Your Google Ads dashboard is lying to you. On paper, everything looks elite: impressions are skyrocketing, and your click-through rate has never been healthier. But in your bank account? The silence is deafening.
While Google’s AI promises efficiency, the reality is often a “black box” that prioritises engagement over intent. When I audit accounts today, I don’t just see high CPCs; I see “AI Max” features effectively dismantling carefully built keyword strategies and bidding on competitor terms without consent.
This article breaks down why the misalignment between Google’s goals and your profit margins is widening and the specific framework I use to stop the wasted ad spend and restore human oversight to automated bidding.
The Misalignment at the Heart of AI Optimisation
The core problem with handing the keys entirely to AI is a fundamental misalignment of goals. Google’s algorithms do not inherently optimise for your revenue or your profit margins. They optimise for engagement — and more specifically, for the metrics that serve Google’s own business model. If people react to your ad, even if they have zero intent to buy, the algorithm interprets that as a positive signal and keeps pushing the ad to similar audiences.
The result is high clicks, low return on investment, and significantly wasted ad spend. Studies suggest that up to 40% of ad budgets can be lost to this kind of inefficiency. That is not a rounding error; that is nearly half your marketing investment evaporating before it ever reaches a genuine potential customer.
This sentiment is now well-documented across the industry. A comprehensive survey of over one thousand PPC practitioners, published in the 2026 State of PPC report, found that 53% of respondents believe managing PPC campaigns is harder today than it was two years ago. The primary reasons cited were striking: 62% pointed to less insight from ad platforms, 62% noted less accuracy, and 43% highlighted a general loss of control . As Google pushes more automated solutions, advertisers are finding it increasingly difficult to steer the ship — and the industry is starting to push back.
The Hidden Risks of AI Max and Broad Match Expansion
One of the most significant drivers of wasted spend right now is Google’s push towards broad match and its flagship automation feature, AI Max. Launched globally in late 2025, AI Max combines search term matching with asset optimisation, packaging existing technologies (Dynamic Search Ad logic, automatically created assets, and final URL expansion) into a single toggle that advertisers can switch on or off.
What Google does not make immediately obvious is what happens when you flip that switch.
Broad-Matchifying Your Carefully Built Keyword Strategy
Research by Mike Ryan, Head of Ecommerce Insights at Smarter Ecommerce, analysed one million AI Max impressions and found that the feature effectively “broad-matchifies” exact match and phrase match keywords. His data showed that exact match expansions accounted for between 27% and 89% of AI Max match type impressions, with an average of around 80%.
In plain English: the tight, intentional keyword structure you have spent months building, the one designed to attract only your most qualified prospects, gets loosened the moment AI Max is active. Your exact match keywords stop behaving like exact match keywords.
Bidding on Competitor Terms You Never Wanted
The implications go further than just broader matching. Data shared by Lunio, an ad traffic verification platform, revealed that in one AI Max-enabled campaign, competitor brand terms accounted for 69% of total impressions — compared to just 77 impressions generated by standard broad match for the same terms. Mike Ryan estimates that including all competitor terms would push that impression share above 80%.
If you are not running a deliberate competitor conquesting strategy, this is a serious problem. You end up paying inflated CPCs to appear in searches where the user is explicitly looking for someone else. Beyond the wasted spend, there is a brand misalignment risk: appearing alongside competitor searches can confuse users and damage your positioning in the market.
Some industry commentators have gone further, suggesting that AI Max may be designed to trigger these “turf wars” deliberately, since competitor terms are traditionally less competitive and represent an opportunity for Google to increase ad revenue by drawing more advertisers into bidding on them.
The Loss of Intent Matching: A Real-World Example
During a recent audit for a client, I encountered precisely the issues that agencies testing AI Max have documented. The account was experiencing a classic “breakdown effect”: impressions had surged by an incredible 81.4% month-over-month (from 54,227 to 98,380), but the conversion rate had plummeted by 54.8% (dropping from 11.14% down to just 5.04%).
Where was all this new traffic coming from? I dug into the Search Terms report and found over $2,585 in wasted ad spend generated by zero-conversion terms. The AI, specifically through Performance Max and broad match keyword expansion, was bidding heavily on irrelevant queries, including competitor terms and generic searches that had no conceivable relationship to client’s service.
Furthermore, 77% of the Performance Max impressions were being allocated to low-conversion channels like Maps and YouTube. These channels consumed over $1,100 of the budget while delivering almost zero conversions.
This is the pattern I keep seeing across accounts: impressions skyrocket, but conversions flatline. The AI is finding traffic, but it is the wrong traffic.
Runaway CPCs Without Guardrails
There is a third dimension to this problem that often goes unnoticed until it is too late: when Smart Bidding is left unrestricted, it can overspend dramatically on individual clicks. The PPC Edge newsletter documented cases of CPCs reaching 20 to 35 times the campaign average, with one practitioner, Dan Chorlton, reporting a single click costing $332.39 — 15x his average CPC.
The algorithm justifies this by reasoning that a conversion is likely. But when the CPC exceeds the cost-per-acquisition target, the logic collapses entirely. Without bid limits, a handful of errant clicks can decimate a daily budget before you have had a chance to intervene.
