Facebook ads reporting: what to track, what to cut, and the reports that actually drive decisions

Most Facebook ads reporting dashboards are built to answer one question: how did we do? That's the wrong question. The useful question is: what do we change next, and why?

The difference sounds small. It isn't. When you optimize your reporting for accountability instead of decisions, you end up drowning in CTR trends, screenshot exports, and weekly decks that no one acts on. The account keeps running. Spend keeps flowing. Nothing changes. That's not a data problem — it's a reporting architecture problem.

This guide is a decision-first facebook ads reporting playbook. It covers the metrics pyramid that separates signal from noise, the diagnostic breakdowns that actually explain performance shifts, a practical weekly/biweekly/monthly cadence, the four reports every media buyer needs, and the post-iOS 14 attribution reality that makes most native Meta reports quietly wrong. Includes a worked example: an €8k/month DTC brand debugging a ROAS drop via breakdowns.

TL;DR: Facebook ads reporting only drives decisions when it's organized around a four-tier metrics pyramid (spend velocity → efficiency → quality → compounding). Native Meta reports are reliable for relative creative comparisons but systematically undercount iOS conversions post-ATT. The fix is blending Meta data with MER and cohort ROAS on a structured weekly/monthly cadence — and knowing exactly when to distrust what the dashboard shows.

Step 0: set up adlibrary before you open Ads Manager

Before pulling a single report, most experienced buyers do one thing first: benchmark the competitive context. You cannot interpret a €12 CPM without knowing whether the category average is €9 or €18. You cannot judge whether a hook angle is exhausted without knowing how long competitors have been running the same creative pattern.

adlibrary's unified ad search gives you that context. You can filter by advertiser, date range, placement, and creative format to see which ads in your category are still in-market — meaning they're likely profitable enough to keep spending behind. Before you diagnose your own numbers, spend five minutes checking which angles competitors have been running longest. That's your baseline.

For creative strategists in particular, this context separates a reporting session from a genuine diagnostic. A hook that performed for six weeks before dying is a different problem than one that never got traction. The ad timeline analysis feature shows exactly how long specific creatives have been active — a direct signal of what's working in your category at a structural level, beyond a seasonal one.

This is the manual step. Claude Code via the adlibrary API can automate the competitive pull as part of a weekly reporting workflow — more on that in the worked example section.

The metrics pyramid: four tiers, one decision at a time

Not all metrics deserve the same attention at the same time. The pyramid forces a sequence: you only move up when the layer below is stable.

Tier 1 — Spend velocity (base)

Is the account spending at the intended rate? Is pacing on track for the week/month? Are any ad sets hitting budget caps prematurely?

This sounds trivial. It isn't. An ad set spending 40% of its daily budget by 8am is running in cheap inventory — usually a sign of poor audience quality or over-narrow targeting. Use the ad budget planner to model pacing curves before campaigns go live so you have a baseline to compare against.

Key metrics: daily spend, budget utilization %, delivery status, impression share by time of day.

Tier 2 — Efficiency

Once spend velocity is confirmed normal, look at efficiency: how much are you paying for each unit of action?

CPM is your auction health signal. Rising CPM with flat CTR means the audience is getting more competitive. Rising CPM with falling CTR is the ad fatigue pattern — creative is losing relevance faster than audience pool is growing.

CPA sits at the top of this tier. It tells you the cost per desired action. But CPA alone without context is noise — a €40 CPA is great for a €200 AOV product and catastrophic for a €45 one. Before you interpret CPA, you need to know your break-even threshold. The break-even ROAS calculator makes this explicit.

Key metrics: CPM, CPC, CTR (link click-through rate, not outbound), CPA, frequency.

Tier 3 — Quality

Efficiency tells you the cost of an action. Quality tells you the value of that action.

ROAS sits here — but as a directional signal, not a ground truth, for reasons covered in the attribution section below. More reliable is MER (Marketing Efficiency Ratio): total revenue ÷ total ad spend across all channels. MER is immune to attribution window games because it uses actual bank-account revenue.

