Platform
Google, the biggest and least forgiving auction
The largest and least forgiving auction on the internet. How we build campaigns that survive the shift to black box automation, protect your margin, and stop Google from spending your budget on irrelevant queries.
The landscape
Scale and the loss of control
Google Ads remains the most reliable demand capture engine available. When someone types a high intent query into a search bar, they are practically raising their hand and asking to be sold to. The trade off is that Google knows exactly how valuable this intent is, and the auction pricing reflects it.
Over the last five years, the platform has fundamentally changed how it sells this inventory. The era of manual bid adjustments and strict keyword match types has been steadily replaced by algorithmic bidding and consolidated campaign types. Google's objective is to maximize total auction liquidity. Your objective is to buy only the clicks that convert profitably. These two goals are fundamentally at odds.
Managing Google Ads today is an exercise in building guardrails. It requires a firm hand on negative keyword lists, a deep understanding of how to train automated bidding models with accurate conversion values, and the discipline to turn off the platform features that optimize for spend rather than return.
Performance Max
What you give up, and how to claw it back
Performance Max (PMax) represents the extreme end of Google’s automation push: a single campaign type that sprays budget across Search, Shopping, YouTube, Display, and Maps based on algorithmic whim.
The trap with Performance Max is the illusion of simplicity. Because it requires very little setup, advertisers launch it, see some initial conversions, and assume the job is done. But underneath the hood, PMax is often cannibalizing existing brand search traffic, taking credit for returning customers, and burying wasted spend in opaque placement networks.
You lose visibility into search term data, and you lose the ability to exclude specific placements easily. To make PMax work safely, you have to constrain it. We do this by feeding it highly restrictive audience signals, explicitly excluding brand terms so it is forced to find net new customers, and heavily segmenting the asset groups by margin rather than by product category.
Default PMax exclusions
Before we let a Performance Max campaign run, we apply an account level negative framework to protect the budget:
- Account level brand exclusions to prevent cannibalization
- Suppression of all known mobile app categories
- Exclusion of poor performing Display network categories
- Negative keyword lists applied via Google rep intervention
The Auction
Broad match and algorithmic bidding
The pairing of broad match keywords with value based bidding is the platform’s current ideal state, but it only works if your measurement layer is flawless.
Historically, paid search relied on exact and phrase match types to tightly control which queries triggered an ad. Broad match was considered a budget trap. Today, Google's machine learning evaluates thousands of signals (time of day, device, past browsing history) to predict conversion probability on broad match queries.
When paired with a bidding strategy like Target ROAS (Return on Ad Spend), broad match can genuinely outperform manual setups. But there is a massive caveat: the algorithm is entirely dependent on the quality of the conversion data you feed it. If you are tracking trivial actions like newsletter signups or page views as primary conversions, the algorithm will ruthlessly optimize your budget to deliver those low value actions at scale.
To succeed with modern Google Ads, the account structure must mirror your business economics. We structure campaigns around conversion value and profit margin, ensuring the algorithm optimizes for actual revenue, not just arbitrary conversion counts.
-
01
Clean the measurement layer
We implement server side tracking and offline conversion imports to ensure Google sees actual closed deals and cleared transactions, not just form fills.
-
02
Isolate match types by intent
High value, exact match queries are kept in dedicated campaigns to ensure they always receive sufficient budget and are not choked out by broad match exploration.
-
03
Set realistic bidding targets
We calculate your break even ROAS and set algorithmic targets that allow the system enough breathing room to enter auctions without destroying your margin.
Measurement
Attribution quirks and the honest verdict
Google Ads is heavily incentivized to claim credit for as many conversions as possible.
By default, Google Ads uses data driven attribution, which looks at the entire path a user took before converting. While this is mathematically sound, it often results in fractional conversions (like 0.4 of a sale) appearing in your dashboard. Furthermore, the platform defaults to a 30 day post click window and a 3 day post view window.
This means if someone clicks an ad, browses your site, leaves, and comes back 28 days later via an organic search to buy, Google claims the credit. We reconcile Google’s reported numbers against your actual CRM or Shopify data to find the true incrementality of the ad spend.
The honest verdict: Google Ads belongs in almost every media mix because the sheer volume of search intent cannot be replicated elsewhere. However, it requires constant vigilance. It is a terrible place to 'set and forget' a budget. If you sell a product with clear search demand and decent margins, it works. If you are trying to create demand for a product nobody knows exists yet, your budget belongs on Meta or native platforms first.
FAQ
Questions about Google Ads
The common queries we get when auditing a Google account.
Do we have to use Performance Max?
No. While Google pushes it aggressively as the default campaign type, it is entirely optional. We often run search only or shopping only campaigns when strict control over placement and search queries is required to hit target margins.
Why is my brand campaign suddenly costing more?
Competitors bidding on your trademark, combined with broad match keyword expansion, often inflate brand CPCs. We isolate brand terms into strict exact match campaigns to defend your baseline traffic at the lowest possible floor price.
Can you see exactly where Display network budget is going?
Yes, but you have to know where to look. We run regular placement reports to audit the URLs and apps where your ads appear, ruthlessly excluding mobile game apps and made for advertising domain networks.
How long does a new automated bidding strategy take to learn?
Typically fourteen to twenty one days, depending on conversion volume. During this period, cost per acquisition will fluctuate wildly. Touching the budget or targets during this phase resets the learning period.
Next step
Find out what your account is wasting
Send us access and we will come back with a written audit: where the budget is leaking, what it is costing you, and the three fixes worth doing first. No charge, no obligation.