A 14-slide Meta report can tell you that 412,000 people saw an ad and still leave you guessing whether the business made money. I’m tearing down a composite monthly agency report because its familiar order reveals the problem: reach, CPM, CTR and Results get the prime real estate; margin-adjusted ROAS, what changed this week and why, and the winning creative test are buried or missing.

The composite follows the order a client would read it: a reach chart, traffic trends, a Results table, platform ROAS, a list of optimisations and next month’s plan. The figures are illustrative; the reporting choices are the point. I’ll give each slide its verdict before moving on. A polished chart is not a business outcome.

Slide 1: Reach and impressions prove the ads ran

The opening slide has a blue line rising to the right: reach 412,000, impressions 680,000, frequency 1.65. Video views and page likes fill out the bottom row. It looks like progress before the reader has seen a sale.

Verdict: this slide shows distribution, not performance. Reach and frequency are estimated from sampled data, not counted the way spend is. And reach, impressions, video views and page likes can rise while profit falls. I used to put this sort of slide first when I managed accounts. I was wrong. It trains a client to celebrate an ad being seen when being seen was never the job.

I would keep the distribution figures as context if they help explain a later result. I would not let them open the meeting. Before anyone admires that blue line, ask how many of the people reached became customers.

Slide 2: CPM and CTR need their definitions attached

Next comes CPM at $18.40, down 12% month over month, and CTR at 1.84%, up a touch. The arrows point the right way. The labels do not tell us enough to follow them.

First, which CTR? Clicks (all) includes reactions, comments, shares and photo expands; link clicks counts destination links. CTR (all) can therefore look better, and CPC (all) lower, than their link-click equivalents. If a report switches between them or never names the measure, its neat trend line cannot support a decision.

CPM needs context too. A lower price for impressions can reflect broader targeting, cheaper placements or more Reels inventory. It does not, by itself, mean cheaper customers. I want link CTR and landing page views side by side. If people still click links but fewer landing pages register a view, check load speed and tracking before blaming the ad. That gives the manager somewhere useful to look instead of another arrow to colour green.

Verdict: keep CPM and CTR as diagnostics. Name the click definition, show the journey to the landing page, and do not present either metric as profit.

Slide 3: The Results total adds unlike things together

Slide 3 looks closer to the money: 1,842 Results at $8.12 per result. The table has a total row, which makes it tempting to treat every entry as the same outcome. They may not be.

Results is the outcome specified by each campaign’s objective and settings. A sales campaign counting purchases and a traffic campaign counting clicks cannot be added into a meaningful performance total. I have seen reports put purchases and landing page views in separate rows, then sum them at the bottom. The spreadsheet can do that. The business cannot bank the answer.

Cost per result is spend divided by those Results, so it inherits the same problem. Say you spend $20k a month. A blended $8.12 per result tells you little if cheap content views sit beside purchases costing $140 each. The average is precise about a category nobody would choose to buy.

Verdict: require one named business outcome and its cost. Show purchases or qualified leads under a fixed definition; keep other events separate. If the definition changes from row to row, the total is not a management result.

Slide 4: Platform ROAS hides its most important setting

This is the screenshot slide: 4.2x ROAS, $36,400 in conversion value on $8,667 spend. In small gray type sits the attribution window: 7-day click, 1-day view. The footnote deserves more attention than the headline.

Website purchase ROAS divides attributed website purchase conversion value by spend and depends on purchase values reaching Meta through the pixel and Conversions API. Change what counts toward attribution and the displayed ROAS can change without a matching change in revenue. The same spend can read 3.5x under 7-day click plus 1-day view and 2.5x under 1-day click. That is a difference in settings, not two different bank balances.

The gap grows when someone puts Meta beside a store report. One worked comparison on $2,000 spend shows 4.5x under the ad-set setting, 3.0x on Meta 7-day click, 2.4x on Shopify any-click and 1.8x on Shopify last non-direct click. Each view answers a different attribution question. None should quietly replace another halfway through a month. Reporting definitions also changed in March 2026, another reason to print the window rather than assume everyone remembers it.

Verdict: keep platform ROAS, but print its attribution window prominently and put store revenue next to it. The first lets you read Meta’s trend consistently. The second stops you mistaking Meta’s claim on revenue for the whole business result.

Slide 5: “Optimisations made” names no decisions

The next slide says: “Optimised audiences, refreshed creative, adjusted budgets for efficiency.” Three verbs, zero campaign names. There is no date, old budget, new budget or reason. I call it the horoscope slide because you could paste it into almost any account and month.