For a deeper treatment of MER as a north-star, the MER budget framework post walks through the full methodology for e-commerce brands.

Key metrics: ROAS (7-day click), MER, conversion rate, AOV, revenue per click.

Tier 4 — Compounding (peak)

The metrics most buyers never look at — and the ones with the most use on long-term margin.

Cohort ROAS: what is the cumulative revenue from a group of customers acquired in a specific week, measured at 30, 60, and 90 days? If your 30-day cohort ROAS is 1.8x and your 90-day cohort ROAS is 3.4x, your MER-based budget decisions should reflect that future revenue, beyond the immediate purchase. This is especially important for view-through conversions — customers who saw the ad but didn't click immediately often show stronger repeat-purchase patterns.

Creative fatigue rate: how quickly does a creative's CPA increase week-over-week after launch? Fast-fatiguing creatives signal a shallow audience match. Slow-fatiguing creatives signal a durable angle worth scaling.

LTV:CAC ratio: the ultimate compounding metric. A 3:1 LTV:CAC with a 6-month payback period has completely different scaling logic than a 3:1 LTV:CAC with an 18-month payback.

Key metrics: 30/60/90-day cohort ROAS, creative fatigue rate, LTV:CAC, repeat purchase rate by acquisition cohort.

Breakdowns that actually diagnose performance shifts

When a metric moves, breakdowns are how you find the cause. The wrong breakdown wastes 20 minutes and produces a shrug. The right breakdown produces a hypothesis in 90 seconds.

The breakdown sequence for a ROAS drop:

1. By placement — Is the drop isolated to Reels, Stories, or Feed? Platform-specific drops often signal a creative format mismatch rather than an audience problem. A Facebook ads dashboard that surfaces placement data inline saves the manual export.

2. By age/gender — Has a specific demographic segment suddenly gotten more expensive? This often indicates a competitor has entered that segment, raising auction prices. Cross-reference with campaign benchmarking data to confirm.

3. By device — iOS vs. Android splits reveal attribution distortion as much as actual performance differences. A ROAS gap between iOS and Android users post-iOS 14 is almost certainly a measurement artifact, not a real behavioral one.

4. By creative — Which specific ad is driving the decline? One exhausted creative pulling down an otherwise healthy ad set is the most common cause of apparent account-wide ROAS drops. This is where ad fatigue shows up concretely.

5. By time of day / day of week — Weekend vs. weekday conversion rates differ significantly for B2B audiences. For DTC, Friday evening through Sunday often shows inflated CPMs with compressed conversion windows.

The sequence matters. Start broad (placement), then narrow (demographic), then specific (creative). Jumping straight to creative analysis before confirming placement is clean wastes time.

For automated diagnostic surfacing, automated ad performance insights covers how to build a pipeline that flags breakdown anomalies without manual exports.

Reporting cadence: weekly, biweekly, monthly

Most buyers check their account daily. Most daily checks produce zero decisions. Here's the cadence that maps review frequency to decision types:

The weekly review should not touch campaign structure. Structural changes reset the learning phase and add noise to any analysis you're trying to run in parallel. If something needs structural surgery, flag it in the weekly and execute in the biweekly window.

Monthly is the only cadence where you should be making budget reallocation decisions across channels. Anything more frequent introduces measurement noise that's larger than the signal you're reacting to.

The 4 reports every media buyer needs

Forget the 47-metric dashboards. Four reports cover 90% of actionable decisions.

Report 1: Creative performance leaderboard

Columns: creative name, spend, impressions, CPM, CTR (link), CPA, ROAS (7-day click), frequency, days active, fatigue score (CPA week-over-week change %).

Sort by spend descending. Flag anything where frequency >3.5 with rising CPA. This is your weekly kill/scale list.

Report 2: Audience efficiency map

Columns: audience/ad set name, reach, frequency, CPM, CPA, ROAS, overlap flag.