How I Tightened Up My Clients’ Accounts
When I saw my clients’ impressions booming with zero return, I knew I had to take back control from the AI. The following is the framework I applied across multiple accounts to eliminate the wasted spend and restore meaningful performance.
1. Portfolio Bid Strategies with Maximum CPC Limits
The most immediate fix for runaway CPCs is setting a hard ceiling on what the algorithm can bid per click. Standard campaign-level bid strategies do not offer this option, but Portfolio Bid Strategies do.
By grouping campaigns into a portfolio strategy (accessible via Tools & Settings > Bid Strategies), you unlock an “Advanced Options” section where you can set a maximum CPC bid limit . The recommended approach is to set this limit at 3 to 5 times your campaign-level average CPC — not your account-level average, which will almost always be too low and will starve the algorithm of the room it needs to perform .
This single change brought immediate stability to several accounts. CPCs stopped spiking, budgets became predictable, and the algorithm still had enough headroom to bid competitively for genuinely high-intent searches.
Important caveat: Portfolio bid strategy maximum CPC limits are not currently available for Performance Max campaigns, which is a significant limitation given how widely PMax has been pushed.
2. Aggressive Negative Keyword Management
Because AI is constantly trying to broaden your reach, you must actively and continuously define what you do not want to target. This is no longer a set-and-forget exercise; it is an ongoing discipline.
I spent considerable time in the search terms reports, specifically filtering for “search terms and landing pages from AI Max” to isolate the traffic the automation was generating independently. The results were instructive. Hundreds of irrelevant terms — vague single-word queries, competitor brand names, and geographically misaligned searches — were consuming budget without any realistic chance of converting.
Adding these as negative keywords (at the exact match level where appropriate) immediately reduced wasted impressions and improved the signal quality being fed back to the algorithm. Lunio’s guidance reinforces this approach: maintaining tight negative keyword lists and audience exclusions is one of the most effective defences against AI Max’s tendency to overreach .
3. Tightening Match Types — With Eyes Open
Reverting to exact and phrase match keywords was a necessary step to regain control of which searches triggered my clients’ ads. However, given the research showing that AI Max effectively broad-matchifies even exact match keywords, this step cannot be taken in isolation. It must be paired with the negative keyword strategy above and, where AI Max is active, with careful monitoring of the search terms report to catch any drift.
The broader lesson here is that match type alone is no longer a reliable control mechanism. The architecture of keyword targeting has fundamentally changed, and practitioners need to adapt their account management accordingly.
4. Disabling Auto-Applied Recommendations
Google’s auto-applied recommendations frequently push for broader reach, additional channels, and expanded targeting — changes that serve Google’s revenue model more reliably than they serve the advertiser’s. Across the accounts I audited, auto-applied recommendations had quietly introduced changes that loosened targeting and increased spend without any corresponding improvement in conversion quality.
Turning these off — or at minimum, reviewing every recommendation before it is applied — ensured that any expansion in targeting was a deliberate, strategic human decision rather than an algorithmic assumption.
5. Tracking Real Conversions, Not Vanity Metrics
Finally, and perhaps most fundamentally: I audited the conversion tracking setup in every account. In several cases, the “conversions” being reported included soft signals — page visits, video views, scroll depth — that the algorithm was treating as evidence of success and using to justify continued spend. Cleaning up the conversion tracking to reflect only genuine business outcomes (form submissions, phone calls, purchases) gave the algorithm accurate feedback and dramatically improved the quality of its subsequent decisions.
AI Needs Human Oversight
AI is an extraordinarily powerful tool for paid search, but it is tactical, not strategic. It does not understand your brand positioning, your competitive dynamics, or the nuances of your customer acquisition cost and lifetime value. It is a machine programmed to seek opportunities and it will find them, whether or not they are opportunities you actually want.
The PPC practitioners who are thriving in this environment are not those who have abandoned AI, nor those who have surrendered to it entirely. They are the ones who have learned to set the rules of engagement: defining clear guardrails, maintaining rigorous exclusion lists, and using automation to execute human-led strategy rather than replace it.
As one industry expert put it succinctly: “Rules need to be set, or else there’s a Waluigi running around your account”. Left unchecked, AI Max will find traffic — but it will be the wrong traffic, at the wrong price, for the wrong reasons.
The good news is that the fixes are not complicated. They require time, attention, and a willingness to push back against the platform’s default settings. But for every client account I have tightened up using the strategies above, the results have been the same: fewer wasted impressions, lower CPCs, and — most importantly — conversions that actually mean something.
Every week you let AI run unchecked, you are funding your competitors. The question is whether you are going to take back control.
Summary: Key Actions to Reduce AI-Driven Wasted Spend
| Action | Why It Matters |
| Set Portfolio Bid Strategy max CPC limits (3–5× average CPC) | Prevents the algorithm from bidding absurd amounts on individual clicks |
| Build and maintain negative keyword lists aggressively | Stops AI from expanding into irrelevant, competitor, and low-intent traffic |
| Audit match types and monitor AI Max search terms separately | Exact and phrase match no longer behave as expected when AI Max is active |
| Disable auto-applied recommendations | Prevents Google from quietly loosening your targeting without consent |
| Track only genuine business conversions | Ensures the algorithm is optimising for real outcomes, not vanity metrics |
| Run AI Max as a campaign experiment, not a full account rollout | Limits exposure while gathering meaningful performance data |
