where we realized that we didn’t need to speak in person.
My main takeaways
Despite all that, you will still feel like you are invading other people’s inboxes, and that’s because you are. Tools like Phantombuster operate in the grey area of morale. It was a matter of how I am using this tool, but I leave it up to you to judge. I strongly advocate against using this methodology once the idea has been validated. Then, you will be able to craft the right story, invest in a proper landing page, and acquire people through proper channels — and you will know which ones to pick because you actually know a bunch of customers in person.
So here are the main things I took from this exercise:
Know what you want
Use technology to enhance your search
Start with 100–150 invites and see what happens
Tell people what you genuinely want right off the bat
Be nice and accept no for an answer
Stop the process once you gain the confidence to build and sell the feature – then build and sell the feature.