A Lead-Gen Account That Is Optimising Toward the Wrong Conversion
A home-security installer uses target CPA of $60 across Search. Three conversion actions are live: "Request a quote" form (primary), phone calls over 60 seconds (primary), and "pricing page view", imported from analytics and set as primary last month. Reported CPA is $42 with 480 conversions; the sales team says about 30% of leads never answer a call, and average sale value has dropped. Since the pricing-page action was added, reported conversions rose 40% and the bid strategy has pushed spend toward a broad "home security" keyword set. CRM data with lead quality and closed-won value exists but has never been imported.
1. Diagnose what changed and explain the mechanism by which it damaged performance.
Read the worked answer
Adding a mid-funnel page view as a primary conversion action changed what the bid strategy is buying.
- A pricing page view is cheap, common and happens before any commitment. It makes the account look efficient ($42 CPA) while the strategy chases volume of shallow events.
- Mixed-cost, mixed-value actions in one count mean the algorithm cannot distinguish a quote request from a browse. It optimises toward the cheapest, highest-frequency action - hence the 40% conversion rise and the drift to broad "home security" queries.
- The symptoms line up: more leads (sales sees 30% uncontactable), lower average sale value, and cheaper reported CPA. All three are consistent with a conversion definition that rewards low intent.
The page view should not be primary. If it is kept at all, mark it secondary - observable but not used for bidding - and confirm the quote form and qualified calls are the only primary actions.
2. How would you rebuild measurement so bidding optimises for business outcomes rather than form fills, and what would you sequence first?
Read the worked answer
Sequence matters: fix the conversion definition before touching targets, or you will tune a broken signal.
- Clean the action set. Primary: quote form, qualified call. Secondary: pricing page view, brochure download. Verify call tracking records duration and, ideally, qualified-call status.
- Import CRM outcomes. Upload lead status and closed-won value keyed to a click identifier. This lets you bid on qualified leads or value rather than raw forms, and gives you a lead-quality metric the sales team will recognise.
- Re-baseline. Run a stable period at the current $60 target with the corrected action set before changing anything, so you can attribute the effect.
- Then consider value-based bidding if closed-won value varies widely by segment - a security install with monitoring has different lifetime value than a one-off hardware purchase.
Expect reported CPA to rise and lead quality to improve. Warn stakeholders in advance so nobody reads the correction as a decline.
3. How would you prove to a sceptical sales director that the change worked, given that reported conversions will fall?
Read the worked answer
Do not argue about CPA. Argue about leads that reach an appointment and close.
- Pre-commit to outcome metrics: contacted rate, appointment rate, closed-won rate and revenue per lead, by campaign and keyword theme. Fix the measurement window before launch.
- Compare matched periods with the same offer and season, since home security has strong seasonal demand. If possible, run a geo holdout or a campaign experiment so the comparison is not just before-versus-after.
- Show the trade explicitly: if qualified leads rise from, say, 340 to 380 while raw form fills drop, the account is better even though reported CPA worsened. That is the whole point.
- Tie it back to the bidding shift: demonstrate that spend moved out of broad "home security" queries toward terms that produce contactable leads, using search terms report plus CRM-matched quality.
Agree the success criterion with the sales director in writing before the change, so the result cannot be reframed afterwards as a CPA debate.