Final URL Expansion and Brand Cannibalization at Copperline Home
Copperline Home has 6,000 SKUs across kitchen, decor, and storage, and runs a single Performance Max campaign with final URL expansion enabled, $9,000/day, tROAS 450%, and no account-level brand exclusions. Over the last 30 days, 28% of clicks landed on non-product pages: blog recipes, store locator, and support articles. Those pages converted at 0.6%, versus 3.4% for product detail pages. Meanwhile, 35% of impressions and 41% of conversions are on queries containing the brand name, which also has its own Search campaign. The team reads the data as evidence that URL expansion is 'finding new audiences' and that brand traffic inside Performance Max is free incremental volume. Last month, non-brand conversion volume fell 12% while total spend rose 8%.
1. Diagnose what the URL expansion and brand numbers actually mean. Use the right denominator.
Read the worked answer
URL expansion. Expansion is designed to send traffic to pages beyond the feed landing page when those pages are likely to convert. Here the marginal pages convert at 0.6% against 3.4% on product detail pages, so at 28% of clicks they are consuming a large share of spend for a small share of conversions. The team's 'new audiences' reading is a clicks-based story; the business question is cost per conversion by landing page group. If blog and support pages produce near-zero conversions, they are pure cost, not discovery. The right check is spend and conversions by landing page, not click share.
Brand. Forty-one percent of conversions on brand queries almost certainly overlaps the brand Search campaign. Performance Max can bid on brand terms cheaply and does so readily when tROAS is the target, which makes campaign ROAS look strong while adding little incremental volume. Combined with non-brand conversions down 12% on higher spend, the likely picture is budget shifting toward cheap brand clicks and low-intent expanded URLs, away from non-brand product demand. That is cannibalization plus waste, not growth.
2. What actions would you take, in what order, and how would you verify that removing brand from Performance Max did not cost you volume?
Read the worked answer
Sequence matters, because the two problems interact.
- Exclude the clearly non-converting URLs first. Add URL exclusions for blog, support, store locator, and cart. This is low-risk and immediately frees budget. Leave category pages eligible only if their cost per conversion is acceptable once measured separately.
- Apply account-level brand exclusions so Performance Max stops competing with the brand Search campaign, and add brand negatives to prevent re-entry.
- Hold tROAS constant for two weeks before judging, so results are not attributed to a simultaneous bid change.
- Watch non-brand conversion volume and total spend, not campaign-level ROAS, since removing brand will mechanically lower reported ROAS while potentially improving total efficiency.
Verification. Use a geo experiment or a campaign draft experiment: hold brand exclusions in test regions and keep the current setup in control regions, then compare total account conversions and total cost, not just Performance Max numbers. If total conversions hold within normal variance and total cost falls, the brand traffic was cannibalized. If total conversions drop, reintroduce brand selectively with a separate low-target campaign rather than folding it back into Performance Max.
3. The team argues that URL expansion should stay on because it may uncover future high-performing pages. How do you weigh that against the current data, and what would change your mind?
Read the worked answer
The argument is legitimate in principle - expansion exists precisely to find converting pages the feed does not cover, such as buying guides or category pages that answer a query better than a product page. The issue is that the current evidence does not support it: 28% of clicks at 0.6% conversion is a large share of spend on a poor conversion rate, and the account is simultaneously showing non-brand conversions down 12% on 8% more spend. When the aggregate trend is deteriorating, optionality is not free.
The way to keep the upside without paying for the downside is to narrow eligibility rather than switch expansion off wholesale. Exclude the pages that cannot convert by their nature - support, blog, store locator, cart - and let category and guide pages remain if their cost per conversion is acceptable.
What would change my mind: page-group data showing that expanded URLs, once separated from product pages, achieve cost per conversion within an acceptable band of product pages, or evidence that expanded-page visitors convert on a later session at a rate that justifies the click cost. Absent either, default to restricted expansion and revisit quarterly with the same page-group report.