The most comfortable Google Ads account we ever audited reported a 9.4x ROAS. It was also, functionally, doing almost nothing. Eighty-one percent of spend sat on branded search — people typing the company's name into Google and clicking the ad above the organic listing they were about to click anyway. The account was not acquiring customers. It was intercepting them, and billing the interception as growth. This is the trap every established brand's account drifts into, and getting out requires a deliberate google ads non-branded search strategy — one that treats brand and non-brand as different businesses with different jobs, different economics, and different reporting.
We manage Google accounts from $8K to $80K/mo. The pattern below is what we apply when a mature account has plateaued behind flattering blended numbers.
Why branded-heavy accounts feel great and stagnate
The mechanics are simple. Branded clicks are cheap ($0.30–2.00 against $2–15 for non-brand in most verticals), convert at 3–8x the non-brand rate, and get last-click credit for demand that was created elsewhere — by your product, your sales team, your Meta spend, your PR. Blend that into one account-level ROAS and you get a number that makes everyone comfortable.
The stagnation is visible only when you decompose:
- Branded volume is capped by branded demand. You cannot bid your way to more people searching your name. If branded impressions grow 4% year over year, that is your ceiling on 81% of the budget.
- Blended ROAS punishes growth. Every incremental non-brand dollar drags the blended number down, so the account "optimizes" by starving the only campaigns that acquire new customers. Smart Bidding on a shared target accelerates this: give tROAS a mixed portfolio and it will shift spend toward brand every time.
- The dashboard rewards the wrong behavior. A buyer judged on account ROAS is structurally incentivized to keep the brand share high. Nobody is lying; the measurement design does the lying for them.
The tell in an audit takes five minutes: segment 12 months of spend and conversions by brand vs. non-brand query. If brand is above 50% of spend in an account that is supposed to be a growth channel, you have found the plateau's cause.
Step one: split the account and split the targets
Everything else depends on separation. Brand and non-brand must live in separate campaigns with separate budgets, separate bid strategies, and — critically — separate reported targets.
- Isolate brand. Exact and phrase match on brand terms and common misspellings, with brand terms added as negatives everywhere else. Verify with the search terms report monthly; broad match and PMax will both leak brand traffic back into "non-brand" campaigns and re-inflate their numbers silently.
- Set asymmetric targets. Brand campaigns get an efficiency target (impression share on brand terms, capped CPC, modest budget). Non-brand campaigns get a growth target expressed in CAC or incremental ROAS, benchmarked against what a new customer is actually worth — usually via contribution margin and repeat rate, not first-order revenue.
- Report them separately, permanently. The monthly report shows two lines. On one B2B account, this single change reframed the conversation from "our Google ROAS is 7x" to "we acquire new pipeline at $210 per SQL and defend brand for $900/mo" — and the second sentence is the one a CFO can act on.
Expect the non-brand number to look worse than anything leadership has seen from Google. It was always this number. It was hidden.
The honest incrementality question on brand terms
Before scaling non-brand, settle how much your brand spend is actually buying. The uncomfortable research consensus — going back to eBay's 2013 large-scale experiments — is that branded search incrementality is often low: many of those clickers would have arrived through the organic listing one position down.
Low, but not zero, and it varies by situation. Brand bidding earns its budget when:
- Competitors bid on your name. If a rival's ad sits above your organic result, defense is cheap insurance. Check the Auction Insights report on your brand campaign; if no one is there, ask what you are defending against.
- Your organic listing is weak for the query — a category page outranking the right landing page, or an unresolved reputation result nearby.
- You need message control — a promotion, a repositioning, sitelinks the organic snippet cannot carry.
The test is straightforward and we run it on most mature accounts: geo-split brand campaigns, turning them off in matched market groups for 3–4 weeks, and measure total conversions (paid + organic) per market. Typical findings in our accounts: 15–40% of branded paid conversions are truly incremental, with the high end in contested categories. On one account, the test let us cut brand spend from $6.5K to $2K/mo with a measured total-conversion loss of under 3% — freeing $4.5K/mo that went straight into non-brand testing. That reallocation, not any clever optimization, was the quarter's biggest lever.
