Why Your Meta CPM Spiked, and Who Caused It

Your CPM jumped 30% on a Tuesday. You did not change the targeting, you did not swap the creative, you did not touch a budget. So you do what everyone does: you blame the creative, you test three new hooks, and a week later you have spent two thousand euros proving that the creative was never the problem.

Every article about this will now explain the second-price auction to you and then tell you to refresh your creative. Here is the part they leave out. Nobody can show you a competitor's CPM. Not Meta, not us, not any tool that claims otherwise. CPM is a number that exists inside one ad account, and you only have keys to your own.

What you can see is the thing that moves it: the money going into the auction. EU and UK disclosure rules force advertisers to publish what they spend and who they reached, ad by ad. So the question "why did my CPM spike" has an answerable version, which is "who started spending, how much, and on what day". The eight advertisers below are the answer for one slice of the Meta index on 13 September 2026. Between them they carry €437,700 of disclosed spend and 45.1 million people reached, and just under €11,000 a day is still going in.

The 8 advertisers buying into the auction

Ranked by each advertiser's single highest-spending live ad. Spend is the disclosed EU and UK total for that one ad, Per day is its average daily rate, and Days is how long it has been running. Every ad below was still live on 13 September 2026.

How to read the numbers

Spend and reach are EU and UK only. They are the slice European and British transparency rules force advertisers to publish, nothing more. A brand can be spending five times these figures worldwide and the disclosed column would not move. So read every euro figure here as a floor, never as a budget.

The rank badge is worldwide, even though the money is not. When the app says an ad is in the top 1% of a brand's 425 active ads, that is computed across everything the brand runs in every market. It is not a European ranking. This is why an ad can be a brand's single best performer on earth and still sit next to a modest euro figure: the rank sees the whole account, the money column sees one region.

Every ad here is still live. The Meta index will happily report an ad as active whose last recorded day was four months ago, so status alone is worthless. These eight were filtered on their end date falling inside ten days of the index date, and on having run at least fourteen days, because a campaign that launched yesterday tells you nothing about whether it is scaling.

None of these numbers is a CPM, and none of them is yours. They are inputs to the auction you share, not outputs of your account. The point is not to benchmark your CPM against theirs. It is to find out whether the thing that changed on Tuesday was you or them.

1. Harvest Right

See this ad in Brandsearch

Harvest Right sells home freeze dryers, and its biggest disclosed ad has been running for 79 days on €132k, the largest single figure in this list. The creative is one flat line of copy, "Harvest Right freeze dryers are now available in Europe", under a Learn more button. No hook, no urgency, no social proof. It is an availability announcement.

The badge is what makes this row interesting. Top 4% of 29 active ads sounds modest until you notice the denominator: this brand runs twenty-nine ads in total. It is not flooding the auction with variations, it is pushing one message very hard into a market it has only just entered. That is a different threat model from a brand running four hundred creatives, which is why the active ad count in your niche is worth knowing before you read anyone's spend figure.

2. Spiral

See this ad in Brandsearch

Spiral sells weather-resistant backpacks, and this video has been live for 88 days, the longest run in the set, carrying €100k and 8.7 million people reached. The offer is doing the work: "Buy ANY Backpack. Get ANY Crossbody FREE! No bundles. No conditions. Two bags. One order."

Eighty-eight days is the number to pay attention to. An ad that survives three months of continuous delivery is not being kept alive by hope, it is being kept alive by a return that still clears at whatever the CPM has drifted to. When you are trying to work out whether a cost rise is temporary, the competitor who has been paying it for three months without flinching is telling you something the auction cannot.

3. Estrid

See this ad in Brandsearch

Estrid runs a razor and deodorant subscription with a vegan body care range, and this is the clearest picture of scaling mode in the list: top 1% of 425 active ads, 75 days, and six versions of the same creative running at once. The copy is a subscription bribe, "A free mystery gift is yours when you subscribe to Estrid. Could be a face razor. Could be a deo."

The versions count is the tell worth learning. One creative duplicated six times is not a brand testing six ideas, it is a brand that has already found the winner and is buying more delivery for it. That is exactly the behaviour that widens the auction under everyone else, and it shows up in the index days before it shows up in your cost per result.

