If you're running Performance Max or Advantage+ and acquiring users who don't convert, the instinct is to call it a funnel problem and go fix the landing pages or the nurture flow. That's usually the wrong fix for the wrong problem, and it's an expensive one. More often it's an acquisition problem, and it traces back to a decision you may not realise you've handed over.

### Why Meta and Google push you toward automated buying

These campaign types can do things a manual setup can't. Take search on Google. With traditional responsive search ads you're capped at three enabled ads per ad group, each pointing to one landing page you picked, so a handful of fixed destinations and the headlines you wrote. Turn on Performance Max with Final URL expansion and Google can instead send someone to whichever page on your site matches their intent, and generate headlines and descriptions to fit that page. Someone searching for a specific product lands on that product's page, not a generic category one.

That's the real appeal: ads get more relevant to each user across more of your site, with far less manual work, and relevance is what drives efficiency. Advantage+ does the equivalent on Meta, assembling and targeting combinations no manual structure could cover.

The trade is control. You no longer pick the audience, the copy is generated, the platform chooses the landing page, and it decides who sees the ad. The most valuable decision, who to go and find, moves from you to the algorithm. That's a good deal when the algorithm is aiming at the right people. Whether it is comes down to one thing: the signal you give it.

![Article image 1](/images/blog/why-meta-advantage-and-google-pmax-keep-sending-you-low-value-users/image-001.png)

### The platform optimises toward whatever signal you send

Meta and Google don't know what a good customer looks like for your business. They know what you told them to optimise for, and they go and find more of it.

> WYSIWYG: what you send is what you get.

Optimise for "leads" and treat every lead as equal, and the platform finds you the cheapest leads it can, because that's what you asked for. It's doing its job perfectly; the instruction was the problem. Pass a value that says this lead is worth 150 and this one is worth 5, and it has something better to aim at, so it starts finding people who look like your 150s.

So the users who don't convert aren't leaking out of the funnel. They're exactly who you asked the platform to acquire.

### What a "funnel problem" actually costs you

The misdiagnosis is expensive in two directions at once.

First, the wasted spend keeps running. If your signal pulls in low-value users, every day the campaign runs it buys more of them. You're not paying once for a bad batch, you're paying continuously for the platform to get better at finding the wrong people.

Second, you pour effort into fixing the wrong thing: rebuilding landing pages, reworking nurture, adding sales follow-up, all aimed at converting people who were never going to convert, while the acquisition instruction underneath stays broken.

Put rough numbers on it to see the shape (illustrative, not a benchmark): you're spending 20k a month optimising for a flat "lead" event, and a large share of those leads have no real intent. A meaningful chunk of that budget is buying pipeline your sales team then spends time disqualifying. Worse, the CAC on your **good** customers looks fine on the dashboard because it's averaged across a pile of cheap leads, so the problem stays hidden. Fixing the funnel does nothing about any of that. Fixing the signal does.

![Article image 2](/images/blog/why-meta-advantage-and-google-pmax-keep-sending-you-low-value-users/image-002.png)

### Decide what a good user is before you optimise

The fix isn't a setting. It's a decision you make before touching the campaign: what makes a user valuable, and what can you observe early enough to act on? That second part is where most teams are too optimistic. Almost no one can see a customer's real value at the moment they convert.

| Your business | What you can see early | The catch |
| --- | --- | --- |
| Ecommerce | An actual purchase and its value | Closest to real value, but a first order isn't lifetime value; most mature teams pass a predicted LTV, not just order value |
| Subscription / SaaS | A signup or activation | An activation isn't value on its own; you need a proxy that correlates with who stays and expands |
| Long sales cycle | Little or nothing at conversion time | You're predicting from early proxies: lead quality, enrichment, first-session behaviour |

The pattern across all three: you're almost always sending a **prediction** of value, not confirmed value. Ecommerce comes closest to the real thing, and even they increasingly model it. So the practical question isn't "what is this user worth", it's "what can I see early that reliably points to what they'll be worth", and you send that.

### Make the value explicit, not just the conversion

Once you have that early proxy, pass it as value, not just a conversion. A conversion tells the platform "this happened." A value tells it "this was worth this much," which is what lets it separate your best users from your cheapest.

| What you send | What the platform learns |
| --- | --- |
| A flat "lead" or "purchase" event, the same for everyone | Every conversion is equal, go find the cheapest |
| "Lead, value 5" for a free-email signup, "Lead, value 150" for an enriched, qualified lead | These are worth ~30x more, go find people like them |

You don't need full lifetime-value modelling to start. You need a defensible view of relative value, and you need to segment it, because the same event can be worth different amounts depending on the campaign and platform it came from. Flatten that and you've handed Meta or Google a blunter instruction than they can follow, and paid full price for it.

### Watch what your signal does to your traffic

Whatever you optimise for, the platform pulls your traffic toward it, and that can narrow you in ways you didn't intend.

One team we worked with optimised for people who called them, because calls closed well. It worked, traffic shifted toward likely callers. But it quietly missed everyone who would have booked directly without ever calling, a whole segment of good revenue the campaign stopped chasing. The signal was doing its job so well it steered spend into one narrow slice of the real market.

So this isn't set-and-forget. Send the signal, then watch what it does to the mix of people coming in, and check you haven't optimised yourself into a corner.

![Article image 3](/images/blog/why-meta-advantage-and-google-pmax-keep-sending-you-low-value-users/image-003.png)

### What this looks like in your results

Done well, it shows up in your numbers as one of two outcomes, both testable:

- **Better users at the same spend:** higher-quality pipeline and more revenue per acquired customer, without paying more per acquisition.

- **The same quality at lower spend:** you tell the platform who **not** to chase, hold acquisition quality steady, and cut the wasted budget.

Run it as a pilot, compare against your own data, and you'll see which one you got, and roughly what it was worth.

And keep the limits in view. None of this is exact; you're acting on early, incomplete information, so you'll live with gaps. The aim was never a perfect read on every user. It's a signal good enough to point Meta and Google at the right people, improved as you learn.

If you want to work out what a good user looks like in your data, whether you can see it early enough to act on, and what better targeting would be worth to your numbers, book a call and we'll take a look together.