Your conversion rate is dropping and Meta might have something to do with it
A falling conversion rate doesn't always come from your site. Meta is spending more and more on warm audiences, inflating recent visitors and mechanically dragging down your blended rate. Before pausing or rebuilding, segment, cross your signals and compare against last year.
- Author
- Axel Corton
- Subject
- Paid media
- Date
- Read time
- 6 min

Key takeaways
- Before summer, cold to warm spend sat at 6 to 1 on one account. It dropped to 1 to 1 in October.
- The same period last year showed 10 to 1, so this isn't seasonality.
- The pattern shows up across several accounts and other media buyers confirm it: the issue comes from Meta.
- Look at a 7 to 30 day window and compare to the same period last year, not to the week before.
- One metric never reveals the root cause: segment by source, placement, landing page and funnel step.
Something's bugging me right now and I'd rather be upfront about it. For the past few weeks, a lot of Meta advertisers have been complaining about weird performance. At first I kind of ignored it, because I wasn't seeing it clearly on my own accounts. On some of them I was actually seeing good stuff, so I figured it was just noise.
Then, digging into one account that had a suspiciously good stretch, I stumbled onto something. And that's when it clicked that the problem and the nice surprise came from the exact same thing.
A low conversion rate in eCommerce doesn't always come from your site
Everyone's first instinct when the conversion rate drops is to eye the site suspiciously. You assume you broke something, that a change went sideways, that there's a bug lurking somewhere.
Sometimes that's true. When your tracking is broken, you make decisions on half the data and yeah, in that case the problem is on your end. Same deal when 18% of your sessions are bots, your conversion rate collapses mechanically and it's got nothing to do with your site.
But there's a third case nobody looks at. A conversion rate that drops because the nature of the traffic Meta sends you has changed. What's changed isn't the volume. It's what those visitors are actually worth.
And that's exactly what I'm seeing right now.
The ratio I watch to understand what Meta is doing
There's a metric I pull up often and that gets way too little attention. The share of budget going to warm audiences, people who already know your brand, versus cold audiences, the new ones.
I watch this ratio to understand how Meta is splitting my budget. At the global level, but also at the ad set and ad level. It tells you where the algo decides to go hunting for the easiest conversion.
On one account I manage, before summer, for every euro spent on a recent visitor, roughly six went to someone new. A 6 to 1 ratio. Makes sense for a campaign meant to grow a brand, you mostly want to reach people who don't know you yet.
Since late August, that gap has been closing. And in October it slammed shut all at once. Over the last few days I'm more like 1 to 1. Just as much spent on people who already know me as on new ones.
My first reaction was to tell myself it was an anomaly, or something I'd triggered. I thought about seasonality, except last year over the same period I was at 10 to 1. So no.
Then I found it on another account I don't even manage. And yet another, with a completely different structure. Other media buyers are confirming it in private groups too. The problem comes from Meta, not from me.
Why this mechanically drives down your conversion rate
The link is direct. A recent visitor who already knows your brand converts better than a stranger. That's obvious. So if Meta spends more and more on warm traffic, your conversions attributed to ads look good, even great.
But your overall site conversion rate often says the opposite. Because if at the same time the volume of cold traffic you're bringing in drops or shifts in quality, and Meta is counting loads of recent visitors in its warm audience pool, you end up with more sessions from people who've already been there and a rate that swings all over the place depending on which segment you look at.
That's the trap. Your ad performance looks great. Your blended conversion rate, meanwhile, moves in sketchy ways. If you only watch a single metric, you miss it completely.
I've already told the story of a case with 65 add-to-carts and zero conversions. Raw volume tells you nothing. Until you turn the numbers into a rate and segment them, you're flying blind.
How to run the diagnosis without blowing everything up
Before you lunge for the pause button, here's the process I follow. In order.
First, separate a real drop from normal variation. A conversion rate bouncing around day to day is noise. Look at a 7 to 30 day window and compare it to the same period last year, not to the week before. Seasonality lies to you constantly if you compare badly.
Next, segment. A global conversion rate is useless on its own. Break it down by traffic source, by device, by landing page, and by funnel step. Add-to-cart rate against sessions, then the move from add-to-cart to payment. You want to know exactly where it breaks.
From there you can land in three situations.
When the add-to-cart rate is weak, the problem is up top, on the audience, creative or offer side. If add-to-cart holds but everything collapses afterward, go look at the funnel, the checkout, the shipping costs, a payment bug. And the last case, the rate per source stays stable but the global drops, that means the mix has changed, more traffic from a source that converts less. That's where Meta's warm/cold ratio comes into play.
To check this on your own, it's not complicated. In Meta's reports, pull a trend chart over the year and break it out by audience segment. You define your warm audience however you want, I use site visitors from the last 30 days. It's not a universal definition, everyone does their own thing, but watch how the curve moves. If the share of spend on warm traffic climbs while you've changed nothing, you've got your answer.
The causes everyone always forgets
The instinct is always the site. When nine times out of ten the cause is elsewhere.
Broken tracking, I covered that. Bots. Meta's shift in how it splits budget. But there's also seasonality, competitors' promos pulling prices down that exact month, and an algo change nobody warns you about.
My favorite theory on what's inflating recent visitors right now is the feature that shows the ad's landing page directly in the comments. It brings people back to the site with no real purchase intent. So it inflates your visitor count, deflates your conversion rate, and hands Meta a bigger pool to file under warm audience.
I do say theory. I don't have rock-solid proof. But the pattern fits.
And that's why you have to cross-check your signals. Conversion rate, abandoned carts, load time, customer feedback to support. A single number never tells you the root cause. It's by cross-referencing that you trace back to the real problem. A CRM that's humming along can actually hide an acquisition engine that's dying, same logic, a green indicator doesn't mean everything is fine.
What I'm doing now
Nothing. Well, not yet.
On the account where I spotted this, performance is good despite everything. The incremental attributed conversions look solid. So I'm not touching anything. The more patient I am and the fewer changes I make, the more I understand how the system actually behaves.
Maybe it's being skewed by warm traffic and it runs out in two weeks. Or maybe it's just a reporting bug, nothing to see here. I don't know yet.
The essential thing remains, and it's concrete. Before you cut a campaign or rebuild your site because the conversion rate is dropping, look at where your data comes from and what the algo is doing with your budget. Half the panic decisions I see come from a number read on its own, without context.
Segment, cross-check your signals, and compare to the same period last year rather than the week before. And the site, you touch that dead last.
Frequently asked questions
How do I know if my conversion rate drop is real or just normal variation?
Look at a 7 to 30 day window, not the day-to-day bounce. And compare to the same period last year, not to the week before. Seasonality lies to you if you compare badly.
Is the problem coming from traffic or from the site?
Segment your rate by source and by funnel step. If the rate per source is stable but the global drops, it's your traffic mix that changed, not your site. If add-to-cart is fine but it collapses at payment, the issue is in the checkout.
What is this warm vs cold audience ratio on Meta?
It's the share of budget Meta spends on people who already know your brand versus new ones. If that share shifts toward warm without you changing anything, your ad performance looks good but your global conversion rate can swing all over the place.
Should I cut my campaigns if I see this?
Not right away. If performance stays good, don't touch anything and observe. The fewer changes you make, the more you understand what the algo is actually doing. Cutting in a panic over a number read without context is the most common mistake.
Should I compare my rate to an industry benchmark?
Marginally. A good rate depends on your sector, your average order value, your device mix. Your real benchmark is your own segmented history. A market average will never tell you why YOU are dropping this month.
Keep reading
Subscribe to our newsletter
Our methods on paid media, eCommerce and CRM, in your inbox.



