Meta Ads KPIs: Formulas and Decision-Focused Reporting

Delivery evaluation should separate repeat appearances from the number of different people reached. Repeated exposure increases impressions; it does not create a new person each time. Summing daily reach does not establish unique people across the full period.

Frequency is an average, not proof that everybody saw the ad equally often. Rising frequency and cost can suggest fatigue without establishing it. Review seasonality, budgets, audience scope, message and purchasing process. Avoid mandatory thresholds such as three or four for every industry.

CPM measures cost per thousand impressions. Competition and delivery mix can affect it; a decline does not prove greater demand or profit. Compare available breakdowns under a consistent objective and period. Paid reach, organic visibility and projected audience size are distinct measurements, not interchangeable market-share figures.

The destination review needs an explicit click definition. All clicks, link clicks and clicks leaving Meta are different counts. A link click does not establish a successfully loaded page or purchase.

Current landing-page guidance describes successful destination loading after a link click; some views can be modelled where data is incomplete. Instant Experience stopped counting within this measure in June 2025. Do not apply the old mandatory-Pixel recipe universally to current implementations.

If clicks and views differ, examine speed, connectivity, redirects, measurement scope and device experience. A single ratio does not prove the website is slow. Name the conversion-rate base: session-to-purchase and click-to-enquiry answer different questions. GA4 bounce rate measures non-engaged sessions; it is not the same as the exit rate.

Content evaluation should preserve the type and quality of likes, comments, shares and saves. More comments can reflect questions, complaints or confusion. More sharing does not automatically produce free sales or an algorithmic discount.

For a custom engagement rate, specify the actions counted and whether reach or impressions is the denominator. Total actions are not unique people. Shares divided by impressions can describe a sharing rate; it does not establish organic reach growth or exponential acquisition. Net follower growth also need not come exclusively from ads.

A defined opening measure could use three-second plays divided by impressions. Retention and completion measures need an explicit denominator and video duration. Do not present hook and hold rates as identical standard columns in every interface. Examine whether viewing actually progresses to appropriate enquiries or purchases.

The measurement-scope review should explain the roles of browser events, server events and company records. Duplicate events or incorrect values can inflate KPIs. CAPI does not recover every lost signal or remove permission and data-use limitations.

Meta’s March 3, 2026 announcement narrows click-through attribution for website and in-store conversions to link clicks and moves other social interactions into engage-through attribution. Rollout is staged. Record active definitions and windows when comparing periods; a reporting shift is not automatically a sales decline.

Meta, GA4 and sales systems can use different models and coverage. Attributed revenue differs from additional revenue that would not occur without advertising. Several channels can claim the same order. Do not add overlapping windows or average disagreeing systems to invent a correct result.

The ROAS, CAC and MER review should distinguish new and returning customers, cancellations, returns and revenue definitions. ROAS is a revenue ratio rather than profit or ROI. High ROAS may reflect a small, constrained demand group.

If a simplified break-even calculation uses pre-ad contribution rate, its ROAS boundary is one divided by that rate; explain included costs. Product, payment, delivery, returns and discounts matter before choosing a target. Turkey’s current 5% Meta location fee sits outside campaign spend and needs separate consideration when reviewing total cost.

In lead campaigns, qualified-enquiry cost and verified new-customer cost differ. Identify unresolved records when sales take time. A competitor’s long-running public advertisement does not reveal its ROAS or profitability.

The customer-quality report can include returns, cancellations, delivery issues and response time as company data alongside platform results. For CSat, define the response scale, survey timing and the satisfied-responses denominator.

NPS subtracts the percentage rating recommendation zero to six from the percentage rating nine or ten; seven and eight form a separate group. A small or selected sample may not represent every customer. These are not automatic Meta ad-quality scores. Avoid presenting old feedback-score thresholds as universal account enforcement rules.

Ad relevance diagnostics compare perceived quality, expected engagement and conversion rankings against relevant competition. The diagnostics themselves are not auction inputs and are unavailable below 500 impressions. An above-average rank does not guarantee profit; a low rank alone is not a stopping decision.