This report exists to catch audience cannibalization — when two ad sets are bidding against each other in the same auction, inflating CPMs for both. The media mix modeler can help model the reallocation scenario before you execute.

Report 3: Cohort ROAS tracker

Columns: acquisition week, customers acquired, 7-day revenue, 30-day revenue, 60-day revenue, 90-day revenue, 30-day cohort ROAS, 90-day cohort ROAS, LTV:CAC projection.

This report is built from your own first-party data (Shopify, WooCommerce, or your CRM), not Meta's dashboard. It's the only way to know whether a campaign that looked weak at 7 days was actually building a high-LTV customer base.

Report 4: Attribution reconciliation

Columns: date, Meta reported revenue, GA4 reported revenue, Shopify actual revenue, MER, Meta/Shopify gap %, iOS conversion estimate.

The gap between Meta's reported revenue and Shopify actuals is your attribution distortion metric. A consistent 25–35% gap is normal post-iOS 14. A sudden spike to 60%+ is a signal that something changed — a pixel event misfired, a new iOS version reduced consent rates, or an attribution window shifted.

How platforms compare on native reporting

adlibrary sits in a different category from attribution tools — it's the competitive data layer that tells you what your category looks like, not a conversion tracker. The AI ad enrichment feature enriches creative metadata with format, angle, and messaging signals so you can spot patterns across hundreds of competitor ads without manual tagging. That context is what makes your cohort and creative reports interpretable, beyond numeric.

Post-iOS 14 attribution reality

Apple's App Tracking Transparency framework, introduced with iOS 14.5 in April 2021, broke the pixel-based tracking model that Meta's reporting was built on. The practical effect: roughly 60–70% of iOS users opt out of cross-app tracking in most Western markets, according to Flurry analytics data.

Meta responded with Aggregated Event Measurement (AEM), which limits advertisers to eight conversion events per domain and uses statistical modeling to estimate conversions that can't be directly attributed. The Facebook developer documentation on AEM describes how this modeling works, but doesn't quantify the error rate — which in practice varies by audience, campaign type, and iOS version.

What this means for your reporting:

• 7-day click window ROAS is a floor, not a ceiling. The modeled conversions Meta adds back are better than nothing, but they're estimates. Your true conversion volume is likely 15–40% higher than what Meta reports for iOS-heavy audiences.

• Android vs. iOS ROAS parity is a measurement artifact. If your iOS ROAS is running at 1.6x and Android at 2.8x, that gap is real but exaggerated by attribution loss. Don't kill iOS-targeted campaigns on the basis of this split alone.

• View-through attribution is structurally undervalued. Post-iOS 14, many view-through conversions vanish from Meta's count entirely. For upper-funnel campaigns, this means Meta reports that look like failures are sometimes building purchase intent that shows up in MER but not in ROAS.

For a deeper treatment of how iOS attribution errors propagate through reporting, the Meta ads attribution error post covers specific diagnostic steps. The broader death of attribution post puts this in the context of marketing measurement's structural shift toward probabilistic models.

The IAB's guidance on measurement standards and Nielsen's annual marketing measurement report both confirm that the industry has moved from deterministic to blended measurement — a shift that Meta's native UI hasn't fully caught up to.

When Meta's native reports lie and how to catch it

There are five specific scenarios where Meta's dashboard numbers are systematically wrong. Not wrong by a rounding error — wrong in ways that produce bad decisions if you trust them.

1. The pixel double-fire

Symptomatic: reported conversion volume significantly exceeds Shopify order volume. Cause: the pixel fires twice on the order confirmation page (common after template changes or third-party app installs). Catch it: compare Meta reported purchases to Shopify actual orders on the same day. A ratio above 1.2:1 is a flag.

2. The attribution window mismatch

Symptomatic: a campaign that looks profitable in Ads Manager is losing money per your P&L. Cause: 7-day click attribution is pulling revenue from customers who would have purchased anyway (high-intent branded search), or from a purchase driven by a different channel. Catch it: run the same date range in GA4 with last-click attribution and compare to Meta's claimed revenue.