A useful report pairs spend and the agreed outcome with cost, prior-period trend, changes, learning and next steps. The change itself should be legible without a call to the manager. For example: On Tuesday, move $400 a day from prospecting broad into the winning UGC test because its cost per purchase is $31 versus $52 for the control over seven days. That is a sample of the detail I want, not a claim about this composite account. It identifies the money, the comparison and the reason for acting.

Verdict: the report needs a dated change log, not a list of verbs. What changed this week, and why? A manager should be able to answer in a line per decision. If the agency cannot name those decisions, “optimised” gives you nothing to audit.

Slide 6: Next month’s plan needs a stopping rule

The final slide promises to “scale winners,” “launch new UGC concepts” and “explore lookalikes.” It sounds active. But it does not identify a winner, attach a budget or say when a losing test stops. That leaves next month’s report free to call almost any outcome a success.

A plan I can assess would say: Spend $1,500 testing four new hooks against the current winner at $50 a day each; pause any concept above $45 cost per purchase after 3,000 impressions. Those are illustrative rules, not universal thresholds. Their value is that the manager commits to a hypothesis, money and a decision point before seeing the result. What was learned and what happens next belong in the monthly report. Otherwise the plan is just a collection of future-tense verbs.

Verdict: make the manager write the test before spending on it. If the plan could have appeared unchanged in last month’s deck, you are paying for monitoring dressed as management.

The missing lines: a creative winner and margin math

After six slides, I can see what Meta reported and what the manager says happened. I still cannot identify the winning ad or tell whether the return cleared the business’s margin. Those omissions matter more than another chart.

First, name the creative test that won. The composite names no ad, control, losing concept or pause decision. In a review of 200-plus accounts, only 5–7% of tested Meta ads met the stated winner standard. That standard requires CPA at or below target for seven days after learning exit; the testing approach described there keeps proven ads in rotation while reserving spend for new concepts. The exact routine is less important here than the reportable decision: which creative earned more budget, against what control, and what stopped running? “Refreshed creative” is not an answer.

Second, put contribution margin beside ROAS. Revenue is not profit. Break-even ROAS is one divided by contribution margin: at 40% margin, break-even is 2.5x; at 30%, about 3.3x; at 25%, 4.0x. The same platform ROAS can therefore clear the bar for one product and miss it for another. A margin-adjusted view makes that bar visible instead of asking an owner to applaud 4.2x on sight.

I also want blended MER: total revenue divided by total ad spend, shown weekly beside platform ROAS. If MER falls while channel ROAS holds, investigate the gap before calling it growth. The two measures do different jobs. Meta ROAS describes revenue Meta attributes to its ads; MER lets you check the wider business result. Neither excuses a manager from explaining what changed.

Checklist: what to demand before you renew

Copy this into an email before approving another month. Ask for the same definitions each week and figures you can read without screenshots. The order starts with whether the spend cleared your margin, then moves through the evidence behind the manager’s decisions.

  1. Spend, revenue and margin-adjusted ROAS: Put Meta spend, store or CRM revenue, contribution margin and break-even ROAS beside the actual return.
  2. A fixed attribution window: Print it on every report; do not switch windows mid-comparison.
  3. Cost per purchase or qualified lead: Name one outcome and keep purchases, leads and landing views out of a blended Results total.
  4. Link CTR and landing page views: Label the CTR and show whether clicks turn into recorded visits.
  5. What changed this week and why: Give the date, campaign, budget before and after, and the reason for each launch, pause or move.
  6. The winning creative test by name: Show its control, cost per purchase, what lost and what receives budget next.
  7. Blended MER beside platform ROAS: Show total revenue divided by total ad spend weekly. Skip this check and you can keep spending while the business-level return slips behind a healthy-looking platform number.

Overall verdict: keep the evidence, replace the presentation

When I compare managers, I would send this checklist before signing and ask for a redacted weekly report filled in. Definitions and dated decisions tell me more than the agency logo on 14 slides. If the answer is another deck with reach first, I know where the manager wants my attention.

I would keep the raw material from the ROAS, optimisation and planning slides. Then I would rebuild them around three lines: margin-adjusted return, what changed this week and why, and the named creative winner. Reach, CPM and CTR can stay in the appendix when they explain one of those lines. They do not get to stand in for it.

That is the standard I hold my own work to now. At groas, the Meta work runs on a flat $1,499 a month, with audiences, budgets and ads worked around the clock, new ads tested and winners scaled. Changes carry their reasoning, so the weekly report is a work log rather than a rewrite of one. Put that beside your current report and ask which one tells you whether you made money.