A client came to us with a question - how do you find 50yo+ millionaires, executives, CEOs online. The group famous for being unreachable and elusive. The age-bracket that uses technology infrequently and with doubts.
My answer - just DM a few hundred of them.
The big problem we were trying to solve is finding people who are generally not very active online nor easily reachable. We wanted them to participate in a high-stakes research project for a technology start-up.
It had to be a numbers game - if you need 50 participants and only 5% are likely to respond back to you, you need ~1000 prospects.
Messaging a 1000 prospects is a logistical challenge. FINDING 1000 prospects without tech is humanly impossibly.
So we did it without humans.
We settled for Twitter as our platform of choice - it’s one of the demographically oldest social platforms and one with relatively open access to its data (even post API changes).
First, we brainstormed what these individuals are likely to follow online - business news, professional organizations, luxury products, services for the wealthy.
We also had a brilliant idea that a lot of the more influential members of our audience will be FOLLOWED BY the same organizations - because they are founders, investors, advisors and board members.
So we downloaded nearly half a million accounts across following and followers of these accounts.
Then, we filtered - for digitally engaged individuals, with a bio, multiple posts, accounts older than 5 years and latest post within 1 month. We knew that these were the people we were most likely to get any response from.
We were left with about 50k accounts. We downloaded all the posts and full info for these over a course of a few days.
And then we had our research intern (an LLM) look through all of them, one by one, and rate them on key numbers. What was their likely age? Were they experts? Were they affluent? Were they a cultural match for a project?
We were left with about 20-30 accounts per key locations - letting our client get 1-3 responses in each key region and snowball sample from there.