Behind the scenes · 7 minute read · June 2026

My audience had no neighbors on the map. Here is what I did

Our system finds trends for an audience by looking at similar audiences. For the property due diligence world, it found none similar enough and returned nothing. So I mapped it by hand. Here is the honest story, and what it taught me about truly niche audiences.

Lera Leonteva
Lera Leonteva
CEO & Founder, Leo AI
Part 3 of 3
72%
similarity needed to find look-alikes
63%
the closest neighbor we found
0
trends the system produced
84
accounts I mapped by hand instead
A 3-part series
01 · The map 02 · The method 03 · No neighbors

Most audiences have neighbors. Find a few similar audiences, see what is working for them, and you have a head start. This one had none.

Here is how our trend engine normally works. To suggest what an audience might want, it looks for similar audiences on a cluster map and borrows their signal. It only trusts a neighbor that clears a 72% similarity score. For the property due diligence audience, the closest thing it could find was a productivity brand at 63%. Under the gate. So the engine did the correct thing and returned nothing.

That left me with a blank page and a real question: what do you do when an audience is so specific it has no look-alikes? I mapped it by hand. These are the things that surprised me along the way.

Surprise 010

The system returned nothing, correctly

My instinct was that a zero result meant something was broken. It was not. A 63% closest neighbor really is too far to borrow from. The honest output for a one-of-a-kind audience is an empty set, not a forced guess. The fix was not better automation, it was a person.

Surprise 024

The target was nearly invisible

I expected to find the closing attorneys themselves. Across all of Instagram, only four individual firm accounts cleared verification. Most do diligence quietly and keep no public feed. The people you most want to reach can be the hardest to see.

Surprise 0316

The institutions carry the audience

The signal was not gone, it had just moved. Title underwriters, escrow firms, environmental consultancies and CLE providers reach this audience every day. Title and escrow alone was 16 accounts, the largest segment. The audience is real, it just gathers around the institutions, not the individuals.

Surprise 0439

The numbers mostly would not verify

Of 39 YouTube channels, I could confidently verify a following for almost none, because the platforms block automated reads. Rather than guess, I left the counts blank and marked confidence. A map with honest gaps beats a map that looks complete and is not.

The number I keep thinking about

63%

The closest look-alike this audience had, under the 72% gate. Too far to borrow from, which is exactly why the map had to be built by hand.

Honest open items

What I could not close

The core segment, individual real estate and closing attorneys, stayed thin no matter how I searched. Four verified accounts, and a long tail of candidates I had to drop because I could not independently corroborate them. I would rather report a thin segment honestly than pad it with accounts I am not sure about.

And almost every follower count on the map is approximate or absent. That is a limitation I chose on purpose: better to show where the data runs out than to publish numbers I cannot stand behind. The full method, including the gate and the cuts, is in the method post.

What I am taking from it

  • A zero result can be the right result. For a truly niche audience, "no similar audiences" is honest, not broken.
  • Some maps have to be drawn by hand. Automation borrows from neighbors. When there are none, a person has to do the looking.
  • Follow the institutions, not the individuals. When the target is quiet, the organisations around them carry the signal.
  • An empty lane is an opportunity. Barely anyone reaches this audience on video. That is not a wall, it is an open door.

If you want the finished map, the 84 accounts and the six segments are in part one.

A note on the numbers. This came out of a human-curated inventory compiled June 1, 2026, after the automated trend source produced nothing for this audience. 167 candidates were reviewed, 84 verified and kept, 78 at high confidence. The 72% and 63% figures are the similarity gate and the closest neighbor the cluster map could find.

Keep reading

Who actually reaches property due diligence professionals

The finished map: 84 verified accounts across 6 segments, and who really carries this audience.

How to map a B2B audience that barely exists online

The repeatable, seven-step method behind the map, so you can run it on your own niche audience.

More from the blog

Our full library of writing on audiences, video, and what actually performs.
Lera Leonteva

Lera Leonteva

I am the founder and CEO of Leo AI. I spend most of my time on one question: how do specific audiences find and watch content, and what makes some of it travel. I write about what the data shows. More about Leo.