Budget-Limited Shopping, Thin Creative, 31% Brand Revenue
An online furniture retailer spends $18,000/month on one standard Shopping campaign with tROAS 450%. Over the last 90 days it returned 520%, but the campaign was budget-limited on 19 of 90 days and holds 68% search impression share on non-brand queries. About 31% of attributed revenue comes from queries containing the brand name. The feed holds 4,800 SKUs; the top 300 SKUs produce 72% of revenue. Leadership wants to replace the campaign with Performance Max next month, using four asset groups split by room type (living, bedroom, outdoor, storage). Creative is thin: five lifestyle images, no video, and product copy reused verbatim from the site. There are no brand exclusions configured and no account-level negative keywords in use.
1. Assess whether a full switch to Performance Max next month is justified, using the numbers given.
Read the worked answer
What supports a switch: the campaign is budget-limited on 21% of days and sits at 68% non-brand impression share, so there is headroom that a single standard Shopping campaign may not be capturing. 72% of revenue from 300 SKUs also suggests the feed has a long tail that is under-served.
What argues against switching next month:
- Attribution risk. 31% of attributed revenue is brand-query revenue. With no brand exclusions, Performance Max can serve brand traffic and book conversions that standard Shopping was already getting. A 520% baseline that is partly brand inflates the comparison and can make a cannibalising launch look successful.
- Creative readiness. Four asset groups against five images, no video and duplicated product copy means each asset group is starved of distinct assets. Google will lean on auto-generated assets, which lowers control exactly where the budget is going.
- Measurement. Going from one campaign to one campaign gives you no clean counterfactual. If results move, you cannot attribute the change.
Verdict: the direction is defensible, the timeline is not. Fix brand exclusions and account-level negatives, add per-room creative, and run Performance Max alongside the standard campaign before considering replacement.
2. Design a test structure and guardrails that would let you measure incremental value without destroying the 520% baseline.
Read the worked answer
Structure
- Keep the standard Shopping campaign running at a reduced but non-trivial budget (e.g. 60–70% of current) so it remains a valid baseline and still captures the headroom.
- Launch Performance Max as a separate campaign with a fixed budget (e.g. $5,000/month), four asset groups, and its own tROAS set slightly below the baseline (e.g. 450%) so it can gather data.
- Split by geography if the markets are separable: Performance Max in matched test regions, standard Shopping in control regions, for at least 4–6 weeks. A simple budget split within the same geography cannot separate cannibalisation from incrementality.
- Pre-register the decision rule: success = incremental non-brand revenue and conversions at or below baseline CPA, not blended ROAS.
Guardrails
- Brand exclusions applied from day one; account-level negative keywords for irrelevant and competitor terms.
- Turn off automatically generated assets or review them before launch; keep the asset groups fed with room-specific imagery.
- Report brand vs non-brand, new vs returning customers, and non-brand impression share, not just ROAS.
- Kill criteria: if non-brand impression share falls or total non-brand conversions drop while spend rises, pause the test.