Why does Meta hide ad spend?

Two reasons: competitive sensitivity (Meta does not want advertisers to know each other's budgets, because it would dampen bidding) and creative-experimentation privacy (advertisers running A/B tests do not want each variant's budget exposed). The Digital Services Act required Meta to expose CREATIVE under EU jurisdictions; it did not require Meta to expose BUDGET. That asymmetry is deliberate. Tools that publish 'spend estimates' are not reading them from Meta; they are modelling them from impression-range data, which has wide confidence intervals.

How accurate are paid tools' spend estimates?

Honestly assessed, the published-spend-range estimates from tools like Pipiads or AdSpy are within roughly 30 to 50 percent of true spend for high-volume ads, and substantially worse for low-volume ads. The error compounds because the estimates are based on impression triangulation, which is itself a triangulation from publicly-visible engagement counts. The number that looks like '€8,000 to €12,000 monthly' might be €2,000 or €20,000 in reality. The directional signal is useful; the specific number is not.

What does ad longevity tell you that spend does not?

Longevity tells you the platform's optimisation engine has decided the ad converts. That decision is made by Meta's algorithm using full conversion data (which you cannot see) and full audience data (which you cannot see). When you read 'this ad ran 60 days,' you are reading the output of a much more sophisticated model than any third-party spend estimator can produce. The longevity is censored at the moment you read it (the ad might run another 60 days or might pause tomorrow) but the signal is still higher-fidelity than modelled spend.

How does the bands framework apply differently per niche?

Niches with fast turnover (fashion, certain DTC categories) have shorter median longevities — a 45 day ad in fast-turnover apparel is high-decile. Niches with slow turnover (supplements, B2B, financial products) have longer median longevities — a 45 day ad in supplements is middle-tier. Read the band relative to the niche median, not absolutely. Our weekly digest does this band calibration per niche so you do not have to.

What other signals does CW Ops publish since not spend?

Per niche per week: dominant hook archetype, dominant emotional trigger, dominant CTA pattern, longevity band distribution, and the pattern half-life estimate. Each is a directly readable, audit-able signal from the public surfaces. None of them require modelling assumptions like spend estimates do. The result is a digest that says 'this is what is working this week, why, and how confident we are' instead of 'this advertiser spent €N.'

competitive-intelligence · 7 min read

Why ad longevity beats ad spend as a performance signal — and why we publish one but not the other

Last updated: May 2026

Fast answer

An ad's longevity (datefirstseen to datelastseen) is exposed in the public Meta Ad Library. An ad's spend is not. Meta keeps spend opaque deliberately. Longevity is a higher-fidelity proxy for performance anyway: Meta's optimisation engine only allows ads above a CTR/conversion threshold to keep running, so a 30+ day longevity tells you the platform has decided the ad converts. Read 4 longevity bands: under 7 days (testing), 7 to 29 (mid-tier), 30 to 89 (high-performer), 90+ (top-decile evergreen). The bands are useful; specific "spend estimates" from tools are mostly modelled noise.

Why we picked longevity, not spend

When we designed CommonWealth Ops, we had to decide what intelligence to publish weekly. The obvious candidate was advertiser spend — "here are the top spenders in your niche this week, here is what they are spending." Every paid competitor publishes some version of this.

We tested the underlying signal. Two problems became unavoidable:

Spend is modelled, not measured. No tool reads spend directly from Meta or TikTok — the public surfaces hide it. Tools estimate it by triangulating impression counts and assumed CPMs. The error bars on this triangulation are large (we measured 30 to 50 percent error on high-volume ads in our own back-tests, worse for low-volume). Publishing a number with that error and presenting it as authoritative would be dishonest by our reading.

Spend is the wrong question anyway. "How much is this advertiser spending" is downstream of the real question, which is "is this ad actually working." A small operator running a winning ad on a €100/day budget is more useful intelligence than a big operator running a losing ad on a €5,000/day budget. The spend number obscures the performance signal it is meant to proxy.

Longevity is different. It is a measured field (Meta exposes the first-seen and last-seen dates directly). It is also a higher-fidelity proxy for what we actually want to know. So we built the weekly digest around it.

