Hi there,
Thanks for reaching out! Read through your situation and honestly, it’s a classic PMax headache. It’s the kind of thing that makes you want to tear your hair out. You have a campaign that’s been your reliable workhorse for two years, performing great, and then overnight it just falls off a cliff for no apparent reason. It’s baffling and incredibly frustrating.
The good news is, you're not going mad. I've seen this exact pattern happen before, especially with heavily automated campaign types like Performance Max. The short answer is that the algorithm has likely found a new, cheaper source of traffic that it *thinks* is good, but in reality, it's junk. Your job now is to figure out where that junk traffic is coming from and steer the machine back onto the right path.
I’m happy to give you some initial thoughts and a bit of a deep-dive into how I'd approach diagnosing and fixing this. It's a solvable problem, but it requires playing detective and taking back a bit of control from the black box.
TLDR;
- Your CPC and Conversion Rate dropping by 80% at the same time is a huge red flag for a sudden drop in traffic quality. The algorithm has likely found a new, cheap, but non-converting audience segment and is now spending your budget there.
- The first step is forensic analysis. You need to dig deep into your reports – specifically Search Categories (the new Search Terms), Location data, and any available placement insights – to pinpoint where this low-quality traffic is originating.
- The fix involves regaining control. This means aggressively adding negative keywords (at an account level), refining your audience signals to be much more specific, and potentially restructuring your campaign to isolate variables and give the algorithm clearer direction.
- You need to stop thinking of PMax as a "set and forget" tool. It needs constant monitoring and strategic adjustments. This article includes a Diagnostic Flowchart to help you troubleshoot future issues and a ROAS Calculator to define your profitability targets more clearly.
- The most important piece of advice is to challenge the algorithm. Don't assume it knows best. Your real-world business data and intuition are your most valuable assets in managing these automated campaigns.
So, what on earth is Google doing?
Before we get into the fixes, it helps to understand the 'why'. You asked what Google is doing, and it's a fair question. When you set a Target ROAS (TROAS), you're giving the algorithm a goal. Its job is to find conversions that, on average, meet that goal. However, its primary directive is also to spend the budget you give it while seeking out the lowest possible cost for clicks and actions.
What likely happened around November 1st is an algorithmic shift. Google is constantly updating its systems. This could have been a broad update, or it could be something specific to your account's learning phase. The algorithm probably identified a new pocket of users or placements that are incredibly cheap to reach. In its world of data points, it saw "cheap clicks" and "some conversion signals" (even if they are weak ones) and thought, "Brilliant! I can hit the TROAS target more efficiently here."
The problem is, the algorithm's definition of a "good prospect" and your definition are two different things. It sees a user who has previously shown some vague interest in a related product. You need a user who is actively in the market, with their credit card in hand, ready to buy *your* product. The 80% drop in CPC is the smoking gun – you're now paying for bottom-of-the-barrel traffic. Your observation that the click:interaction rate is the same is sharp, but it doesn't mean the quality within that traffic is the same. It just confirms the clicks are coming from similar formats (i.e., Shopping listings), not that the *people* clicking are of the same quality.
This is the fundamental trade-off with PMax: you give up control for automation. When it works, it feels like magic. When it breaks, it's a complete black box. Our job now is to shine a light inside that box.
We'll need to look at your data forensics...
To fix this, we need to become detectives. We have to follow the money and figure out exactly where your budget is now being spent. We're looking for anomalies and sudden changes from October to November.
1. Scrutinise Your Search Categories (The New "Search Terms")
You mentioned your keywords are still relevant, which is good. But PMax now uses "Search Categories," which are a bit more vague. I'd challenge you to look deeper. Are the *themes* of the search categories the same? For example, have you shifted from high-intent "buying" terms to lower-intent "research" terms? Both might seem relevant, but their conversion potential is worlds apart.
