Facebook Ads Metrics That Actually Move Money (And the Ones I've Stopped Watching)
After about $1.2M in tested spend across SaaS, fintech, and e-commerce campaigns over the last two years, here's what I actually look at — and what I've learned to ignore.
Ads Manager will throw 50+ columns at you the moment you open it. Most of them are noise dressed up as signal. I spent my first year as a media buyer staring at CTR and Relevance Score, "optimizing" things that didn't move revenue an inch. Then I almost killed a campaign that ended up being our best performer of the quarter, because its CTR was 0.6% and I panicked.
This is the framework I wish someone had handed me back then. It's the one I use now to make scale/kill decisions in under 60 seconds.
The hierarchy: read top-down, not left-to-right
Every metric you'll see fits into one of three buckets, and they're not equally important:
- Money metrics — ROAS, CPA, revenue. These either work or they don't. If ROAS is hitting target, nothing upstream matters.
- Quality metrics — CTR, hook rate, video hold rate, CPM. These tell you why the money metrics look the way they do.
- Volume metrics — impressions, reach, frequency. Useful for catching fatigue, useless for judging performance.
The discipline: when reviewing accounts, I always start at money and only drill down if money's broken. If your $30 CPA campaign is profitable, I genuinely don't care that the CTR is 0.6%. The market clicks differently than your aesthetic preferences.
Every junior media buyer I've trained does the opposite — opens Ads Manager, sees a low CTR, and starts "optimizing" creative on a profitable ad set. That's how you accidentally kill winners.
Core metrics, what they mean (and what they don't)
ROAS — the only one that pays your bills
Revenue divided by ad spend. Spent $1,000, generated $3,200 — ROAS is 3.2x.
The catch: ROAS attribution depends entirely on which window you're using. Facebook's default 7-day-click + 1-day-view will systematically over-count compared to ground-truth. I run a private rule: take Facebook's reported ROAS, multiply by 0.7, and that's roughly what shows up in the actual revenue dashboard.
For one e-commerce client running $8k/day, Facebook reported 4.2x ROAS through last month. Shopify said 3.0x. We'd been making decisions on the fake number for three weeks. Now we run weekly reconciliation in a separate sheet.
CPA — what you're actually willing to pay
Cost per action — purchase, lead, install, trial start. The number that defines whether the unit economics work.
Most accounts I audit have a "target CPA" pulled out of thin air. Real targets come from LTV math: if a SaaS user is worth $180 over 18 months and you want a 3:1 LTV:CAC, your blended CPA target is $60. Not $30 because $30 "feels safe."
Cold tier CPA will run higher than your blended target. That's fine. Retargeting and brand search make up the difference. If you optimize every channel to the same CPA target, you'll starve cold acquisition and die in 6 months.
CTR — a diagnostic, not a goal
Percentage of impressions that turned into clicks. Useful for spotting creative issues, but only when paired with downstream data.
Realistic ranges by funnel position:
- Cold prospecting: 0.8% – 1.5%
- Warm/lookalike: 1.2% – 2.5%
- Retargeting: 1.5% – 4%
Above 3% on cold? Probably clickbait. Below 0.6% on retargeting? Creative is dead — refresh.
One trap: CTR (Link) vs CTR (All). The "All" version counts likes, shares, profile clicks. I've seen agencies report "2.1% CTR" that was actually 0.4% to the link. Make sure your dashboard uses CTR (Link Click-Through Rate).
CPM — what you can't directly control
Cost per 1,000 impressions. Driven by auction competition, audience overlap, ad quality, and placement. CPMs in 2026:
- US Tier-1: $18 – $35 cold, can spike to $60+ in Q4
- UK/AU/CA: $14 – $28
- EU mixed: $8 – $20
- LATAM: $3 – $8
- India/SEA: $1.50 – $5
If your CPM jumps 40% week-over-week with no other change, three things are usually happening: audience saturation (frequency creeping above 2.5), creative fatigue (Facebook lowers your auction priority), or auction competition (Black Friday, election cycles, major launches).
The "is this winning?" decision framework
I run this 4-question check on every ad set before I touch it:
- Is CPA at or below target? If yes, leave it alone unless ROAS contradicts.
- Is ROAS above 1x at minimum (and meeting goal)? If no, dig into AOV and conversion rate before blaming the ad.
- Is frequency under 2.5 in the last 7 days? If above, audience is saturating — refresh creative or expand audience.
- Is the trend stable or improving over 3 days? If declining sharply, action. If just noisy, hold.
This sounds simplistic. It is. But the number of times I've watched someone kill a profitable ad because "CTR dropped from 1.4% to 1.1% yesterday" — and then the next day it bounced back to 1.5% — would horrify you.
Metrics that look smart but lie to you
"Engagement Rate"
Useful for organic. Useless for paid. A high engagement rate just means you got cheap engaged audiences. Doesn't translate to purchases.
