Most Facebook advertisers track too many metrics and act on the wrong ones. They optimize for reach when they should be watching hold rate. They kill ads because CTR is low when the real issue is the offer. They scale because ROAS looks good this week, without checking whether that ROAS holds as spend increases.
The metrics that predict scale are specific, and most dashboards don't surface them clearly. Here's what to actually watch.
The problem with vanity metrics
Impressions, reach, and post engagement are easy to report and nearly useless for decisions. High reach tells you Facebook is serving the ad. It tells you nothing about whether anyone cares.
Likes and comments are worse. They measure social friction, not purchase intent. An ad that generates 500 comments saying "tagging my friend!" and converts nobody is a waste of budget with a great engagement report.
Even frequency, which used to be a key metric, has become less useful as Facebook's audience expansion has made it harder to define who you're actually reaching. Track it as a sanity check, not a primary signal.
Hook rate: the first 3 seconds
Hook rate is the percentage of people who watch at least three seconds of your video after it appears in their feed. The formula is 3-second video views divided by impressions.
A good hook rate for Facebook is generally above 25%. Below 20%, the opening isn't stopping people. Above 35%, you've found something that pulls attention effectively.
Hook rate is a creative metric, not an audience metric. If your hook rate is low, the problem is the first frame, the first line of voiceover, or the first text card. Not the targeting. Not the offer. Not the landing page. Fix the hook. The breakdown of what makes hooks work is a useful companion to this metric.
Hook rate is also your fastest feedback signal. You'll see it move within 24 to 48 hours of a new creative going live. Use it to quickly eliminate bad hooks before they burn budget.
Hold rate: the real predictor of ROAS
Hold rate measures how many people who watched the first three seconds watched all the way through (or to the 15-second mark, depending on how you define it). The formula is ThruPlays divided by 3-second video views.
This metric predicts ROAS better than link CTR because it measures intent. Someone who watches 80% of a two-minute video is far more likely to buy than someone who watched three seconds and clicked out of curiosity.
A hold rate above 25% is solid. Above 40% indicates the ad is delivering on the promise of its hook. If hook rate is high but hold rate is low, the hook is working but the body of the ad isn't delivering. That's a different problem than a bad hook.
For image ads, the equivalent is time-on-ad or secondary actions (link clicks, saves, comments) relative to impressions. Images don't have a hold rate per se, but you can approximate the same logic with engagement depth.
CTR link vs CTR all: what the difference reveals
Facebook reports two types of click-through rate, and conflating them is a common mistake.
CTR (all) includes every click: link clicks, likes, comments, shares, video plays, profile views. It measures total engagement with the ad unit. CTR (link click-through rate) measures only the clicks that go to your landing page or destination.
The difference between these two numbers tells you something. If CTR (all) is high but CTR (link) is low, people are engaging with the ad but not clicking through to buy. That usually means the creative is interesting but the offer isn't compelling, or the CTA is buried or weak.
For most purchase-focused campaigns, target a link CTR above 1%. Below 0.5% usually means something fundamental isn't connecting. The full CTR guide covers how to read this in context.
Cost per purchase vs cost per initiated checkout
CPP (cost per purchase) is your bottom-line conversion metric. It's the one that ultimately tells you if an ad is profitable.
But cost per initiated checkout tells you where the funnel is breaking. If your CPP is high but cost per initiated checkout is reasonable, the problem is on the checkout page, not in the ad. If cost per initiated checkout is also high, the problem is upstream: the ad isn't driving purchase intent at all.
Track both. The ratio between them is your checkout completion rate. A healthy checkout completion rate is generally above 60%. Below 40% usually points to a landing page or checkout friction problem, not a creative problem.
ROAS thresholds by context: what "ready to scale" actually means
There's no universal ROAS threshold for scaling. The right number depends on your margins, your customer acquisition cost targets, and whether you're in growth mode or profitability mode.
That said, here are rough reference points for DTC ecommerce brands:
- Below 2x ROAS: almost always unprofitable unless you have exceptional LTV and a high repeat purchase rate.
- 2x to 3x ROAS: breakeven range for most brands with 40-50% gross margins. Not scaling territory unless your contribution margin math says otherwise.
- 3x to 4x ROAS: profitable for most brands. Scaling is worth exploring if the ROAS holds as you increase spend.
- Above 4x ROAS: strong signal. Scale aggressively unless you're hitting frequency caps or seeing rapid creative fatigue.
The critical word in all of this is "holds." ROAS at $500/day spend rarely holds at $5,000/day. As you scale, you move from your best audience to progressively less qualified audiences. ROAS almost always declines with scale. The question is whether it declines to below your breakeven threshold.
How to use these metrics to identify winning creatives before scaling
The sequence matters. Use early signals (hook rate, hold rate) to filter out obvious losers in the first 48 to 72 hours. Then use downstream signals (link CTR, initiated checkout rate, CPP) to identify which of your early survivors are actually converting.
Don't scale anything before it has at least 20 to 30 conversions at your target CPP. Below that, you're acting on noise. If you're unsure how much data you actually need before a result is real, the guide on statistical significance in creative testing breaks this down precisely.
When you find a creative that passes all these filters, it's worth understanding why. Pull similar ads from competitors using Spreshapp's ad browser to see if there's a pattern. What's the hook structure? What's the offer framing? Save the ads to your library and tag them by what's working. Over time, your saved ad library becomes a map of what your market responds to.
Build a creative scorecard
Tracking these metrics ad hoc in Ads Manager gets messy. Build a simple scorecard you update weekly. The columns:
- Creative name and format
- Days live
- Spend
- Hook rate (3-sec views / impressions)
- Hold rate (ThruPlays / 3-sec views)
- Link CTR
- CPP
- ROAS
- Status: testing, scaling, fatiguing, killed
This doesn't need to be sophisticated. A spreadsheet works fine. What matters is that you're tracking the right things consistently, not whatever Ads Manager happens to show you on the default view.
When the metrics do say scale, the next challenge is executing it without collapsing performance. The guide on how to scale Facebook ads without killing performance covers the mechanics.
Scaling decisions should be boring and systematic. If the numbers say scale, scale. If they don't, don't. The advertisers who build durable performance are the ones who let data make these calls, not instinct.