Think about the difference between these two types of search terms for, say, a company selling high-end running shoes:
| High-Intent (Buying Now) | Lower-Intent (Researching) |
|---|---|
| "brooks adrenaline gts 22 size 9 uk" | "best running shoes for flat feet" |
| "buy hoka clifton 8 online" | "hoka vs brooks comparison" |
| "on running cloudmonster sale" | "what is pronation in running" |
The algorithm could have shifted your spend from the left column to the right. The terms on the right are still 'relevant', but they attract people who are weeks or months away from a purchase, if they ever make one. This would absolutly expain the drop in conversion rate. You need to go through your search category insights and see if there’s been a thematic shift like this. If there is, it’s a strong signal we need to add negative keywords to block this research-based traffic.
2. Interrogate Your Location Data
This is often the biggest culprit. A sudden drop in CPC can happen when the algorithm starts showing your ads much more heavily in a lower-income country or region. The people there might click, but they may not have the disposable income to purchase, or your shipping costs might be prohibitive. It’s a classic way to get cheap clicks that go nowhere.
Go into your location reports and compare October to the first week of November. Create a report that shows Clicks, Cost, and Conversions by country (or by region/city if you only target one country). Are you seeing a massive spike in clicks from a specific location that has zero or very few conversions? If you find a country or region that is soaking up budget with no return, exclude it immediately. Don't hesitate.
Here’s a conceptual look at what you might find. Imagine this is your click distribution:
3. Check Your Audience Signals
Audience signals are your main way of guiding PMax. Have these been too broad? Are you using generic interests? If your signals are weak, you give the algorithm more leeway to go off-piste and find these weird pockets of cheap users. You might need to tighten these up considerably, using custom segments built from high-intent keywords or competitor URLs, and most importantly, uploading your customer list. A list of past purchasers is the most powerful signal you can give Google.
To help you systematically walk through this, I've created a little diagnostic flowchart. It's a simplified version of the process I'd go through myself.
Create & apply comprehensive negative keyword lists at the account level.
Immediately exclude underperforming countries, cities, or regions.
Refine signals: Upload customer lists, create custom segments based on high-intent searches.
I'd say you need to regain control...
Detective work is one thing, but now you need to take action. Your goal is to force the algorithm out of its bad habits and back towards the profitable traffic you were getting before.
1. Go Overboard with Negative Keywords
This is your most powerful weapon. PMax makes it difficult, but you can apply negative keyword lists at the Account Level. Go through your search category insights and be ruthless. Any term that smells of research, comparison, or is just plain irrelevant needs to go on a list. Think of terms like "reviews," "free," "vs," "alternative," "how to," etc. Build a "master" negative list and apply it. This is absolutly critical.
2. Restructure Your Asset Groups
If you have one big PMax campaign with one asset group for everything, you're giving the algorithm too much freedom. It's lumping all your products and signals together. Consider restructuring. A common and effective approach is to create separate asset groups (or even separate campaigns) for different product categories.
For example:
- Asset Group 1: Best-Sellers. Use a product feed selection for only your top 5-10 products. Give it really tight audience signals based on past purchasers of those specific items.
- Asset Group 2: Category A (e.g., Men's Shoes). Signals focused on interests and custom segments related to men's footwear.
- Asset Group 3: Category B (e.g., Women's Apparel). Signals focused on interests and custom segments related to women's clothing.
This segmentation forces Google to be more specific. It stops it from using signals for women's apparel to find customers for men's shoes, which can lead to the kind of wasteful spending you're seeing. I've seen this segmentation strategy work very well for eCommerce clients. For example, one campaign we worked on for a women's apparel brand saw a significant increase in returns by implementing a similar structure, as it allowed us to give the ad platform's algorithm much clearer signals about who to target for each product category.
3. Reboot with Better Audience Signals
Don't just tweak your existing signals; reboot them. Pause the old ones and create new, higher-quality ones.
- Customer List: If you aren't using this, start now. Export a list of your past customers' emails and phone numbers and upload it. This is the highest quality signal you can provide.
- Custom Segments (Intent): Instead of using Google's pre-built interest categories, create custom segments based on "people who searched for any of these terms on Google." Fill this with your absolute best, highest-intent keywords (the ones from the left column in the table above).
- Custom Segments (Website Visitors): Create segments for people who visited specific competitor websites. This allows you to target users who are actively shopping around in your niche.