3. The audience overlap bleed

Symptomatic: two ad sets show strong individual ROAS but combined account ROAS is flat. Cause: both ad sets are winning impressions from the same users, counting the same conversions in both sets' reports. Catch it: run the Facebook Audience Overlap tool. If overlap exceeds 20%, you're likely double-counting. The facebook advertising insights dashboard post covers how to structure the account to minimize this.

4. The modeled conversion inflation

Symptomatic: Meta reports strong ROAS for an iOS-targeted campaign, but Shopify revenue didn't grow proportionally. Cause: Meta's AEM modeling over-attributed conversions to the campaign — a known issue when modeling parameters don't match your actual customer behavior. Catch it: cohort analysis. If the customers Meta claims to have acquired don't appear in your Shopify customer list in the same volume, the modeled count is inflated.

5. The delayed attribution spike

Symptomatic: a campaign's ROAS jumps two days after you paused it. Cause: Meta's 7-day click window is still attributing conversions to the paused campaign as users convert days after their last click. This is attribution working as intended, but it looks like a reporting anomaly. Catch it: note the pause date and expect a 5–7 day tail of attributed conversions.

For a systematic approach to catching these patterns with decision intelligence tooling, meta advertising decision intelligence covers the infrastructure layer. For teams running the full marketing agency tool stack, integrating a pixel audit step into the monthly reporting cadence catches most of these before they compound.

Worked example: debugging an €8k/month DTC ROAS drop

This is a real scenario, anonymized. A DTC apparel brand spending €8,000/month on Meta noticed their reported 7-day ROAS drop from 3.2x to 2.1x over three weeks. Their instinct was creative fatigue. The actual cause was different.

Step 1: Spend velocity check

Budget pacing was normal. Daily spend matched targets. No budget caps being hit early. Tier 1 was clean — the drop wasn't a delivery problem.

Step 2: Efficiency breakdown by placement

Breaking down by placement showed Feed ROAS at 2.8x and Reels ROAS at 0.9x. Three weeks earlier, Reels had been at 2.4x. Reels now represented 38% of total spend (up from 18%), driven by Meta's automatic placement optimization shifting budget toward Reels inventory.

Step 3: Identify the mechanism

The Reels creatives were all repurposed Feed videos — 1:1 or 4:5 aspect ratio content reformatted to 9:16. The format mismatch was driving a high swipe-away rate, depressing CTR, which was training the algorithm to deliver to lower-quality audiences, which compressed ROAS further. A classic marketing funnel problem: the mid-funnel was leaking through a format-audience fit failure, not creative message exhaustion.

Step 4: Confirm with creative leaderboard

The creative performance report confirmed it. The two native 9:16 Reels creatives (shot vertically) had CPA of €34 and €29. The repurposed Feed videos running in Reels had CPA of €71–€89. The audience was the same. The creative format was the entire delta.

Step 5: The fix

Three actions: (1) exclude Reels from the reformatted Feed video ad sets, (2) increase budget on the two native Reels creatives, (3) brief three new native vertical creatives for the next two-week cycle.

Two weeks later, blended ROAS recovered to 2.9x. Not quite the 3.2x peak, but competitive CPMs in the category had risen approximately 12% in the same period — confirmed by checking competitor ad activity via adlibrary's campaign benchmarking workflow.

What would have gone wrong with a different diagnosis:

If the team had paused their highest-spend Feed creatives on the assumption of creative fatigue, they would have killed the accounts' best-performing inventory and lost another week of data. The breakdown sequence — placement first, then creative — prevented a €6,000–8,000 misallocation.

The diagnostic process, automated, looks like this:

# Weekly reporting diagnostic prompt (for Claude Code / GPT-4o + Ads Manager API)
Given the following Ads Manager breakdown data [paste CSV]:

1. Flag any placement where ROAS is >30% below the account average
2. For flagged placements, list all active creative IDs and their aspect ratios
3. Calculate the CPA delta between native-format and repurposed-format creatives
4. Recommend: exclude from placement, pause, or scale for each creative
5. Output a Slack-ready summary with 3 bullet decisions

For teams integrating this into a broader ops stack, Claude Code for marketing ops covers how to wire Ads Manager API data into automated diagnostic pipelines. The automated ad performance insights post shows the output layer.

