The Situation
Berkshire Hathaway HomeServices Northwest Real Estate, a Pacific Northwest brokerage within the BHHS network, was running Facebook advertising to drive buyer engagement. The team understood their market and had real instincts about creative. Performance was not the problem of a brokerage that did not understand its category.
The problem was that paid social was not built to scale.
In 2017, the entire year of Facebook advertising produced 916 property searches. Not 916 a month. 916 across twelve months. The campaigns were running, the budget was being spent, and buyer demand existed in the market. The system itself capped what the spend could do.
The Primary Challenge
They were not running out of ideas. They were running out of tests.
Campaigns relied on manually promoting individual property listings. Listings had to be updated by hand as inventory changed. Only a handful of creative and audience combinations could be tested at any one time. Every property got shown to every potential buyer regardless of behavior, intent, or fit.
The constraint was not creative quality. It was testing capacity. When the system can only evaluate a few variations at a time, high-performing combinations stay hidden and underperforming ones keep eating budget. Optimization becomes guesswork because there is not enough data to optimize against.
That is a structural ceiling, not a creative one. No amount of better ad copy fixes it.
Key Outcomes
- Property searches grew from 916 in 2017 to over 120,000 in 2018
- Cost per property search dropped from $39.38 to $0.30
- Overall advertising costs reduced by 40% year-over-year
- 171,765 link clicks generated through paid social
- Click-through rate climbed to 6.62%
- Engagement increased by more than 491%
Our Approach
We rebuilt the system around two structural changes that worked together: dynamic catalog advertising to handle inventory at scale, and a testing framework that could evaluate hundreds of combinations instead of a handful.
Dynamic catalog advertising eliminated the manual work of promoting individual listings. Facebook’s real estate catalog system was implemented to automatically synchronize property inventory with active campaigns, which meant ads dynamically displayed relevant homes based on each user’s browsing behavior rather than showing the same listing to everyone.
The testing framework rebuilt how creative and targeting decisions got made. Instead of testing one variable at a time and waiting weeks for clean signal, combinations of images, headlines, audiences, and placements were evaluated simultaneously. The goal was not just to identify what worked, but to identify what worked in which context, for which audience, against which inventory.
Execution Highlights
Dynamic Listing Infrastructure
Property inventory was synchronized with Facebook’s catalog system, which made it possible to serve dynamic carousel ads tailored to each user’s browsing behavior. Manual ad creation for individual listings was eliminated entirely. The brokerage stopped fighting against inventory turnover and started using it as fuel for the system.
Scaled Creative Testing Framework
Testing capacity expanded from a handful of variations per campaign to hundreds. Combinations of images, headlines, copy, audiences, and placements were tested in parallel rather than serially, which meant winning combinations surfaced in days instead of months. The signal density made every subsequent decision sharper than the last.
Automation and Optimization
Automated rules paused underperforming ads as data accumulated, and budget was reallocated toward top-performing combinations in real time. The system ran continuously rather than waiting for manual intervention, which meant performance compounded between active touchpoints rather than stalling between them.
Full-Funnel Campaign Structure
Cold campaigns introduced new audiences to listings. Retargeting campaigns re-engaged high-intent users who had already shown signal. Dynamic ads personalized inventory based on observed behavior across both. The funnel did the qualification work the manual approach had been forcing budget to do.
Results
Property searches grew from 916 in a full year to over 120,000 in the following year. Cost per property search dropped from $39.38 to $0.30. Overall ad spend was 40% lower than the prior year, meaning the lift came from system efficiency rather than budget.
Click-through rate climbed to 6.62%, total link clicks reached 171,765, and engagement increased by more than 491%. The brokerage moved from inconsistent performance to a paid social system that generated sustained buyer demand at a fraction of the previous cost.
Strategic Takeaway
Most campaign inefficiency does not come from weak creative or poor targeting. It comes from the inability to test enough meaningful combinations to identify what actually works.
Testing capacity is a performance ceiling. When the system can only evaluate a few variations at a time, optimization becomes guesswork because there is not enough signal density to optimize against. Spend produces activity but not insight. Decisions get made on incomplete data and the account stays stuck against a wall it cannot see.
When testing scales, the wall comes down. Combinations that would have stayed hidden surface fast. Budget moves toward winners in real time. Performance stops being something you hope for and starts being something the system produces. That is what turned 916 property searches into 120,000.
