A B2B SaaS Paused Brand Search for 30 Days. Sign-ups Barely Moved.
A common belief: brand search ads protect you from competitors. We ran the test.
A Series B SaaS company was spending roughly $18,000/month on Google Ads bidding on its own brand name. The argument was familiar: “If we don’t bid, a competitor will, and we’ll lose customers.”
We helped them run a 30-day geo-holdout test across 8 matched DMA pairs. Brand-search ads were paused entirely in the holdout DMAs while nonbrand and other channels stayed live.
What we measured
- Direct sign-ups from organic search and direct traffic in each DMA
- Total qualified leads (form fills + demo requests)
- Net new ARR attributed to each region
The result
In the holdout DMAs, where brand search was completely off:
- Sign-ups dropped by 6%, not the 40-60% the team expected
- Of that 6% drop, organic search captured roughly two-thirds of the lost paid clicks
- Net incremental impact of brand search: ~2% of conversions
In dollar terms, the team was paying about $18,000/month for $400 worth of incremental sign-ups. Cost per truly incremental signup was over $900, vs. the $35 the platform reported.
What about competitors?
This is the question everyone asks. The team monitored auction insights during the test. Two competitors did briefly start bidding on the brand term in the holdout DMAs — and got almost no clicks. Brand-loyal users scrolled past the ads to the organic result.
“We were essentially paying Google to run defense against a threat that didn’t exist.” — VP of Growth
What they did next
- Cut brand search budget by 80% (kept a small spend for new product launches and competitor incursions)
- Reallocated $14k/month to nonbrand search and a YouTube test
- Re-ran the test 90 days later to confirm the result held. It did.
The lesson
“Brand search converts well” is not the same as “brand search drives incremental revenue.” Conversion rate measures who clicked the ad on their way to converting. Incrementality measures who would not have converted without it.
The two numbers can differ by an order of magnitude. They did here.