Where Estrid's money actually lands

The Adspend view splits the same ad into the numbers that decide whether it touches you at all: €49.7k total, 4.3 million reached, €1.2k average daily, and a country breakdown reading GB 100%. If you sell into Germany, this particular ad is not your problem no matter how impressive the badge looks. If you sell razors in Britain, it is competing with you every hour of every day.

That country split is the step most people skip. A competitor doubling spend in a market you do not sell in changes nothing for you, and treating it as a threat is how budgets get moved for no reason. Check which markets a competitor's spend is actually landing in before you touch anything.

4. Les Petits Culottés

See this ad in Brandsearch

Les Petits Culottés sells French eco-friendly nappies and an organic infant milk, and this video is putting €1.8k a day into the auction, the highest daily rate in the list, across 33 days and two versions. The hook leads with a rating, "Notée 4,9/5 par les parents", then offers a free trial pack.

This row carries no rank badge, and that absence is information rather than a gap. A high daily rate with no worldwide ranking means the brand is spending hard on something that is not yet its best performer, which is what a scaling test looks like from the outside. Set it against Spiral two rows up: same auction, opposite phase, and the correct response to each is different.

5. Beavertown Brewery

See this ad in Brandsearch

Beavertown Brewery sells craft beer, and the entire creative is fifteen words: "Fancy a case of Neck Oil? Discover Neck Oil & more from Beavertown Brewery. Order now." Thirty days, €36.2k, no badge, no duplicated versions.

The panel for this one has only two figures on it, days and spend, because the brand has not stacked versions and the ad has not ranked. That is what a steady, unremarkable, permanently present advertiser looks like in the data, and most of the money in any auction belongs to advertisers exactly like this. The brands that spike your costs are newsworthy, but the ones setting the floor are not.

6. 47 Skin

See this ad in Brandsearch

47 Skin sells anti-blemish and scar-repair skincare, and its top disclosed ad is at €989 a day, the lowest daily rate here, while still carrying a top 2% badge across 266 active ads. The creative offers a trial: "Skin confidence. Now available for you to trial."

This is the row that most cleanly breaks the assumption that spend and performance are the same axis. €989 a day is a rounding error next to Harvest Right, and yet within its own account this ad beats 98% of everything the brand runs worldwide. The euro column measures disclosure. The badge measures performance. Reading the first as if it were the second is the single most common mistake made with this data.

What 47 Skin repeats, and how often

The Hooks view counts every opening line the brand runs: 35 opening lines across 371 of the ads we scanned, with a further 1,169 catalog ads where Meta fills in the text per product. One line, "47 Skin: The Breakout Solution That Actually Works", is carrying 242 ads on its own. The next one down has 31.

A distribution that lopsided is a brand that has stopped exploring. When a competitor's creative volume collapses onto one hook, they have found their answer and the rest of the spend is going behind it, which is the moment auction pressure in that niche stops being noisy and starts being sustained. Find the hooks already carrying your competitors' ad volume.

7. INITIO

See this ad in Brandsearch

INITIO makes perfume, and this is the newest ad in the list at 24 days, already at €28.2k and already top 2% of 111 active ads. The copy is unhurried and product-led: "OUD FOR GRACE, a new expression of Oud : deep yet lifted, powerful yet perfectly balanced."

Twenty-four days to a top 2% placement is fast, and this is the profile that should worry you most if you sell fragrance into Italy, where 82% of the disclosed spend on this ad landed. A new ad climbing quickly is the leading indicator; the 88-day veteran is the lagging one. If you only check who is spending the most, you will find the veterans and miss the climbers entirely.

8. air up

See this ad in Brandsearch

air up sells a water bottle that flavours water through scent, and its top disclosed ad runs three versions across 32 days on €14.6k, with 86% of that landing in France. The hook is a question that argues with the category: "Si c'est sain, c'est forcément pas bon ?", roughly, if it is healthy it must taste bad, right?

At €14.6k this is the smallest total in the list and it still represents €1.3k a day of live pressure, more per day than Estrid or Spiral. Totals reward ads that have been running longest. Daily rates tell you what is happening now. When you are diagnosing something that started on Tuesday, sort by the second one.

