
The Brand-Term CPA Creep: A Scenario SA Retailers Will Recognise
Picture a mid-sized South African e-commerce brand - homeware, R80m turnover - ranked #1 organically for its own name for three years. Brand paid search was cheap insurance. Then brand-term CPCs climb, and Auction Insights shows a marketplace and an affiliate bidding on the brand name. Platform ROAS still looks flattering because the ad clicks convert - but some would have arrived anyway.
This is the 2026 brand-search dilemma: not whether to buy your own name, but how much of it you need to rent. The answer is a query-by-query incrementality judgement backed by measurement that separates paid from organic contribution.
Why Brand Queries Stopped Being a Cheap Win in 2026
Three forces converged. AI Overviews now answer some branded informational queries directly, compressing clickable space. Broad-match expansion under AI Max means brand campaigns can quietly match non-brand queries unless brand inclusion and exclusion are configured correctly. And competitors and affiliates have learned that conquesting a known brand is cheaper than building demand. None of this makes brand spend useless - each raises the cost of getting the decision wrong.
SA brand-term CPCs are generally lower than UK/US equivalents, but trending upward as competitor bidding rises. Treat any figure as a range to measure in your own account.
What Actually Changed: AI Max, Brand Controls and Branded AI Overviews
AI Max changes the control surface: broad-match expansion reaches queries you never added, so brand inclusion and exclusion settings are strategy, not housekeeping. Incomplete brand exclusions let non-brand campaigns bid on your own name and inflate apparent brand CPA. Branded AI Overviews reduce paid click volume you can defensibly attribute, strengthening the case for incremental revenue over last-click. Audit brand inclusion lists quarterly and treat branded AI Overview impressions as an organic metric, not a paid failure.
The Cannibalisation Argument, Stated Honestly
Observational studies from Ahrefs, SEMrush and Search Engine Land repeatedly find paid and organic results on the same brand query largely overlap, suggesting much brand spend is cannibalistic. Google's own research argues genuine incremental lift exists. Neither is definitive: observational studies cannot control every confounder, and platform research has an obvious interest. Incrementality is real but site-specific - test it in your account.
Build Your Brand Query Taxonomy Before You Bid or Block
You cannot decide on the phrase 'your brand' alone. Break the brand universe into query groups by intent modifier.
| Brand query group | Intent modifier | Recommended action |
|---|---|---|
| Exact brand name | None | Test pause or holdout |
| Brand + product category | Category | Defend if organic weak |
| Brand + reviews/legit | Trust | Defend |
| Brand + coupon/discount | Price | Test, then likely pause |
| Brand + competitor | Comparison | Pause unless margin justifies |
Mapping Intent Modifiers: When 'Brand + Reviews' Is a Different Decision
Defensive Spend: Four Cases Where Brand Bids Still Earn Their Place
Brand bids are justified when they protect revenue that would otherwise leak to a third party: (1) competitors or marketplaces appear in Auction Insights on your exact brand query and displace your organic sitelinks; (2) you sell a considered or high-ticket product where trust modifiers dominate; (3) your organic result is an AI Overview summary omitting price or availability; (4) you need a controlled landing page for a promotion organic cannot mirror in time. The test: does the paid click produce revenue organic would not have captured?
Four Cases Where You Should Pull Brand Spend Back
Pull back when: (1) you hold a stable #1 organic result and no competitor appears on the exact brand query; (2) loose AI Max brand controls have you bidding your own brand inside non-brand campaigns; (3) affiliates bid your brand and you double-pay for the same sale; (4) the query is post-purchase or customer-service where paid adds no acquisition value. Each points to a measurable holdout, not a hunch.
Designing an Incrementality Test You Can Defend to Finance
Geo-split or time-split brand campaigns, hold the control steady, and compare total revenue (paid plus organic) between test and control. The metric that matters is incremental ROAS, not platform ROAS.
iROAS = (Revenue_test - Revenue_control) / Brand_ad_spend_test
# Revenue must include organic and direct, not just paid-attributed conversions.
# Use a holdout or paused-geo period of at least 4 weeks to reduce noise.Warning: platform-reported conversions inside a brand campaign are not incremental revenue - they include clicks that would have converted organically. Only a holdout or pause test produces a defensible iROAS figure.
The Measurement Layer: Server-Side Data, Consent Mode v2 and Brand Segments
Incrementality testing depends on clean pipelines. Server-side tagging reduces lost conversions from ad blockers and browser restrictions; Consent Mode v2 models consent-safe conversions where users decline cookies. In GA4, create a brand versus non-brand segment so revenue and conversion rates compare without blending. Roll up to MER and blended CAC so brand spend is judged against total business revenue. POPIA and GDPR require consent-safe joins that do not expose personal data - keep the brand segment aggregated.
Decision Rules: Scoring Each Brand Query Group on Incrementality and Risk
Score each group on expected incrementality and competitive risk (share of impressions where a competitor or affiliate appears), then apply a default action.
| Query group | Expected incrementality | Competitor share | Default action |
|---|---|---|---|
| Exact brand | Low | <10% | Pause / test |
| Brand + reviews | Medium | 10-30% | Defend |
| Brand + category | Medium | >30% | Defend with capped bids |
| Brand + coupon | Low | Low | Pause |



