What this checkpoint covers
Northbank Removals is an invented removals firm: customers request a quote online, a surveyor visits or calls, and the move is booked weeks later. The questions draw on the enhanced conversions guide, the guide to connecting a CRM to ads and analytics and the target CPA versus target ROAS comparison.
Task
Write a short answer and a reason for each question. Count an answer as right only if the reason matches. For each miss, reread the linked page.
Questions
- Northbank turns on enhanced conversions for leads. The developer suggests sending the hashed email with every quote request, including from visitors who declined advertising cookies, because hashing makes it anonymous. What do you say?
- Offline imports by click identifier match far fewer booked moves than the CRM records. Most unmatched leads arrived on a landing page and submitted the form on the pricing page. What is the likely cause, and the fix?
- The account manager switches bidding to the booked-move stage. A fortnight later the last few days show almost no conversions and she wants to switch back. What do you tell her?
- Northbank can import quote requested, survey completed and move booked. Which should be primary for bidding, and what happens to the other two?
- Booked moves range from small flat moves to office relocations worth many times more. The account runs a cost target on surveys. Should it move to a return target, and what has to be true first?
- An analyst proposes importing booked moves into Google Analytics (GA4Google's event-based analytics platform.) as a key event and then into Google Ads, alongside the existing direct import from the CRM. What is the risk?
- After a month of imports, the Google Ads column shows more booked moves than the CRM has for the same period. What do you check?
- Enhanced conversions show a recording status, but conversion counts barely moved. The owner calls the project a failure. Is it?
Answers explained
- No. Hashing reduces risk; it does not replace consent. Denied consent should stop both the conversion and the enhanced data, and the privacy notice must describe the sharing. See consent and privacy requirements.
- The identifier was lost between landing and submission. It must be stored in first-party storage on arrival and read back into a hidden field at whichever form is submitted, with a lifetime at least as long as the click window. Test with a real click and read the CRM record. See capturing identifiers at the form.
- Do nothing yet. Conversions are credited to the click date and booked moves arrive weeks after the click, so recent days are always incomplete. Read the change by conversion time or exclude open windows, and let the strategy finish learning; switching back restarts it. See the note on conversion lag in the comparison of cost and return targets.
- The completed survey, in most cases. It is the latest stage that still arrives within weeks and in enough volume to learn from; booked moves are the truth but arrive late and rarely. Quote requests and booked moves stay secondary, for volume and for checking that survey values still predict revenue. See which stage to send.
- Yes, eventually. Under a cost target every survey counts the same, so bidding cannot favour the office moves that pay most. A return target needs values that reflect margin by job type and enough valued conversions to learn from. Validate the values first, change the goal and the target in separate steps, and start the ROASRevenue generated per unit of ad spend. target near the achieved return rather than the ideal. See the decision table on target choice.
- Double counting. Two primary actions describing the same booked move count it twice and inflate whatever the bidding sees. Keep one route primary, usually the direct import the account matches to the click, and leave the other secondary for reporting. See the analytics routes in offline conversion imports.
- Duplicates and definitions. Look for the same lead uploaded by both the click identifier and the hashed email route, re-sent rows from a failed upload, counting set to every instead of one, and a CRM report cut by a different date. Reconcile monthly and track the unmatched share, as closed-loop reporting describes.
- Not necessarily. Recording means hashed data arrives, not that extra matches were made. The gain depends on what was being lost: long delays, strict browsers, cross-device journeys. Judge it on the actions and devices where the click identifier was weakest, before and after, and keep the cost target's CPACost of each conversion. view stable while you do. See validating enhanced conversions.
Where to go next
If you missed the stage or target questions, redo the conversion values exercise for a solar installer. If the delay question caught you out, the next module, which starts with the geo split bidding exercise, builds on it. To revise the foundations, return to the intermediate Google Ads level.