What the eight have in common

Three things, all computed from the same pull rather than asserted.

Duplication tracks confidence, not budget. Estrid runs six versions of one creative, air up three, INITIO and Les Petits Culottés two each. The three brands with the largest totals, Harvest Right and Spiral and Beavertown, run one version apiece. Stacking versions is what a brand does once it is sure, and it is visible in the index the day it starts.

Long-running ads and high daily rates are different populations. Spiral has been live 88 days at €1.2k a day. Les Petits Culottés has been live 33 days at €1.8k. The first is a proven winner being maintained; the second is a bet being pressed. Both add to your CPM, and only one of them is likely to still be there in a month.

Disclosed spend concentrates by country, hard. Six of the eight put 82% or more of their disclosed spend into a single country: GB for Estrid, Spiral, 47 Skin and Beavertown, FR for Les Petits Culottés and air up, IT for INITIO. Auction pressure is not a global weather system. It is local, and the first useful question about any competitor's spend is which market it landed in.

The check, in ten minutes

This is the diagnostic the whole article exists for. Run it when your CPM moves and your account did not.

Step 0. Decide whether this is you or everyone. Before you go looking for a culprit, find out whether there is one. A CPM that moves more than about 20% off your normal range is worth investigating and anything under that is noise, which means you need a CPM benchmark for your niche before a deviation can mean anything at all. If the advertisers in your niche are scaling through the same window, spending flat or up on ads that are still running, you are looking at a market shift and you can scale into it with some confidence. If they are testing, pulling back or killing winners, the problem is more likely to be inside your own account.

Step 1. Sort the niche by disclosed spend and read the daily column. Filter Meta ads to your category, sort by total spend, and then look at the per-day figure rather than the total. The total tells you who has been here longest. The daily rate tells you who is buying right now.

Step 2. Look for climbers, not leaders. Rank movement is the signal that a snapshot cannot give you. An ad that has jumped a hundred positions in the last week or two is getting budget pushed into it right now, and that is the one about to show up in your costs. Tracking a brand is what accumulates the daily rank history that makes this readable; the Ad Rank tab links to it directly.

Step 3. Check the versions count. One creative duplicated several times, as with Estrid's six, is scaling mode. A brand adding genuinely different creatives is still exploring. The first costs you money soon, the second might not cost you anything.

Step 4. Match the dates, then check the market. Find the day your CPM started climbing and compare it with the competitor's spend ramp. If they line up inside 48 to 72 hours you have your answer. Then check the country split before you act, because a surge in a market you do not sell in is not your surge.

Step 5. If nobody moved, go back to your own account. This is the branch most versions of this advice never close. If no competitor in your niche shows a spend ramp, a rank climb or stacked versions, then the cause is probably internal: audience overlap between your own ad sets, a relevance or quality drop, or a creative that has simply run past its useful life. Competition is a common explanation, not a universal one, and deciding it was competition when it was not costs you the week you spend watching the wrong dashboard. If the ad in question is one of your own long-running winners, the question is whether to kill it or refresh it, and that is a different decision from the one this article is about.

Check which advertisers in your niche are scaling this week.

What the rank and runtime chart shows

The Ad Rank tab plots rank against how long each ad ran, which is this whole argument drawn by the product. On Harvest Right the shape is unambiguous: the cluster at the bottom left is short-lived ads that never ranked, and everything reaching the top of the chart has been running past sixty days. Nothing arrives at rank one quickly and stays there cheaply.

This is also the fastest way to read a competitor's discipline. A brand with a dense low-left cluster kills fast and tests constantly. A brand with a thin chart and a few long survivors, like this one, is running a small number of bets for a long time. See how long the winning ads in your niche have been running.

Four ways to respond

Pick by your margin and your creative capacity, not by reflex.

Match the spend, but only on your best ad sets. If your unit economics can absorb a higher CPM, raise the daily budget to hold impression share. The arithmetic is simple enough to do in your head: a CPM that rises 40% needs roughly 40% more budget to buy the same delivery, so €2,000 a day becomes €2,800. Do it on your two or three strongest ad sets rather than across the account, and check first whether you are still profitable at the ROAS that implies for your category. A competitor spending heavily may be running a loss-leading launch you have no reason to match.