Quality Ranking / Engagement Ranking / Conversion Rate Ranking
Facebook's three diagnostic rankings (Above/Below average). I check them only when an ad set's CPM is unusually high. Otherwise: noise. I've had ads with "Below Average" Quality Ranking print money for 6 months straight.
Relevance Score (the old one)
Deprecated, but I still see consultants reference it in 2026 reports. If your media buyer mentions Relevance Score unironically, get a new one.
Reach
Vanity. "We reached 4 million people." OK, did they convert? Reach inflates with bad audiences. A campaign hitting 200k engaged retargeting users will outperform one reaching 2M cold.
Industry benchmarks (2026, US Tier-1)
Take these as starting points, not gospel. Account-level variance is huge — the right benchmark is your own historical numbers from 90 days ago, not someone's blog.
| Vertical | CPM | CTR (Link) | CPC | Conv. Rate |
|---|---|---|---|---|
| B2B SaaS (trial) | $22 – $40 | 0.8% – 1.4% | $2 – $4.50 | 3% – 7% |
| E-commerce (purchase) | $14 – $28 | 1.0% – 2.2% | $0.80 – $2 | 1.5% – 3.5% |
| Lead-gen (insurance/finance) | $18 – $32 | 0.9% – 1.6% | $1.80 – $3.50 | 5% – 12% |
| App install (utility) | $6 – $14 | 1.4% – 3% | $0.40 – $1.20 | 8% – 18% |
| Education (course/book) | $12 – $24 | 1.2% – 2.4% | $0.60 – $1.80 | 2% – 6% |
If you're 2x off these ranges in either direction, something's specific to your account — either better targeting, broken pixel, wrong bid setup, or you're in a niche where the benchmarks don't apply.
The reporting setup I actually use
Three views. Nothing more.
Daily glance (5 minutes). Spend, CPA, ROAS at the campaign level. Just the totals. If something shifted >25% from yesterday, I drill in. Otherwise close the tab.
Weekly review (45 minutes). Ad-set level. CPA trend, CTR trend, frequency, audience saturation. Decide what to scale, what to refresh, what to kill. Document the call in a sheet so I can audit my own decisions a month later.
Monthly reconciliation (2 hours). Compare Facebook attribution to actual revenue source-of-truth (Shopify, Stripe, CRM). Calculate true blended ROAS. Adjust target CPAs for the next month. This is the one most teams skip — and the one that catches attribution drift before it costs you a quarter.
The attribution problem nobody fixes
Here's the part that's gotten worse since iOS 14, not better.
Facebook reports conversions based on its own model — view-through, modeled, click-attributed. Your source of truth (Shopify orders, Stripe charges, CRM-confirmed deals) reports actual money in the bank. These two numbers will never match.
What I see in 80% of accounts: Facebook over-reports by 20-40%. Sometimes more. The gap grows with broad audiences and shrinks with retargeting (because those users have a stronger pixel signal).
The fix isn't to "trust Facebook" or "ignore Facebook." It's to maintain two views in parallel. Use Facebook attribution for in-platform optimization (Facebook is making the bidding decisions, so its data is what it acts on). Use source-of-truth for budget allocation, ROAS targets, and "is this account actually profitable" calls.
If you're running >$10k/month and you don't reconcile weekly, you're flying blind. Build the sheet, even if it's ugly.
FAQ
What's a "good" CTR for Facebook ads?
Depends entirely on funnel position. 0.6% on a profitable cold campaign is fine. 1.5% on retargeting that's losing money is not fine. CTR is a diagnostic — read it next to CPA, never alone.
How long do I wait before judging a new ad?
I want at least 50 conversions or $1,000 spent (whichever comes first) before making a kill/scale decision. Anything less is noise. For high-AOV / low-volume offers, double both thresholds.
Should I optimize for purchases or add-to-cart?
Purchases, always, if your account has the volume (50+ purchase events/week). Below that, optimize for the closest upper-funnel event with sufficient volume — usually add-to-cart or initiate checkout. Optimization for events with fewer than 50 weekly conversions makes Facebook's bidder essentially random.
Why does my CPM keep climbing?
Three usual culprits: audience saturation (frequency >2.5 in last 7 days), creative fatigue (CTR declining for 5+ days), or auction competition (Q4, election year, major industry launches). Diagnose by checking frequency first — it's the cheapest fix.
Is Quality Ranking worth optimizing for?
Only when your CPM is unusually high. Otherwise: noise. I've had "Below Average" ads run profitably for months. Don't optimize for Facebook's diagnostic; optimize for the money.
Bottom line
Five numbers matter: ROAS (real, not Facebook-reported), CPA, CTR, CPM, frequency. Everything else is a diagnostic for those five. The discipline is reading top-down — start at money, drill into quality only when money's broken — and reconciling Facebook's reality with your bank account at least weekly.
Most account problems aren't metric problems. They're attention problems. Watch fewer numbers more carefully, and you'll outperform 90% of agencies that are tracking 30 things and acting on none of them.