Where adlibrary fits in the reporting stack

Facebook ads reporting tells you what your account is doing. adlibrary tells you what your category is doing. Both are necessary. Neither replaces the other.

The specific gap adlibrary fills: before a monthly budget reallocation review, you need to know whether a ROAS decline is account-specific or category-wide. If CPMs across your category rose 15% in the same period, your ROAS drop is a market condition, not an execution failure. Those require completely different responses — don't change creative strategy when the auction got more expensive; change bid strategy.

The unified ad search gives you the category view in minutes. Filter by your primary competitors, sort by first seen date, and look at the last 30–90 days. Which ads are still in-market? Which angles have multiple competitors running simultaneously (usually a sign that angle is proving out commercially)? Which placements are your competitors concentrating on?

For media buyer daily workflow integrations, adlibrary functions as the pre-reporting context layer — the five-minute competitive scan that makes the subsequent 55 minutes of Ads Manager work interpretable.

For agencies managing multiple clients, media buying software comparison covers how adlibrary fits alongside attribution tools, creative testing platforms, and budget management software in a complete stack.

The ROAS calculator and Facebook ads cost calculator are the quick-reference tools for in-meeting number-checking — useful when you need to translate a ROAS figure into a revenue projection on the fly without exporting.

Frequently Asked Questions

What metrics should I track in Facebook ads reporting?

Track metrics in four tiers: spend velocity (daily spend, pacing) at the base; efficiency (CPA, CPM, CTR) above that; quality (ROAS, MER, conversion rate) in tier three; and compounding signals (cohort ROAS, LTV:CAC, creative fatigue rate) at the top. Most buyers over-report on efficiency and under-report on compounding signals, which is where real margin hides.

How accurate is Facebook ads reporting after iOS 14?

Facebook's reported conversions after Apple's App Tracking Transparency prompt are modeled estimates, not raw pixel signals. Meta uses Aggregated Event Measurement (AEM) and statistical modeling to fill gaps from users who opt out of tracking. Independent studies estimate 20–40% of iOS conversions go unattributed in native reports. Cross-referencing with first-party data and MER (revenue ÷ total ad spend) is the current standard for ground-truth measurement.

What is the difference between cohort ROAS and last-click ROAS in Facebook ads?

Last-click ROAS assigns all revenue credit to the final click before purchase, which systematically undervalues top-of-funnel and view-through touchpoints. Cohort ROAS groups buyers by their first-touch acquisition date and tracks their cumulative revenue over 30, 60, or 90 days. For subscription and repeat-purchase businesses, cohort ROAS is a better predictor of long-term profitability because it captures reorder revenue that last-click never sees.

How often should I pull Facebook ads reports?

Weekly reviews should cover spend pacing, CPM trends, and top-creative performance. Biweekly reviews are for audience fatigue, frequency caps, and ad set structural changes. Monthly reviews are the right cadence for attribution model audits, cohort ROAS comparisons, and budget reallocation decisions. Daily checking is common but rarely produces actionable data — Meta's delivery algorithm needs 48–72 hours to stabilize after any change.

When should I trust Meta's native reporting vs. third-party attribution?

Meta's native reports are reliable for relative performance comparisons within the platform — comparing creative A vs. B, or audience segment X vs. Y. They are unreliable for absolute revenue attribution, especially for iOS users, because of the ATT framework and the 7-day click / 1-day view default window. For absolute spend-to-revenue decisions, blend Meta data with first-party analytics (GA4 or Shopify) and calculate MER as your north-star metric.

---

Facebook ads reporting is not a scoreboard. It's a diagnostic instrument. Build it to produce decisions, not documentation — and the reports that don't point to a specific action this week are reports you don't need.