Study the creative instead of outbidding it. Better creative at the same budget beats more budget behind the same creative, and a competitor who has been running one ad for 88 days has already paid for the testing. You are not copying the ad, you are taking the angle: the structure of the hook, the objection it answers, the offer shape. That is their research, and it is sitting in the index for free.

When the auction is quiet, do the opposite. If your CPM rose and no competitor is spending through it, flooding is not the problem and matching spend will not help. Narrow to high-intent audiences rather than broad ones, run fewer creatives at higher budgets, because Meta needs more data per creative to find a winner when delivery is expensive, and weight roughly 70% of budget to proven performers and 30% to tests. Holding back 10 to 15% of the monthly budget as a contingency is what makes this possible at all; if every euro is committed, you have no move to make.

Change the timing, not the spend. Some brands scale on a fixed weekly cadence. If a competitor's disclosed spend consistently climbs from Thursday through Sunday, launching your own tests Monday through Wednesday buys you cheaper delivery for nothing. This costs no extra budget and it is the option most often ignored, because it is the only one that is not about money.

The response nobody in this category writes about

Every response above is about spend, creative, audience or timing. None of them touches conversion rate, which is the cheapest lever a thin-margin operator has: if the CPM rises 20% and your conversion rate rises 20%, your cost per acquisition has not moved.

Estrid's landing pages, ranked by how many ads point at them, show what that looks like when someone is doing it properly. The top page has 758 ads and €707k behind it, the second has 266 and €91k. They are the same page in two languages, and the price is localised: "A great shave for only £7.95" in English, "Eine erstklassige Rasur für nur 9,95€" in German. Competitors often rework landing page messaging during expensive periods, adding urgency, proof or a stronger guarantee, and that change is visible here before it is visible anywhere else. Pull any competitor's landing pages, ranked by the ads pointing at them.

How long will it last

The honest answer is that you cannot know, but you can watch for the tell. If competitors keep their winning ads live through an expensive stretch, their unit economics still work at the higher cost and they are treating the spike as temporary. If they start killing long-running winners, the squeeze is reaching brands with more margin than you and it is likely to last.

Cross-platform movement is the other early signal. If the brands in your niche are shifting budget off Meta, the pressure on Meta drops and your CPM may fall on its own. If their activity is rising everywhere at once, prepare for higher costs across the board rather than waiting for this one to pass. One note on that: TikTok in this index carries organic engagement only, followers, views and engagement rate. There is no TikTok spend figure and no TikTok ad rank anywhere in the product, so cross-platform reading here means comparing ad counts and organic signals, not spend against spend.

Five minutes on a Monday prevents most of this. Open the three to five competitors you actually compete with, check two numbers for each, the disclosed daily spend and the count of ads launched in the last seven days, and note whether either is climbing. When both climb for the same brand in the same week, you have 48 to 72 hours before it reaches your Ads Manager. Check your own numbers against the brands in your category.

Facebook CPM spike or Meta CPM spike

They are the same thing and the same auction. Meta renamed the company in 2021 and the ad platform kept both names in circulation, so "Facebook CPM spike" and "Meta CPM spike" describe one event: the price of an impression in an auction you share with everyone targeting the same people. Everything above applies whichever word you searched for. The one place the distinction still matters is placements, since Facebook feed, Instagram feed, Reels and Audience Network clear at different prices, and a shift in your placement mix can move a blended CPM without any competitor doing anything at all.

Method

Every figure in this article came from one query against the live Meta ad index on 13 September 2026, and every figure in a numbered section was read off the screenshot beside it rather than recomputed.

The pull asked for active Meta ads with at least €800 a day of disclosed EU and UK spend, sorted by total disclosed spend, six thousand rows deep. That produced 311 brands, of which 101 ads across 66 brands survived the liveness filter: an end date within ten days of the index date, and a run of at least fourteen days. The rest were ads the index still labels active whose last recorded day was months ago, which is a documented trap rather than a surprise.