By providing these much stronger signals, you are effectivly telling the algorithm, "Stop looking for cheap weirdos in the dark corners of the internet. These are the people I want you to find. Go get more of them."
You probably should reset your targets...
When a campaign goes this far off the rails, simply waiting for the existing TROAS to fix it is a mistake. The campaign is now learning from bad data, reinforcing a negative feedback loop. You need to reset its expectations. This means first understanding what a realistic and profitable ROAS target actually is for your business.
Many businesses set a TROAS target based on a guess or what "sounds good." You need to calculate it based on your actual profit margins. A 300% ROAS sounds great, but if your margins are thin, you could still be losing money. Use this calculator to find your break-even point and set a healthy target.
Once you have your true, calculated Target ROAS, I would consider lowering your current TROAS target in the campaign settings temporarily. Yes, this sounds counterintuitive. But your campaign is failing to hit its current target so badly that it's just flailing. By lowering the target to something more achievable (but still profitable), you might give the algorithm a chance to stabilise, find *some* conversions, and relearn from better data. Once performance has stabilised for a week or two, you can start gradually increasing the target back up to your desired level.
You'll need a long-term fix...
The immediate problem is the fire in front of you. But this whole episode should be a wake-up call. Relying 100% on a single, opaque PMax campaign is a massive risk, as you've just discovered. A long-term strategy needs to be about building resilience and diversification.
You shouldn't necessarily abandon PMax—when it's working, it's very effective. But you should build a safety net around it. Consider carving out some budget to run standard Shopping campaigns for your most important products. This gives you direct control over bids and keywords for your cash cows, ensuring they always have visibility, regardless of what PMax is doing on any given day. Standard Shopping is more manual work, but that work translates into control and predictability.
Furthermore, this is a Google-only problem. What are you doing on other platforms? The visitors who come from this cheap PMax traffic might not be ready to buy *now*, but they've shown some level of interest. A well-structured retargeting campaign on Meta (Facebook & Instagram) could be a very cost-effective way to bring them back later when they are ready to purchase. Building a multi-channel presence means you're not held hostage by the whims of a single algorithm. I've worked on campaigns for everything from subscription boxes to high-end apparel where building a multi-channel presence creates a system that is far more profitable and stable. For example, one eCommerce campaign we ran on Meta for a client selling maps generated an 8x return, showing how powerful the platform can be for reaching the right audience.
This is the main advice I have for you:
| Phase | Action | Why This is Important |
|---|---|---|
| 1. Diagnosis (Immediate) | Deeply analyse Search Categories, Location data, and Audience performance for November vs. October. | You can't fix a problem you don't understand. This step identifies the source of the low-quality traffic. |
| 2. Stabilisation (Next 24-48 Hours) | Aggressively exclude underperforming locations and apply a comprehensive negative keyword list at the account level. | This stops the bleeding. It immediately cuts off the most obvious sources of wasted ad spend. |
| 3. Relaunch (This Week) | Restructure into theme-based asset groups. Replace broad audience signals with high-quality ones (customer lists, custom intent). | This gives the algorithm clearer, better instructions, forcing it to target higher-quality users and relearn good habits. |
| 4. Long-Term Strategy (Next Month) | Launch parallel Standard Shopping campaigns for top products and consider Meta retargeting to build a more resilient system. | Reduces reliance on a single, volatile campaign type and creates a more robust, multi-channel advertising strategy. |
I know this is a lot to take in. Managing a PMax campaign that’s gone rogue is not a simple fix. It takes time, expertise, and a willingness to constantly test and analyse. It's not the "set it and forget it" solution Google often sells it as. The algorithm is a powerful tool, but it's still just a tool. It needs a skilled operator to guide it, correct it when it makes mistakes, and provide it with the right data to make better decisions.
If you'd like to go through your account on a call and have me take a look at the data with you, feel free to book in a free, no-obligation strategy session with us. Sometimes a second pair of expert eyes on the actual data is the fastest way to get to the bottom of it.
Hope this helps!
Regards,
Team @ Lukas Holschuh
