Pinning, Ad Strength, and 600 Impressions a Day
A project-management SaaS runs one Responsive Search Ad with 15 headlines and 4 descriptions. Three headlines are pinned to position 1, one per persona, and the team reviews "Ad strength" weekly; it currently reads "Average". The ad's CTR is 4.2%, the best in the account, and cost per trial is $88 against a $70 target. Legal requires the phrase "Free 14-day trial" to appear in a headline whenever pricing is referenced in a description. The team wants to unpin everything so Ad strength rises to "Excellent", and also wants to fold the new "AI assistant" feature into the same ad. Traffic is about 600 impressions per day.
1. Should the team unpin the three headlines to improve Ad strength? Argue the decision with the evidence available.
Read the worked answer
No - not on that reasoning. Ad strength is a completeness and diversity signal, useful as a checklist, not a performance forecast. The account has actual evidence: 4.2% CTR, best in account, and a cost per trial of $88 against a $70 target. Unpinning to move a label from Average to Excellent optimises the wrong number.
Pinning is justified when there is a reason to control the message: persona routing, legal requirements, or a compliance phrase that must appear. Legal already forces partial control here. The real problem is the $88 trial cost versus a $70 target - that is a landing page, offer or keyword-quality issue, and unpinning will not fix it.
A defensible position: unpin the two persona headlines that carry no compliance or routing duty, keep the one that must govern a legal claim, and judge the change on cost per trial and trial-to-paid rate - not on Ad strength.
2. What would you do about the request to add the AI assistant feature to this ad, given the volume of traffic?
Read the worked answer
Adding a fourth message to an ad with 600 impressions a day and three existing pinned variants will produce no readable result and dilutes the message that currently earns 4.2% CTR.
Better options:
- Keep the test unit small and clean. Duplicate the ad as a challenger in the same ad group, swapping the weakest asset for the AI assistant headline, and let the existing ad hold. Do not mix two changes (unpinning plus a new feature) in one step or you learn nothing.
- Consider whether this belongs in the same ad group at all. If the AI assistant is a distinct use case with its own query intent, it may deserve its own ad group and landing page rather than competing for a slot in an ad that already converts.
- Set a decision threshold in advance. With roughly 18,000 impressions a month and a 4.2% CTR, you are looking at hundreds of clicks, not thousands; define what difference would be believable before you start.
3. Given roughly 600 impressions per day, how would you design and judge an ad test?
Read the worked answer
At this volume, a conventional A/B on small CTR differences will never reach significance before the offer or the market changes.
- Estimate first. Around 18,000 impressions and roughly 750 clicks a month is enough to detect a large difference in trial starts, not a 0.3-point CTR move. Test offers and messages, not punctuation.
- Use the metric that pays. Cost per trial and trial-to-paid rate, not CTR. A higher CTR with worse trial quality is a loss.
- Run one change at a time and hold it long enough to cover a full sales cycle; trials take days to convert, so early results will flatter the challenger.
- Sequence rather than parallelise if traffic is thin: run the challenger for a defined period, then revert, and compare periods with matched promo and seasonality.
- Write down the decision rule before launch: what lift, over what window, would justify keeping the change.