Then the list was cut by hand, on evidence, and the rejections are worth naming. Regional storefronts were dropped because their ad sets and spend figures do not survive being checked. So was a brand whose own store description reads "The STIHL online shop has moved", two flagged as dropshipping, one generalist marketplace whose store description is literally "My Store", and three whose descriptions were too thin to verify what they actually sell. The largest spender in the original set, at €9,165 a day, was cut for two reasons: it is a regional storefront, and its panel rounds to the same €100k and 8.7 million reach as Spiral's, which would have read as a copy-paste error whether or not it was one.

What each brand sells was taken from its own store description on its brand card, never from memory. That check exists because a previous article in this series described a collagen-mask skincare brand as a skateboard company, and all three of the mistakes it caught read perfectly plausibly.

Three limits worth stating. Rank badges are per brand and worldwide, so "top 1% of 425 active ads" means within that brand's own 425 ads across every market, not across the index. Spend and reach are the EU and UK disclosed slice only. And no CPM appears anywhere in this data, including ours, because no advertiser's CPM is published by anyone. This article measures what goes into the auction, which is the closest observable thing to the cause.

Where the list came from

Discovery sorted by total adspend is the same query in the interface, with reach, total spend and daily rate on every card. The two at the top of the whole Meta index when this was pulled belong to Shopify, at €617k and €554k, running €11.9k and €10.7k a day. They are a useful reminder of the scale above these eight, and of who is setting the floor in a lot of auctions.

API and MCP

This is the query that produced the list above, unedited:

curl -H "X-API-Key: $BRANDSEARCH_API_KEY" \
  "https://api.brandsearch.co/v1/meta-ads/search?status=active&daily_spend_min=800&sort_by=eu_total_spend&sort_order=desc&page_size=100&fields=id,brand_id,status,start_date,end_date,eu_total_spend,eu_daily_spend,eu_total_reach,reach_rank,duplicate_count"

Two parameters carry the weight. daily_spend_min filters on disclosed spend per day rather than lifetime total, which is what separates an advertiser buying the auction this week from one that accumulated a large number slowly. duplicate_count is the versions figure, and it is the cheapest scaling signal in the index: filter on duplicate_count_min=5 and you have a list of brands that have already found their winner.

Always request end_date and status together. An ad that stopped ten months ago is still reported as active, so liveness has to be decided on the date rather than the label. The same index is available over MCP, so Claude, Cursor, ChatGPT or any other tool can run this check without a browser. Connect it to your own AI.

Brandsearch indexes 223M ads across Meta and TikTok, and 13.7M stores (9M Shopify).

Questions people ask

Why did my Meta CPM increase overnight?
If nothing changed in your account, the most likely cause is that the auction got more expensive rather than your ads getting worse: another advertiser started buying the same impressions. Check whether the brands in your niche show a spend ramp or a rank climb in the 48 to 72 hours before your costs moved.

How do I find out which competitor is driving my CPM up?
Sort the Meta ad index in your category by disclosed daily spend, look for ads climbing in rank or stacking duplicate versions, then match the dates against your own CPM chart. If the ramp and the rise line up and the spend landed in your market, you have your answer.

Can I see my competitor's actual CPM?
No, and neither can any tool that says it can. CPM is reported inside an advertiser's own ad account. What is published, under EU and UK transparency rules, is the spend and reach on each ad, which is the input to the auction rather than the price any one advertiser paid.

Is a CPM spike always caused by competitors?
No. Seasonality, audience overlap between your own ad sets, a change in placement mix and creative fatigue all raise CPM without anyone else doing anything. That is why the check above starts by asking whether anybody else moved at all, and ends by sending you back to your own account if nobody did. If what actually moved was your return rather than your cost, a ROAS drop has its own diagnostic.

Should I pause my ads when CPM spikes?
Usually not, and never on the strength of the CPM alone. Work out whether you are still profitable at the higher cost first. If the competitors in your niche are running their winners straight through the expensive period, the economics evidently still work for them, and pausing hands them the impressions you were arguing over.

How long do CPM spikes last?
That depends on whether the cause is a single scaling advertiser or a seasonal shift with several. The observable tell is whether competitors keep their long-running winners live. If they do, they expect it to pass. If they start killing ads that have run for months, it is squeezing brands with more margin than you and it is likely to last.