The four longevity bands

Meta's optimisation engine continuously evaluates every running ad. Ads above a CTR-and-conversion threshold (specific to the objective, audience, and creative format) keep delivering. Ads below the threshold get throttled and eventually paused. The advertiser can manually intervene (push budget, restart paused ads) but the default behavior of the system is performance-gated.

So an ad's longevity is literally the platform's judgment of its performance — censored at the moment you read it (the ad may continue or may have just paused), but with the platform's full optimisation logic baked in.

The bands we use:

Band 1 — under 7 days active

The advertiser is TESTING. Either the ad has just launched and the platform has not yet decided its fate, or the ad failed early and was paused. From a pattern-reading perspective, this band is mostly NOISE — it tells you what the advertiser THOUGHT might work, not what actually works.

Operator action: ignore band 1 ads unless you are tracking a specific competitor's test cadence (which is useful as a behavioural signal — a competitor running 12 band-1 ads in a week is gearing up for a campaign).

Band 2 — 7 to 29 days active

MID-TIER. The ad cleared the initial 7-day threshold but has not yet established as a sustained performer. This band has the highest noise-to-signal ratio because it includes both "ads that will become winners" and "ads that are about to pause." For the weekly read, treat band 2 as DIRECTIONAL — the patterns visible here are emerging.

Operator action: watch for archetypes that appear in band 2 across multiple advertisers. If 3+ advertisers have band-2 ads using the same hook archetype, you are likely seeing the leading edge of next week's dominant pattern.

Band 3 — 30 to 89 days active

HIGH PERFORMER. The ad has cleared the most-restrictive performance gate and is in active sustained delivery. The platform has made enough conversion measurements to determine this ad is in the top 20 to 30 percent of its peer group. From the operator's perspective, band 3 is the GOLD signal.

Operator action: catalog every band-3 ad you see in your niche weekly. The hook structures, offer framings, and CTAs in band 3 represent the platform-validated winners. Copy the STRUCTURE (not the words) when designing your next test.

Band 4 — 90+ days active

EVERGREEN TOP-DECILE. Only a small fraction of ads survive the 90-day mark. The advertisers running these ads have a winning creative formula AND a sufficiently broad audience to sustain delivery at scale for 3+ months. These are the ads worth studying in depth — they represent the most-validated creative patterns in the niche.

Operator action: read every band-4 ad in your niche slowly. If a band-4 ad exists for your specific subcategory, your CTR baseline should be set against the level needed to compete with that ad.

Calibrating the bands per niche

The bands are universal in DEFINITION but not in IMPORTANCE. Fast-turnover niches (apparel, certain DTC categories) have shorter median longevities — a 45 day ad in fast-turnover apparel is high-decile. Slow-turnover niches (supplements, courses, financial products) have longer median longevities — a 45 day ad in supplements is middle-tier.

The reading rule: read the longevity band of an ad RELATIVE to the niche median, not absolutely. Our internal calibration for the two niches we currently cover most heavily:

Fitness: band 3 (30 to 89 days) starts being meaningful around 45 days due to fast hook saturation in the category. Band 4 starts around 75 days.
Skincare: band 3 is meaningful from the standard 30 days because skincare hooks have longer half-lives.

For niches we do not yet cover, you can roughly estimate by reading the top 20 ads in the Meta library for the niche and computing the median active period. Set your band-3 threshold at roughly 2x the median.

What we do NOT publish, and why

We do not publish:

Spend estimates — modelled, not measured, with wide error bars. See above.
Specific CTR or conversion rate — not exposed on any public surface. Anything published as CTR or CVR from public data is fabricated.
A/B test variant relationships — not exposed. We can see all variants but not which ones are grouped by the advertiser as a test.
Predictive forecasts — we are descriptive intelligence. Predicting next week's winner is structurally beyond what the data supports. Saying "this hook archetype will keep dominating" requires assumptions the data does not justify.

