Does geo-fencing for hyper-local seller leads actually surface motivated sellers

The intent-signal premise is real but the execution gap is enormous, and what you described is almost exactly what happens when the vendor is reselling audience segments they bought wholesale rather than actually building a custom geofence around those specific locations in your zip codes. I ran a test on something adjacent in Q3 last year, not Columbus but a Midwest suburban market, targeting people who had physically entered self-storage facilities within a 12-mile radius of our highest-turnover properties. Paid $3,100 for eight weeks, got 21 form fills, and the operator I handed them to said roughly the same thing you said: a third uncontactable, a handful confused, one agent. The two that were real were people already on our direct mail list anyway. What I think is happening is the device ID matching is messy at the physical location level, so the audience you're actually reaching is anyone whose phone GPS put them within 200 meters of a storage facility at any point, which in a dense suburb is a lot of people who just parked nearby or drove past. The signal degrades fast.

The portability thing you mentioned at the end is the real lesson I had to learn the expensive way. When the audience data, the pixel, and the creative all sit in their platform, you are not building anything, you are renting eyeballs for six weeks and then starting from zero. I now will not run any paid campaign where I do not own the pixel outright, the CRM integration is direct, and the audience list gets exported to me in a format I can actually use. That condition alone eliminates probably 60 percent of the vendors I have talked to, which tells you something about what most of them are actually selling.

Paid $3,100 for almost identical garbage in Akron, same estate sale foot traffic pitch.

The portability problem is the one that comes back every time. Owning the pixel and the audience segments has to be settled before anything gets signed, and a vendor who walks rather than agree to that has told you what you needed to know. On the intent signal question specifically, it does work in some markets. Phoenix is one where it has held up, and there the vendor was pulling actual deed and probate filing data layered on top of the geo-fence rather than foot traffic at a storage unit. That combination brings cost per real conversation closer to $600. Pure location ping targeting with no public records layer is mostly noise.

The estate-sale foot traffic signal is real, but three zips is too thin a sample to matter statistically.

The Columbus outer suburbs specifically are rough for this because you are dealing with a lot of retirees and older homeowners who click on anything that looks official, and storage facility visitors in that ring are mostly renters in transition, not owners with equity. I ran a similar test in the Indianapolis east side, not geo-fencing but similar intent-signal logic, and the problem was identical. The signal tells you something happened near a building, it does not tell you the person owns anything or is thinking about selling. Probate attorney offices sound good until you realize half the foot traffic is heirs, attorneys themselves, paralegals, and people there for completely unrelated estate matters. You basically geo-fenced a building type and called it seller intent.

The $2,400 per conversation number is actually what this product costs when it works, which tells you everything. I have seen GCs on the service side pay similar rates for qualified homeowner leads through direct mail in distressed zip codes and still find the math tight. The real tell for me was when you said the vendor held the creative and the audience data. That structure is designed so you cannot test whether their audience actually performed or whether your offer and creative just did not land, because you will never know which variable failed. Anyone running these programs seriously should be hosting the pixel on their own domain and pushing leads into their own CRM day one so at minimum you exit with a suppression list and some behavioral data you can actually use next time.