FR
← Blog
Advertising

Facebook ads report: how to read what it really says

September 2026 · Advertising · Quebec

A Facebook ads report answers exactly one question: what Meta was able to observe and claim credit for inside its own attribution window. Your accounting answers a different one: what you invoiced. Both numbers can be correct at the same time and never meet. Reading a report well means holding both at once and never letting one stand in for the other. Everything else, every column, rate and chart, follows from that distinction.

The confusion is expensive in both directions. One company kills a campaign that was producing phone calls because the results column looked thin. Another lets a campaign run for eight months because the platform shows plenty of results, and never checks how many contracts were signed. The full method sits on the Facebook and Meta advertising page. This page is only about reading the report.

What Meta says about its own columns

Meta documents each of these metrics in its developer reference. The definitions are worth reading word for word, because they are more modest than what people make them say in a meeting.

Two things stand out. First, the word estimated keeps coming back, spend included. Second, the definition of results points at the objective somebody selected. A result is not a unit of business value. It is whatever the person who built the campaign designated as the goal. Change the objective and the column changes meaning, with no warning and nothing in the report to flag it.

So that is the first question to put to anyone presenting you a table: what exactly is a result here? If the answer takes more than one sentence or stays vague, the conversation should end there.

The attribution window is half the number

Meta documents a default window: 7 days after a click, plus 1 day after a view. Nineteen window values are documented in total, including 1-day, 7-day and 28-day windows, for clicks and for views. The same month, on the same ads, does not produce the same result count depending on which window was used.

One documented detail deserves to be better known: inline link clicks use a fixed 1-day-click attribution window. Two columns in the same table can therefore be computed over different periods. Comparing them as if they covered the same interval produces false conclusions, and it is a common mistake.

What to demand is simple. That the attribution window be written on the report itself. And that it never change from one month to the next without being told first. A rise in results that comes from a wider window is not an improvement. It is a change of measuring unit.

The click columns do not count the same thing

Meta documents several click metrics. Clicks, all types, counts clicks on your ads, which includes actions that bring nobody to your website. Outbound clicks counts clicks that take people off Meta-owned properties. Link clicks counts clicks to selected destinations. A click-through rate built on one of those columns does not compare with a rate built on another, even though both get called CTR in conversation.

More important for a small company: Meta documents link clicks and landing page views as two distinct metrics, and even documents a ratio between them, the average number of landing page views per link click. The existence of that ratio as a published metric says the essential thing: the two counts differ. When the gap is large, something is being lost between the ad and the page, and that is a question for whoever owns the website, not a rounding error. I will not explain here why the gap happens, because Meta does not document the mechanism on any page I was able to verify.

What the report cannot know

A custom conversion, as Meta documents it, tracks actions by parsing your website's referrer URLs. Meta gives its own example: a custom conversion that counts visitors who viewed any page with /thank-you in the URL. The documented limit is 100 custom conversions per ad account.

Read the mechanic and you can see the boundary. That metric knows a page loaded after a click. It does not know whether the request came from a real buyer, whether your team called back, whether the quote went out, or whether it was accepted. Nothing in the report can know it, because none of that happens on the platform. This is not a flaw in Meta. It is the limit of what any advertising system can observe from outside your company.

Hence the only honest reading: the platform report on one side, your own count on the other, over the same period. How many requests came in. How many were qualified. How many were invoiced. When the two series diverge, the divergence is itself the useful information, and that is what the monthly meeting should be about.

Picture a process equipment manufacturer in Mauricie seeing far more results in the report than serious requests in the inbox. Both counts can be accurate. The right meeting is about the gap between them, not about the third decimal of the cost per result. Here, the person who shoots the videos is also the person who runs the ad accounts, so the report is not relayed through an account manager who has seen neither the campaign nor the client.

No published benchmark tells you whether your report is good

Meta publishes no average cost per click, no average cost per thousand impressions and no reference click-through rate, for Quebec or anywhere else. Every good CTR figure you have read comes from a third-party aggregator that adds up countries, industries, objectives and audience sizes with no relation to yours. That is not a methodological footnote. It is why those benchmarks contradict each other from one article to the next.

The only benchmark that holds is your own history, read on the same attribution window, with the same definition of a result. A supplier comparing your account to an industry average is comparing you to something nobody measured.

When this report does not answer your real question

For a company whose sales close months later, over the phone, often after a plant visit, the report cannot answer the question the owner is actually asking. The default window closes 7 days after a click. A contract signed in March off an ad seen in October will appear in no column, no matter how good the work was.

What is missing then is not in the platform. It is in your own records: the date each request arrived, what the person said when asked where they heard about you, and what the request became. Kept honestly for a year, that log is worth more than any column, and it belongs to you. No platform will replace it, and nobody should promise you otherwise.

A second, plainer limit. If your campaign produces a handful of requests a month, no percentage swing in the report means anything. At that volume you read the requests one by one rather than reading rates, and the report mostly serves to confirm that delivery is working.

Sources

If you want your report read next to your real sales numbers, we can do that together on a short call.

Book my call
Home · Portfolio

B2B Marketing · Industrial Marketing · Uncompressed · ClientGallery

© 2026 MarketingSTRAT