Your cost per thousand went up. Check these, in order.

The cost to show your advert to a thousand people went up, and the account feels broken. Before you change anything, work out which of four things happened, because the fixes are different and three of them are commonly applied to the wrong problem.

Cost per thousand impressions is a price, not a performance metric. It tells you what the auction charged to put your advert in front of a thousand pairs of eyes. It says nothing about whether those were the right eyes, and treating it as a scorecard is how accounts get optimised into cheap, worthless reach.

Is it the market, or is it you?

Start here, because it is the cheapest question to answer and it stops you from fixing something that is not broken. Every advertiser in your auction is bidding for the same attention, and when more of them arrive with more money, the price goes up for everyone including you.

Seasonality is the biggest version of this. Gupta Media’s tracker of social media advertising costs reports seasonal rises of as much as 66% during the holiday shopping period, with competition around Black Friday and Cyber Monday doing most of the work. If your price rose in the second week of November, the market raised it, not your creative.

How to tell the difference in your own account, without buying a benchmarking tool: look at a campaign you have not touched in weeks, ideally a slow retargeting one. If its cost per thousand impressions rose by roughly the same proportion as everything else, the movement is external. If it held steady while a specific campaign spiked, the problem is that campaign.

Have you run out of new people?

The second cause is audience exhaustion, and it looks like a pricing problem because it is one. When a defined audience has mostly seen your advert already, the platform has to work harder to find someone who has not, and hard work costs money.

Two numbers tell you. Frequency is the average number of times each person has seen you. First-time impression ratio is the share of today’s impressions that reached somebody new. Triple Whale’s creative fatigue framework puts the warning line at frequency drifting past about 3.0 on a seven-day window with first-time impression ratio falling under roughly 50%, and notes that prospecting fatigues far earlier than retargeting does.

The fix is size, not cleverness. Broaden the targeting, remove interest layers that were never doing much, or add a country. A narrow audience is a comfort blanket — it makes early results look precise and then it runs out of people, at which point the price rises with nothing to show for it.

Is the creative tired, or was it never that good?

Creative fatigue and audience exhaustion feel identical from the dashboard and need opposite responses, so separate them properly. Fatigue means the same people are ignoring something they used to react to. Exhaustion means there are no new people left to react at all.

Test it directly. Put a genuinely new piece of creative into the same audience with the same budget. If its price to reach a thousand people comes back down, the creative was tired. If the new piece prices the same as the old one, the audience is finished and no amount of production fixes it.

The uncomfortable version of this diagnosis: sometimes the advert was never strong and it was cheap for a while because the audience was fresh. New creative that immediately performs the same as the old creative is not always fatigue. Occasionally it is a message problem that reach was hiding.

What in the account could be doing this?

The fourth category is self-inflicted, and it is the most common one in accounts run by people who care. Every item below raises the price of reaching a thousand people, and every one of them is something an advertiser did on purpose.

Work down that list before rebuilding a campaign. Most accounts we open have at least two of these active simultaneously, and the owner has spent the month blaming the creative.

When should you leave a rising cost alone?

Here is the case nobody writes about, because it does not sell an audit. Sometimes the price to reach a thousand people rises, and the correct action is nothing.

If your cost to reach a thousand people is up and your cost to acquire a customer is flat or down, you are buying more expensive attention that converts better. That is a good trade and every instinct to optimise it away is wrong. It happens routinely when you move from cheap, distracted inventory towards places where people are actually paying attention, and it happens whenever you start optimising for purchases instead of clicks.

• Reach price up, cost per customer down: leave it. You are buying better people.
• Reach price up, cost per customer up, sales flat: this is a real problem — go back to the top of this list.
• Reach price down, cost per customer up: the worst pattern of the four, and the one people celebrate. You have found cheap reach that does not buy.
• Both down: something changed for the better. Find out what, and write it down before you lose it.

The test is always the same. A price is only bad relative to what it buys. An account managed to the cheapest possible reach will reliably find the cheapest possible audience, which is usually the one least interested in paying you.

What this order costs you when it is wrong

We will concede the weakness of this method: it is sequential, and real accounts break in several places at once. A seasonal rise, tired creative and a budget increase in the same fortnight will each partly explain the others, and working down a list will make you confident about the first cause you find. The discipline that saves you is changing one thing at a time and giving it a few days, which is slower than everybody wants to be.

The one thing not to do is rebuild the account. It resets everything the platform has learned, it destroys the comparison you needed to diagnose the problem, and it feels like progress for about a week. Diagnose, change one thing, wait, measure. Boring, and it works.