Meta Ads Metrics That Actually Matter for E-Commerce (And 3 That Don't)

What Meta Ads metrics actually matter for e-commerce in 2026?

Five metrics actually move e-commerce unit economics: ROAS, CPM, hook rate, scroll-stop rate, and CTR. Three commonly-cited metrics are noise: raw reach, raw impressions, and engagement without conversion. This is the operator-side view of the metrics — what to optimize for in your own Ads Manager, and what to ignore.

The competitive-intelligence-side view is different. The public Ad Library exposes the advertiser identity and creative but NONE of the performance metrics. Competitor metrics are inferred from observable proxies (run duration, ad count, hook stability). For your OWN ads, the 5 metrics below are the actual KPIs.

The 5 metrics that matter

Metric 1 — ROAS (return on ad spend). The only number that directly answers "is this profitable?" Revenue from the ad divided by spend on the ad. ROAS above your contribution-margin floor = scaling signal. ROAS below = kill the creative. Every other metric on this list is a leading indicator of ROAS; ROAS itself is the outcome.

Metric 2 — CPM (cost per 1,000 impressions). The auction-pressure proxy. Rising CPM in your niche while ROAS holds means competitors are bidding harder for the same audience; you need sharper creative or wider audience. Falling CPM in your niche means competitors are pulling; the auction is friendlier. CommonWealth Ops's multi-advertiser convergence tracking surfaces the structural auction pressure even before your CPM reflects it.

Metric 3 — Hook rate (3-second video view rate / impressions). The algorithm's first-quality-signal. Hook rate above 35-40% on cold audiences means the creative is earning attention; below 25% means the algorithm will de-prioritize. CommonWealth Ops's persistent-ad threshold (14+ days) correlates with strong hook rates because Meta's algorithm punishes weak hooks fast.

Metric 4 — Scroll-stop rate (impressions where viewer stops scrolling). A more granular signal than hook rate; some platforms expose it as "thumb-stop rate." Indicates whether the visual + first-2-seconds-of-audio earn enough to halt the scroll instinct. Improving scroll-stop rate by 5pp typically produces a larger ROAS shift than improving CTR by 5pp.

Metric 5 — CTR (link click rate post-impression). The post-hook conversion proxy. CTR above niche-average means the creative earned the click; below means even good hook performance isn't translating to action. The 2026 e-commerce benchmark sits around 1.0-1.8% for cold-audience prospecting; warm retargeting carries higher.

The 3 metrics that distract operators

Distractor 1 — Raw reach. "Our ad reached 500,000 people." Reach without conversion is bragging rights, not unit economics. Meta's algorithm can deliver high reach on creative that doesn't convert because reach itself is the optimization target only when you specifically ask for it (and you usually shouldn't).

Distractor 2 — Raw impressions. Same problem as reach, with an additional confusion: high impressions on a small audience drives frequency, which drives fatigue, which drives ROAS down. Impressions ARE useful when tracked against frequency (frequency-adjusted impressions show how much audience saturation is happening) but raw impressions alone is noise.

Distractor 3 — Engagement without conversion. Likes, comments, shares on an ad that isn't generating revenue. Engagement metrics serve organic content optimization; they don't serve paid-conversion optimization. The "Comment 10 for link" hooks captured in CommonWealth Ops's data USE engagement (comments) AS the conversion mechanism — that's a different play; pure engagement-without-action is the distractor.

How competitor metrics relate (and what's actually observable)

Here's the structural divide: your OWN Meta Ads Manager exposes the 5 metrics above; the public Ad Library does NOT. So competitor-intelligence work uses observable proxies for performance:

Run duration as ROAS proxy. An ad still running 14+ days after launch has cleared the brand's internal ROAS kill-switch at least twice. CommonWealth Ops uses the 14-day persistence floor as the working signal that "this creative is making the unit economics work for that advertiser."

Active-ad count as scaling-signal proxy. A brand running 7 active ads vs 2 in the prior month has tripled spend behind validated creative. Frequency acceleration (one of the 5 product-scaling signals CommonWealth Ops codifies) reads the metrics indirectly through ad-count delta.

Hook stability as hook-rate proxy. An ad running 21+ days on the same hook variant indicates the brand's hook rate is holding under repeat exposure. Hook variants that rotate every 7 days indicate hook fatigue.

The proxies aren't as precise as the raw metrics, but they're sufficient for the operator-side decisions competitive intelligence supports: which niche to enter, which competitor's audience to target, which hook archetype is currently winning.

How does CommonWealth Ops surface this for operators?

CommonWealth Ops scrapes Meta Ad Library and TikTok Ad Library weekly, persists the observable proxies (run duration, ad count, hook stability, frequency acceleration), and surfaces them in the per-niche intelligence report. Subscribers get the proxy-level signals; their own Ads Manager provides the 5 direct metrics for their own creative.

The methodology is fully documented in our how-CommonWealth-Ops-collects-intelligence post. For the technical walls limiting what's exposed publicly, see our meta-ads-library-api-what-it-returns-and-what-it-doesnt post.

See pricing →