YourGoogleAdsAccountIsProbablyCompetingAgainstItself
There's a structural problem running through most DTC Google Ads accounts that doesn't show up in dashboards, doesn't trigger alerts, and rarely gets flagged in agency reports. Campaigns bidding against each other for the same queries. Brand Search and Brand Shopping in the same auction simultaneously. PMax quietly absorbing branded demand that a dedicated brand campaign should own. Generic broad match competing for the same non-brand terms as PMax, with no mechanism to tell them apart.
We call it bid stacking. The effect is a 40-50% artificial inflation of CPCs - paid entirely by the advertiser, to nobody's benefit except the auction. It's one of the most common structural problems we find in accounts, and because it doesn't produce an error message, most brands have been paying it for months or years without knowing.
It's also entirely fixable. But fixing it requires being honest about how these accounts actually work, which means having a different conversation than most agencies are willing to have.
The structure problem
The standard DTC Google Ads architecture - brand, generic, shopping, PMax, retargeting - is right in theory. The problem is in the execution. When campaigns don't have clearly defined query ownership, and when there's no cross-campaign negative framework enforcing those boundaries, the structure becomes a suggestion rather than a system.
PMax is where this gets most expensive. PMax doesn't show you which queries it's serving in any useful detail. It reports channel-level splits - Search, Shopping, Display, YouTube - but not the query-level data that would tell you whether it's harvesting branded demand that your Brand Search campaign should own. The signal that something's wrong isn't an error. It's Brand impression share declining while PMax ROAS looks strong. By the time that pattern is visible, budget has already drifted in the wrong direction.
The fix is a dual exclusion: a brand exclusion list at account level, plus campaign-level negatives on core brand terms inside PMax. The exclusion list alone isn't enough - it catches exact brand terms but leaves long-tail branded queries exposed. Both layers together close the gap.
The same logic applies across the whole account. Generic Search needs a weekly search query report enforcement gate. Without it, broad match keywords drift into PMax territory within weeks, attribution becomes unreadable, and you lose the ability to tell which campaign is actually driving anything. Data-driven attribution helps, but DDA distributes credit across whatever structure you give it. If the structure has overlapping campaigns chasing the same queries, DDA distributes credit across the overlap. It doesn't fix it.
The measurement problem
Most DTC brands we work with are optimising against numbers that are partially wrong.
GA4 typically overclaims revenue relative to Shopify. The gap varies by account, but it's rarely trivial - in some cases it runs to five figures a month. The practical consequence is that every POAS target, every ROAS benchmark, every Smart Bidding signal in the account is calibrated against a number that includes revenue GA4 is inventing. The account looks like it's performing better than it is, until a founder asks why the Google Ads numbers don't reconcile with what finance is seeing in Shopify.
Shopify is the ground truth. Performance targets should be set against Shopify revenue, not GA4, and this should be agreed in writing before any engagement starts. The conversation about which number you're being held to is uncomfortable to have. It's considerably more uncomfortable to have it after three months of reporting against the wrong one.
Server-side tracking - implemented properly, with GCLID passback verified - is what closes the gap. But more importantly, it's the foundation that everything else depends on. The customer state model, the incrementality tests, the suppression framework - none of it is reliable until the conversion data is clean. The sequence is non-negotiable: fix the tracking first, establish a baseline, then build.
The social-to-search question
Here's the conversation most performance agencies avoid: for brands with strong social and influencer presence, non-brand search may not be driving new demand. It may be capturing demand that social already created.
A user discovers a brand on TikTok. They watch three videos, follow the account, see an influencer they trust using the product. Two weeks later they search generically - "eyebrow gel", "brow sculpt" - and click a paid search ad. Google's reporting attributes the conversion to paid search. The ROAS looks strong. But the demand existed before the search happened. The search was navigation, not discovery.
This doesn't mean generic search has no value. A brand with strong organic visibility might still benefit from paid non-brand coverage - protecting against competitor intercepts, capturing high-intent queries, accelerating conversions that would otherwise take longer. But the value is as a conversion accelerator or competitive defence, not as a demand creation channel. And the only way to know which it is for your brand is to test it.
A geo holdout - pausing non-brand spend in matched regions and measuring Shopify revenue per visitor in holdout versus control - is the only measurement methodology that gives you a credible answer. Not Google's Conversion Lift tool, which is Google marking its own homework. A properly designed holdout, run for a minimum of six weeks against a clean baseline, measured against the source of truth. Most brands don't run it because they're not sure they want to know the answer. The ones that do tend to find the result changes where they put money.
The customer definition problem
New versus existing customer is the right instinct but the wrong resolution for most DTC brands with retail presence.
Someone who has bought three times from a department store or a beauty retailer isn't a prospect. They're a brand loyalist who hasn't yet bought direct. Their acquisition economics in a paid search campaign are completely different to someone who has never encountered the brand. Treating them as the same category makes acquisition cost look better than it is - you're averaging the effort of genuinely cold acquisition with the relatively easy job of converting an already-convinced buyer into a direct customer.
The distinction that matters is new to brand versus new to DTC. A customer state model that separates these two - along with active customers, replenishment buyers, and lapsed - gives you a different performance target for each segment, a different suppression and targeting logic for each campaign tier, and a reporting framework where blended CPA is context rather than the optimisation target.
Customer Match is the mechanism. Multi-key matching (email, phone, postal address combined) achieves 50-70% match rates. Email-only is 40-50%. That means 30-50% of existing buyers remain invisible to suppression lists and will see prospecting ads regardless. Layering a pixel-based audience on top closes some of that gap. Full suppression isn't achievable. The goal is to minimise leakage, document it transparently, and never let it become a surprise in a quarterly review.
Where Google Ads actually sits
The most important shift in how we think about paid search is this: Google Ads no longer sits in marketing. It sits at the intersection of marketing, operations, and finance - and accounts that are built without input from all three are built on assumptions that eventually break.
The retail versus DTC channel question can't be answered by a media planner. It requires contribution margin data from finance and operational context on fulfilment, returns, and customer lifetime value. The POAS targets that determine whether an account is profitable can't be set without COGS data, regional margin variation, and a clear view of what first-order profitability needs to look like for the business to scale.
The brands that get the most from paid search are the ones whose leadership treats it as a commercial infrastructure question rather than a marketing spend question. The brief isn't "grow our Google revenue." The brief is "build a paid search system that fits how this business actually makes money, and prove that it's working."
That's a harder brief to answer. It requires more from the agency and more from the client. But it's the only brief worth taking seriously.
JudeLuxe is a performance marketing consultancy specialising in paid search for DTC brands. We work with founders and CMOs who want to understand what their Google Ads account is actually doing to their margin.