The essentials
- One main KPI per objective: unique reach and cost per thousand for awareness, cost per outbound click for traffic, cost per qualified lead for prospecting, ROAS on margin for sales.
- Three levels of reading: result (conversion, ROAS), diagnosis (CTR, conversion rate, frequency), context (CPM, impressions). Never decide on the context level.
- Meta alert thresholds for 2026: outbound CTR below 0.8%, frequency above 3 to 4 in prospecting, platform / GA4 gap above 40%.
- To be ignored: the number of likes, "clicks (all)", raw reach with no cost, ROAS with 7-day view-through attribution.
Meta Ads Manager offers more than 300 metrics, LinkedIn Campaign Manager around a hundred. A dashboard that displays twenty of them is useless: decisions are made on two or three figures, chosen according to the campaign objective, and the rest serve to explain their variations. Here are the KPIs to keep by objective, the alert thresholds observed in France in 2026 and the order in which to read them.
The main KPI depends on the objective
| Objective | Main KPI | Secondary KPIs | 2026 benchmark |
|---|---|---|---|
| Awareness | Cost per thousand people reached (unique reach), not per impression | Frequency, 50% video view rate, estimated ad recall | Target frequency 2 to 4 over the period; Meta reach CPM €4 to €8 |
| Traffic | Cost per outbound click, then cost per GA4 session | Bounce rate, session duration, pages per session | Click → session loss below 40%; otherwise a page problem |
| Lead generation | Cost per qualified lead (after sales screening) | Cost per raw lead, qualification rate, meeting rate | Meta €8 to €40 raw, LinkedIn €40 to €150 raw; qualification 30 to 60% |
| E-commerce sales | ROAS calculated on margin, or cost per purchase compared with the margin | Conversion rate, average basket, share of new customers | Median Meta ROAS 2.5 to 4; threshold specific to each margin |
| App | Cost per install, then cost per active user at 7 days | Retention rate, in-app events | Install €1 to €4; 7-day retention 20 to 40% |
Cost per qualified lead and ROAS on margin require data from outside the platform: the CRM for the first, the margin per product for the second. Without them, you steer on raw leads and a revenue-based ROAS, two metrics that can look good while the campaign is losing money. The calculation is detailed in ROAS in SMA.
The three levels of reading
Each KPI belongs to a level, and the rule is simple: you decide on level 1, you explain with level 2, you set the context with level 3.
- The result: conversions, cost per conversion, ROAS, qualified leads. These are the only figures that justify increasing, reducing or cutting a budget.
- Diagnosis: outbound CTR, page conversion rate, frequency, video view rate, cost per outbound click. They say why level 1 is moving: creative, page, saturation.
- Context: impressions, reach, CPM, clicks (all), engagement. They describe the buying conditions but are never enough for a decision.
A classic mistake is to cut a campaign because its CPM has risen by 30%, when its cost per conversion is stable; or to increase it because its likes are soaring, when it is not converting. The reading traps are developed in the mistakes in interpreting KPIs.
The 2026 alert thresholds by platform
| Diagnostic metric | Meta | TikTok | What it reveals below the threshold | |
|---|---|---|---|---|
| Outbound CTR | Below 0.8% | Below 0.35% | Below 0.6% | Creative or promise not suited to the audience |
| Frequency (prospecting, 7 days) | Above 3 | Above 4 (30 days) | Above 2.5 | Saturated audience, refresh or widen it |
| Page conversion rate | Below 1% in e-commerce, below 3% in lead generation | Landing page, offer or off-target traffic | ||
| 3-second video view | Below 25% | Below 20% | Below 30% | Hook too slow |
| Platform / GA4 conversion gap | Above 40% | Measurement problem (duplicate, view-through attribution, consent) | ||
The cost ranges (CPM, CPC) that provide the context are set out in understanding CPM and cost per click in social ads.
The metrics to remove from reports
- The number of likes and followers gained: no reliable correlation with sales; see vanity KPIs.
- Meta's "clicks (all)": they include image expansions and profile opens; use outbound clicks.
- Reach without the cost: a reach of 500,000 people says nothing if you do not know what it cost and who it reached.
- ROAS with view-through attribution: an impression seen without a click followed by a purchase within 24 hours counts as a conversion; in retargeting, this inflates ROAS by 30 to 100%. Read click-based ROAS at 7 days, then incremental ROAS if you can test it.
- The relevance score and the quality rankings: useful for comparing two creatives, useless for steering a budget.
Building the weekly report
A social ads management report fits on one page: per campaign, the spend, the main KPI, its change over 7 and 28 days, and a one-line comment. The diagnostic KPIs appear only for the campaigns whose main KPI has moved by more than 20%. The platform and GA4 data appear side by side for the cost per conversion. The report templates are described in the advertising campaign KPIs.
How GreenRed helps
Rather than juggling several tools, GreenRed's SMO / SMA social media module brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.
Frequently asked questions
How many KPIs should you track per campaign?
One main KPI linked to the objective, two or three diagnostic KPIs, and that is all for weekly management. The other metrics remain available in the manager for the monthly analysis. A report that displays more than six figures per campaign ends up not being read, and decisions are then made on intuition.
Is ROAS a good KPI for lead generation?
No, unless each lead has a reliable estimated value. For services, cost per qualified lead is the right metric, complemented by the meeting and signature rates coming from the CRM. ROAS applies to online sales, where the value of the conversion is known at the moment it happens.
Should you compare your KPIs with industry averages?
With caution. The published averages mix different advertisers, objectives and countries; the gap between two accounts in the same industry commonly reaches a factor of three. Use them to spot a gross anomaly, then build your own benchmark from three months of data.
How do you measure the real contribution of social ads to sales?
Three complementary methods: click-based ROAS in the platform, data-driven GA4 attribution for the share actually attributed, and an incrementality test (geographic holdout or a Meta conversion lift test) to measure the sales that would not have happened without advertising. The third figure is the only one that really measures the contribution.