How to read Facebook Ads Manager: seven numbers that matter, three that mislead

Ads Manager ships with more columns than anyone reads, and the default view is not the useful one. Most are the same number expressed differently, and several mean nothing until a second number sits beside them. You need about ten.

The seven numbers that carry signal

These seven are a chain, not a list: money buys impressions, impressions land on people, the ratio between the two is how often each person sees you, and it all ends in one output number. Read them in that order and the account starts explaining itself.

1. 1.Amount spent: the one number that is billed rather than inferred. When two numbers disagree, anchor on this one.
2. 2.Impressions: how many times the ad was on screen. Events, not people: one person scrolling past you four times is four impressions.
3. 3.Reach: how many distinct people those impressions landed on. The denominator that says whether you are addressing a crowd or the same room repeatedly.
4. 4.Frequency: impressions divided by reach for the range on screen. Arithmetic on the two rows above, not a separate measurement.
5. 5.CTR (link): the share of impressions that produced a click to your site. Use the link version; plain CTR is one of the three below.
6. 6.CPM: what a thousand impressions cost. The entry price for your audience, and it moves for reasons that have nothing to do with your ad.
7. 7.Cost per result: spend divided by results. The output of the six above, which is why it moves last and why it is the worst column to open when you want a cause.

No number means anything on its own

A 3% CTR sounds better than 1.2% until you put the price next to it. A thousand impressions at 3% produce 30 link clicks; if those impressions cost $40, each click cost $1.33. A thousand at 1.2% produce 12 clicks, and at a $9 CPM each cost $0.75. The weaker CTR is buying clicks at a bit more than half the price. CTR says how well the ad lands on the people who see it, CPM says what those people cost, and neither answers alone.

Frequency needs the date range beside it rather than a second metric. A frequency of 3.0 across 30 days is a person seeing you every ten days; 3.0 across 7 days is a person seeing you every other day. Comparing frequency between two reports built on different ranges tells you nothing.

What a rise means, what a fall means

When a number has moved and you need a first hypothesis, start here. These are not verdicts, only the direction to look first.

Three columns that mislead

Results

Results is not a metric, it is a placeholder. It reports whatever event the ad set was optimized for: purchases, leads, landing page views, video plays. Two campaigns can sit side by side, both reporting 340 results, and describe completely different businesses, because one was buying leads and the other page loads. Check the objective before comparing cost per result to anything, and add the conversion column you care about by name. Results is a summary column, and summary columns drop the unit.

CTR (all) and clicks (all)

Clicks (all) counts every click on the ad unit: reactions, comments, shares, taps on the page name, image expands, and somewhere in there the click that reaches your site. On a post that collects engagement it sits well above link CTR while your site sees nothing new. Keep the link versions and drop the (all) ones, then put link clicks next to landing page views: a wide gap between those two is a page problem, not an ad problem.

Conversions

Reported conversions are an attributed, deduplicated, partly modeled estimate. That is not a complaint about Meta, it is what the column is built to be. It answers how many conversions can be credited to ads under the current attribution setting, which is a different question from how many customers you have.

Why Ads Manager and your CRM will never agree

Four mechanisms pull the two counts apart, and they stack rather than cancel.

1. 1.Attribution window. The default is 7-day click and 1-day view, and a conversion is credited to the day of the click, not the day of the sale. Someone who clicked Monday and bought Friday sits in Monday's row here and Friday's row in the CRM. Change the window and the same past week reports different numbers.
2. 2.Deduplication. With the pixel and the Conversions API both running, the same action arrives twice, and Meta collapses the pair by event name and a shared event ID. A missing ID, or two sources generating different ones, turns one purchase into two.
3. 3.Lost events. Tracking prevention, ad blockers, app tracking opt-outs, and every conversion on a channel the pixel cannot see: phone calls, DMs, someone walking in. The account undercounts what it cannot observe and models part of the difference back in.
4. 4.The CRM counts different objects. Duplicate submissions, spam, test entries and one person filling two forms are one conversion to Meta and several rows in most CRMs.

Reconciling the two to zero cannot be done, and hours disappear into trying. Make the CRM the source of truth for money, use the account for direction, and watch the ratio between them. Whatever that ratio has been for six weeks is your normal, and the week it halves, tracking broke. The question is never which system is right, it is whether the gap changed while you changed nothing.

Server-side events through the Conversions API close part of that gap, because they carry what the browser could not send. Two rules keep them from making the count worse: an event that duplicates a browser event needs the same event ID, and an event marking a later step in the funnel needs its own name, or the same person is counted twice. That is the seam Aevin sits on: a lead reaching a stage marked qualified or won fires the event from the server, under its own name, with email and phone hashed first.

A ten-minute weekly read

1. 1.Set the range to the last 14 days and break one campaign out by day. A single row of totals hides everything that matters.
2. 2.Check spend first. In a summary row, a campaign that stopped working and a campaign that stopped spending look identical.
3. 3.Read frequency and link CTR together. Frequency up with CTR down is fatigue. Frequency flat with CTR down is the audience or the offer.
4. 4.Read CPM on its own, then against last month. CPM up while CTR holds means the auction got more expensive and there is nothing in the ad to fix.
5. 5.Confirm what Results is counting before you read cost per result at all.
6. 6.Compare the account's conversion count with the CRM for the same period, and that ratio with last week's. Then change one thing, and write down the date.

None of this needs intuition. It is a fixed sequence of comparisons against the account's own history, and it is the first thing to fall out of a busy week. That is the argument for handing it to something that never gets busy: the analyst inside Aevin runs these comparisons on a schedule and reports which numbers moved, with the figures behind the finding.