Google Ads counted forms. Sales needed qualified demos.
The operations SaaS company spent approximately €28,000 a month on Google Ads in France and the UK. Headline CPL looked acceptable, but sales had limited capacity and too many demos lacked fit or buying intent. The account rewarded form submissions even when the commercial team could not turn them into opportunities.
Repair the quality signal before bidding
Campaign identifiers and acquisition context were captured in the CRM. Sales and marketing agreed what counted as a qualified demo, then checked event timestamps and duplicate checks before imported outcomes could influence the account.
Sampled lead journeys exposed the difference between cheap submissions and accepted demand. That made search-query quality and downstream qualification part of the allocation review rather than judging every campaign by headline CPL.
Match searches, ads and pages to the buyer’s need
High-intent and exploratory searches were separated, with brand performance kept distinct so existing demand would not mask nonbrand inefficiency. Negative keyword coverage removed queries repeatedly associated with poor outcomes, including superficially relevant terms that did not produce useful sales conversations.
Role-specific landing-page variants matched the operational problem behind the search. Controlled tests examined that message continuity, while the account retained an interim conversion signal until qualified-demo imports were dependable enough for a deeper-funnel bidding pilot.
Review lead quality alongside advertising costs
Search-query quality, qualified-demo rate, sales acceptance and opportunity progression were reviewed together. A superficially relevant query could remain a poor investment if its downstream outcomes repeatedly failed, while a higher-CPL campaign could deserve budget when its accepted demand was more valuable.
The transition to qualified-demo bidding followed validation of the imported signal. That preserved learning volume while the deeper outcome accumulated and gave the team a pilot the team could evaluate rather than an account-wide change whose result would be harder to diagnose.
Teach the account what a worthwhile demo looks like.
The CRM signal needed a consistent definition, the correct acquisition record, a valid timestamp and protection against repeated imports. Bidding changed only after those checks held up.
| Validation layer | What needed to be checked | Why it affected the account |
|---|---|---|
| Commercial definition | Sales and marketing apply qualified demo consistently | Bidding should receive a meaningful quality distinction |
| Record relationship | Acquisition context belongs to the correct CRM outcome | Avoid rewarding the wrong campaign or journey |
| Timing | The event timestamp represents the intended outcome | Interpret delay and reporting progression coherently |
| Duplicate checks | The same qualified event is not repeatedly counted | Avoid inflated conversion counts and misleading costs |
| Learning readiness | Volume and delay permit a limited test using qualified outcomes | Protect delivery from an unsupported account-wide switch |
Qualified demos arrived less often and later than forms. Keeping an interim signal protected learning while the deeper outcome was tested.
Test qualified-demo bidding before a wider switch
Qualified demos were initially too sparse and delayed to support an immediate account-wide bidding change. I kept a stable interim conversion while validating identifiers, timestamps and duplicate checks, then introduced the deeper signal through a controlled pilot.
Measurement quality came before algorithm changes. Inconsistent sales qualification would teach the platform an unreliable distinction, while a premature switch could weaken delivery even if the chosen business outcome was conceptually correct.
Ninety days with flat spend
Within ninety days, cost per qualified demo fell from €420 to €285, approximately 32%. Demo-to-opportunity conversion rose from 18% to 27%, a 50% relative improvement, with approximately flat media spend.
These are combined before-and-after outcomes; landing-page tests, branded demand, sales response and seasonal conditions remain relevant to the interpretation.
The account learned from what sales accepted, once the CRM signal was dependable. Query control and landing-page relevance supported that change; deeper bidding came after measurement was ready.