When Meta ad costs rise, the cause is in one of three places.

1. Make sure the rise is real

Open your store's order report and Meta Events Manager for the same period. Count how many orders the store recorded and how many Purchase events Meta received. Our rule of thumb is that the gap should be no more than about five percent.

The comparison leads to one of three situations:

• Store sales unchanged, Meta shows fewer purchases. The problem is tracking. The advertising may work exactly as before, but Meta does not see every purchase and reports their cost too high.
• Store sales down across all channels. The problem is demand or the store: season, price, stock, a competitor or a technical fault on the site. Meta is the messenger, not the cause.
• Only Meta-driven sales down. The problem is in the ad account or in tracking: a setting changed, the audience wore out or tracking broke specifically on Meta's side.

The check takes ten minutes and often saves a month of wrong fixes. With GreenGoKauppa, a sales collapse was traced to two separate causes, tracking and a change of season. Without the comparison it would have been easy to read them as one problem: tired ads.

”Just Media noticed a problem in our reporting that was distorting the results.”

The five-minute tracking check

If the gap between orders and purchase events is more than five percent, go through these five points before you touch the campaigns:

1. Does Purchase fire on the right page? The event must fire on the order confirmation, not at the start of checkout. If it fires at checkout, Meta counts abandoned orders as purchases and learns from the wrong people.
2. Does it fire twice? Two scripts, or an app and theme code on top of each other, double the purchases. Place a test order and see how many events it produces.
3. Are browser and server connected? If you run both the pixel and the Conversions API, both must send the same event_id so Meta merges them into one purchase. Without it, purchases double, and if one side is missing entirely, some purchases disappear.
4. What is the match quality of the purchase event? Meta rates on a scale of 0 to 10 how well it can connect an event to a person. If your Purchase event is clearly below 6, events are missing identifiers such as email, and Meta cannot use the purchases for learning.
5. Do express checkouts show up? Apple Pay, Shop Pay and PayPal can skip the page where the pixel is installed. Place a test order with an express checkout as well.

Fixing tracking does not improve a single ad. It restores the numbers you judge ads by. That is why we go through tracking and data quality before the first new ad.

2. The auction: price of impressions or purchase rate

Once the numbers are reliable, break the cost per purchase into its parts. It is made of three numbers, and each has a different cause and a different fix:

CPM tells you what a thousand impressions cost. Click-through rate tells you whether the ad is interesting. Conversion rate tells you whether the clicker buys. See which of the three changed, then read on:

• CPM up, frequency rising, but reach not growing. The audience is saturated. The same person sees the same ad again and again, and Meta charges more for their attention. Broaden the targeting or change the angle, not just the image.
• CPM up, frequency not. There is more competition in the auction. From October to December that is normal, and the fix is to accept the more expensive impression or move tests to a quieter time.
• CPM unchanged, CPA up. The problem is not in the auction. It is after the click: in the ad, on the page or at checkout. Go to step three.

The most common mistake here is to raise the budget when CPM rises. That buys more expensive impressions for the same tired audience.

3. The store: where the buyer disappears

If CPM does not explain the rise, the buyer disappears somewhere between the click and the order confirmation. Three numbers tell you where:

• Click-through rate fell. The ad or its promise has worn out or is hitting the wrong buyer. This is the only case where the fix is a new ad, and even then a new angle rather than a new image.
• Page conversion fell, clicks did not. The problem is on the landing page: price, offer, delivery time, availability, reviews or a slow mobile page. First check whether something changed. A new theme, a new pop-up or the most popular size selling out is enough.
• Checkout completion fell. The problem is technical or in the costs: shipping is revealed only at checkout, a payment method is missing or a field does not work on a phone.

Place a test order on a phone, not a computer, because most Meta traffic comes from phones. Walk the whole path from ad to order confirmation and time it.

What not to do

• Change everything at once. If you swap the ads, the audience and the budget on the same day, you never learn what helped. Next time you start from zero again.
• Raise the budget on top of the problem. A bigger budget fixes none of the causes above. It makes the same mistake with more money.
• Pause the ad set for a week. A break of more than seven days restarts Meta's learning from scratch. Lower the budget rather than switch off.
• Read ROAS without margin. A more expensive purchase can still be profitable if the average order grew. Compare cost per purchase with your break-even, which we calculate in Set your Meta ads budget from purchases, not gut feeling.

The order, in short

1. Compare store orders with Meta purchase events. If the gap is more than five percent, fix tracking first.
2. Break the cost per purchase into parts: CPM, click-through rate, conversion rate. See what changed.
3. If CPM rose, check frequency and reach. If not, place a test order on a phone and find the point where the buyer disappears.
4. Only then decide whether the ad needs to change. If it does, change the angle, not the color.

Most of the time the ad did not go stale overnight. Something around it changed, and in this order that something is found faster than by guessing.