Housing.com · Platform health

No Site Visit Required

Only 1 in 10 brokers verified their listings. They wanted the badge. They just couldn't get to it.

Verified listings 12% → 18% Manual review −80% 4 months
See the screens
Housing.com verification options, with the new Self Verify option

ACT 1

A badge almost nobody earned

On Housing.com's search results, the verified tag is a promise: this listing is real. It should have been everywhere. Instead, just 12% of listings carried it.

Brokers had exactly two ways to earn it. Drive to the property and take geo-tagged photos on site. Or ask Housing's data collector team to go on their behalf, which was slow, costly and painful for everyone involved.

Two routes. Same dead end: someone had to physically stand inside the property.

ACT 2 Empathize

First, does anyone care about the badge?

Pushing a metric nobody values is just busywork. So before chasing more verified listings, we went to buyers and asked two blunt questions: what does "verified" actually mean to you, and do you trust the tag when you see it on a search result?

Their answers set the bar for what "verified" had to guarantee, and turned a vanity number into something worth growing.

ACT 3 Empathize

Then, the brokers

If buyers wanted the badge, why weren't brokers earning it? Nobody said "I don't care." Everyone said some version of "I can't."

"I have too many properties."

Visiting every listing just to verify it doesn't fit a broker's day.

"There's no network there."

Patchy signal on site made geo-tagged uploads unreliable.

"My tenant still lives there."

You can't just turn up and start photographing someone's home.

ACT 4 Define

The barrier was the trip, not the broker

Our closest competitor sat at around 20% verified listings, and they weren't moving. At 12%, we were behind, but the reason was obvious: the process, not the people. Remove the site visit, and the gap was ours to close.

There was a cost case too. Every field visit by the data collector team was our most expensive way to verify a single listing.

HOW MIGHT WE

…let brokers prove a listing is real without anyone visiting the property, and without letting fake listings slip through?

ACT 5 Ideate

Easy part: what. Hard part: where.

The core idea came quickly. Let brokers upload geo-tagged photos from anywhere, including photos the owner shares with them. The question we wrestled with was where this should live, so it felt like part of the job, not one more chore.

Option A

A separate verification flow

Its own journey, reached after the listing is live.

  • Clear and self-contained
  • One more task to remember
  • Asks brokers to come back later
What we chose

Inside the photo upload

Verification happens while brokers add photos to a new listing.

  • Zero extra steps
  • Brokers are already holding the right photos
  • A soft callout, never a blocker

ACT 6 Prototype

From a detour to a side effect

Brokers now see Self Verify as a new option alongside the old routes. And while they add photos to a listing, we show which images are eligible for verification, right in the upload step. Brokers could publish either way. Verification simply became something that happened while they were already doing the work.

Verification options: Self Verify (new), self verify at the property, or request a Housing field visit
The choice: the new Self Verify sits alongside the old site visit and field visit
Photo upload step showing which photos can verify the listing
The flow: verification built into the photo step
Search results with a Verified tag on a listing
The payoff: the Verified tag buyers see on search results
Self Verify card: get up to 3X more leads, earn credits for verifying listings
The nudge: sellers see why it's worth it: more leads, plus credits for verifying

ACT 7 Test

Swapping a review team for two models

Remote uploads only work if you can trust the photos. Until now, a manual team checked every one. We piloted replacing them in a single city with an ML pipeline, and my job was to decide how good it had to be: the thresholds and acceptance criteria that separate "verified" from "try again."

Geo-tagged photosuploaded while listing
→
ML taggerwhich room is this?
→
ML auditorquality & integrity checks
→
Verified taglive on search results

ACT 8

The pilot talked back

Speed turned into delight

The tag arrived faster and more accurately than manual review ever managed. Brokers noticed, and they were genuinely happier to see their listings verified so quickly.

The model wouldn't bend

A human reviewer can make a judgment call. A model can't. Honest edge cases got rejected, so we reworked our thresholds and acceptance criteria, with ops helping us tune them, before going pan-India.

ACT 9

Four months later

Once the pilot held up, we scaled it across India. Here's what moved.

Verified listing share

12%
18%
BeforeAfter 4 months

+50% growth in share. Bar heights scaled to a 0–30% axis.

Manual review workload

100
20
Manual teamML pipeline

Indexed to 100 before the change: an 80% reduction.

Closing the gap with our main competitor

Housing.com, before12%
Housing.com, after 4 months18%
Main competitor, holding flat~20%

Bars scaled to a 0–30% axis. We grew while they stood still.

THE FINAL WORD

Meet people where they already are

Brokers never needed convincing that verification mattered. They needed it to stop costing them a trip. The breakthrough wasn't a clever new feature. It was moving an existing step into a moment they were already in, and letting technology carry the weight a person used to.

Read another case study Get in touch