What we DO publish, per niche per week:

Dominant hook archetype (rhetorical question, imperative, identity, social proof, pattern interrupt) — see the hook archetype framework.
Dominant emotional trigger (aspiration, pain relief, curiosity, FOMO, nostalgia).
Dominant CTA pattern (comment-keyword, link-in-bio, shop-now, learn-more).
Longevity band distribution for the top 20 advertisers in the niche.
Pattern half-life estimate (how many weeks the dominant archetype has held).

What this method does NOT do

It does not replace your own ad testing. The longevity-band read identifies validated patterns; your own €5/day hook tests determine which validated pattern works for YOUR product.

It does not give you a competitor's strategic plan. Reading their longevity bands tells you what creatives are working for them; it does not tell you their full marketing mix (email, organic, partnerships, retail), pricing strategy, or product roadmap. Those exist in places the public ad library cannot see.

It does not work for short observation windows. The longevity bands are a 30 to 90 day frame. If you only read your niche once and never again, the bands tell you a snapshot; the COMPOUNDING value is in reading the same niche weekly so you watch ads transition between bands.

How CW Ops uses the bands operationally

Our weekly digest for each covered niche reports the longevity band distribution across the top 20 advertisers, identifies the bands where the dominant hook archetype is most concentrated, and flags newly-entering band-3 or band-4 ads. The result is a 5-minute read that tells you which patterns crossed the performance gate this week.

Pricing details: EUR 49/month plus 20% of net profit (a single rate for every operator, no threshold, and EUR 0 share in any month you do not profit). If the longevity-based intelligence does not surface a pattern you would not have found yourself within 30 days, the first month refunds and the account freezes.

Frequently asked questions

Why does Meta hide ad spend?: Two reasons: competitive sensitivity (Meta does not want advertisers to know each other's budgets, because it would dampen bidding) and creative-experimentation privacy (advertisers running A/B tests do not want each variant's budget exposed). The Digital Services Act required Meta to expose CREATIVE under EU jurisdictions; it did not require Meta to expose BUDGET. That asymmetry is deliberate. Tools that publish 'spend estimates' are not reading them from Meta; they are modelling them from impression-range data, which has wide confidence intervals.
How accurate are paid tools' spend estimates?: Honestly assessed, the published-spend-range estimates from tools like Pipiads or AdSpy are within roughly 30 to 50 percent of true spend for high-volume ads, and substantially worse for low-volume ads. The error compounds because the estimates are based on impression triangulation, which is itself a triangulation from publicly-visible engagement counts. The number that looks like '€8,000 to €12,000 monthly' might be €2,000 or €20,000 in reality. The directional signal is useful; the specific number is not.
What does ad longevity tell you that spend does not?: Longevity tells you the platform's optimisation engine has decided the ad converts. That decision is made by Meta's algorithm using full conversion data (which you cannot see) and full audience data (which you cannot see). When you read 'this ad ran 60 days,' you are reading the output of a much more sophisticated model than any third-party spend estimator can produce. The longevity is censored at the moment you read it (the ad might run another 60 days or might pause tomorrow) but the signal is still higher-fidelity than modelled spend.
How does the bands framework apply differently per niche?: Niches with fast turnover (fashion, certain DTC categories) have shorter median longevities — a 45 day ad in fast-turnover apparel is high-decile. Niches with slow turnover (supplements, B2B, financial products) have longer median longevities — a 45 day ad in supplements is middle-tier. Read the band relative to the niche median, not absolutely. Our weekly digest does this band calibration per niche so you do not have to.
What other signals does CW Ops publish since not spend?: Per niche per week: dominant hook archetype, dominant emotional trigger, dominant CTA pattern, longevity band distribution, and the pattern half-life estimate. Each is a directly readable, audit-able signal from the public surfaces. None of them require modelling assumptions like spend estimates do. The result is a digest that says 'this is what is working this week, why, and how confident we are' instead of 'this advertiser spent €N.'

Become an operator

Stop guessing what to sell.

CommonWealth Ops turns your market's competitor activity into ranked, data-backed intelligence — and protects your capital before you spend a euro on ads. EUR 49/mo + 20% of net profit. No free trial: skin in the game both ways.

Join the waitlist

See pricing Real and aspirational stories

← All posts