Building non-brand: intent clusters, not keyword dumps
Non-brand search fails most often because it is built as one undifferentiated keyword list feeding one landing page. We build it as intent clusters — groups of queries that share a problem state and a stage of decision — because bid strategy, ad copy, and landing page should all change with intent, and clusters are the unit at which Smart Bidding gets clean signals.
The clusters, in descending order of intent:
- Category purchase intent ("buy X", "X for Y", "X near me"). Highest CPC, shortest payback. Start here; this is where the first reallocation dollars go.
- Comparison and alternatives ("X vs Y", "best X for Z", "[competitor] alternative"). Mid-funnel, exceptional economics when the landing page actually compares instead of redirecting to the homepage. On a SaaS account similar to our B2B SaaS case, alternative-queries delivered SQLs at 60% of the category-term CPA.
- Problem-aware queries ("how to fix X", "why does X happen"). Cheap clicks, long payback. Only fund these once the first two clusters are saturated, and only with a lead-capture step in the path.
Operating rules that keep clusters honest: one cluster per campaign (or per tightly themed ad group), broad match only when paired with tROAS/tCPA and a mature conversion history, a weekly search-terms pass with aggressive negatives for the first 6–8 weeks, and landing pages matched to the cluster's question — comparison traffic to comparison pages, not the homepage. The full build sequence is what our Google Ads service runs as standard; the mistake to avoid is launching all clusters at once and diluting the learning data across all of them.
PMax guardrails so it scales non-brand instead of eating brand
Performance Max belongs in this strategy — it is often the cheapest source of incremental non-brand volume for e-commerce — but unguarded, it does exactly what the blended account did: harvest brand and retargeting demand and report it as prospecting.
Non-negotiable guardrails:
- Brand exclusions on, always. Apply account-level brand lists to every PMax campaign. Verify in the search terms insight; exclusions leak on close variants.
- Judge PMax on account-level incrementality, not its own dashboard. The test: did total non-brand conversions rise when PMax spend rose, or did Shopping/search conversions fall by a matching amount? We run a 4-week before/after on this for every PMax launch. Roughly a third of the time, early PMax "performance" is pure cannibalization.
- Feed it your real objective. New-customer bidding modes and cart-margin values where available; otherwise PMax happily buys cheap repeat purchases.
- Keep exact-match search campaigns on your proven money terms alongside PMax. When both compete, exact match wins the auction internally and keeps your best queries under explicit control.
Offline conversion import: the B2B unlock
For lead-gen and B2B accounts, one intervention outranks everything above: stop optimizing to form fills. A form fill is not revenue; in most B2B funnels, 60–90% of raw leads never become pipeline. Smart Bidding pointed at form fills dutifully finds more of the cheap 90%.
Offline conversion import (via GCLID/enhanced conversions from the CRM) closes the loop: qualification events, opportunities, and closed revenue flow back into Google as the optimization signal, typically with values attached and a 30–90 day lag the system handles natively. The effect is not subtle. On a B2B services account spending $28K/mo, switching the bid signal from leads to CRM-qualified opportunities raised cost-per-raw-lead 22% — and cut cost per opportunity 47% over two quarters, because the system stopped buying students, job-seekers, and tire-kickers who fill forms beautifully.
Implementation is a plumbing project, not a marketing project: consistent GCLID capture on every form, a CRM field mapping, a daily upload job, and a fallback for leads that arrive by phone. Budget 2–4 weeks of setup. It repays that in the first quarter, and it also fixes the reporting layer — the same closed-loop logic that post-iOS attribution forced onto every other channel. If you want the brand/non-brand decomposition done on your own account before committing to any of this, that split is the first deliverable of our audit.
The sequence, compressed
For a branded-heavy account, the order of operations matters more than any single tactic: split brand from non-brand and set separate targets; run the brand incrementality geo-test and reallocate what it frees; build non-brand as intent clusters, highest intent first; add PMax behind brand exclusions and judge it on incremental account-level volume; and, for B2B, wire offline conversions before scaling anything. Six months of that sequence converts a 9x-ROAS interception machine into a smaller-looking number that is actually a growth channel — which is the trade every established brand eventually has to make on purpose, because no dashboard will ever suggest it.