01What Andromeda changed about measuring Meta ads
Meta’s Andromeda update changed how ads find people. Before it, you chose an audience and Meta showed your ads to it. Now Andromeda reads what’s in each ad, the visuals, the words and the format, and uses that to shortlist which ads each person could see. The creative does most of the targeting.
That changes what your numbers mean.
Before
Audience and ad set metrics told you who was working. You tested audiences and judged campaigns.
Now
Ad-level metrics tell you who the ad is reaching and why. You test creative and judge the business.
At the same time, Meta has narrowed what it counts. In 2026 a click-through conversion means a link click, and shares, saves and engaged video views are reported separately. The long view-through windows are gone. Meta’s numbers are tidier than they were, but they still only describe Meta. They don’t tell you whether the business made money.
So we look at Meta ads on four levels, and we never judge one level with another level’s numbers.
Creative metrics are now targeting metrics
A low hook rate doesn’t just mean fewer views. It means Meta has less to learn from about who your ad is for.
Platform ROAS is a steering wheel, not a bank balance
Use it to compare ads and campaigns. Use your store’s revenue and margins to decide if Meta is working.
02The four levels of metrics
Every metric on Meta answers one of four questions. Work down the list: if an ad isn’t stopping people, there’s no point studying its conversion rate.
Discover · Is the ad stopping people?
CPM, hook rate, hold rate and CTR. These grade the creative and tell you whether Meta can find the right people with it.
Nurture · Is it building intent?
Link CTR, cost per click, landing page view rate, add-to-cart rate and frequency. These tell you whether the right people are taking a closer look.
Convert · Is it selling?
Conversion rate, cost per purchase, ROAS and average order value. These tell you whether the ads and the site are closing sales.
Profit · Is it making money?
MER, new-customer acquisition cost, LTV, payback and contribution margin. These tell you whether the business is growing profitably. They’re the ones that decide.
The first three levels live in Ads Manager. The fourth lives in your store, your accounts and a spreadsheet, which is why most brands never look at it. It’s also where every decision about budget should be made.
03The scorecard at a glance
These are the bands we work to for eCommerce brands on new audiences. They’re a sense check, not a pass mark. Categories, prices and placements all move them, so your own account’s last 90 days is always the first benchmark. ROAS and cost per purchase have no outside benchmark at all: the numbers to aim for come from your own margins, and chapter 06 shows the maths. Each metric is explained in the chapters below.
| Metric | Weak | Normal | Strong |
|---|---|---|---|
| Hook rate | Under 25% | 25–40% | 40%+ |
| Hold rate | Under 30% | 30–50% | 50%+ |
| CTR (all) | Under 1.2% | 1.2–2.4% | 2.5%+ |
| Link CTR | Under 0.8% | 0.8–1.5% | 1.5%+ |
| Landing page view rate | Under 70% | 70–85% | 85%+ |
| Add-to-cart rate | Under 6% | 6–12% | 13%+ |
| Conversion rate | Under 1% | 1–2% | 2.5%+ |
| Cost per purchase | Above your break-even CPA | At or below break-even | At or below your target CPA |
| ROAS (Meta) | Below your break-even ROAS | At or above break-even | At or above your target ROAS |
| MER | Under 3x | 3–5x | 5x+ |
| New customers from Meta | Under 70% | 70–85% | 85%+ |
| LTV:CAC (12 months, on margin) | Under 2:1 | About 3:1 | 4:1+ |
| CAC payback | Over 12 months | 3–6 months | On the first order |
| Contribution margin after ads | Under 15% | 20–35% | 35%+ |
The one rule above all of these
A metric is only good if it moves the one below it. A 35% hook rate on an ad that doesn’t sell is a weak ad. A 1.4x ROAS on an ad that brings in new customers who come back is a strong one. Read every number in this guide as a clue, not a verdict.
04Discover: is the ad stopping people?
Discovery metrics grade the creative. After Andromeda they also tell you how much signal Meta is getting: every time the right person stops on your ad, Meta learns more about who to show it to next.
CPM (cost per 1,000 impressions)
What it is. What you pay for 1,000 impressions. Spend ÷ impressions × 1,000.
What it tells you. How much competition there is for the people Meta is showing your ads to. It isn’t a creative score: a low CPM can just mean Meta found cheap, low-intent people.
What good looks like. There’s no good or bad CPM on its own, so watch the trend. For Australian eCommerce brands on new audiences, we usually see somewhere around $12 to $25. Australia generally runs cheaper than the US. Expect it to climb sharply in the weeks before Click Frenzy and Black Friday and stay high until Christmas, then ease off.
Watch out for. CPM rising on one ad while the rest of the account stays flat. That usually means Meta has run out of the people that ad suits.
Hook rate (thumbstop rate)
What it is. The share of people who watched the first three seconds. 3-second video plays ÷ impressions.
What it tells you. Whether the opening of a video stops the scroll. It’s the purest score for the hook, and the main discovery metric for video.
| Hook rate | Weak | Normal | Strong |
|---|---|---|---|
| New audiences | Under 25% | 25–40% | 40%+ |
We treat 25% as the baseline: below it, the opening isn’t doing its job. 40% is what we aim for, and the hooks that get there are the ones worth turning into more ads. Beauty and food tend to run a few points higher than fashion and pets, and Reels usually beats Feed. Give an ad at least 5,000 impressions before you judge it. And remember that a strong hook rate on its own predicts very little about sales. It’s the first gate, not the finish line. Our guide to Meta ad hooks covers how to write better ones.
Hold rate
What it is. The share of people the hook stopped who kept watching. We work it out as ThruPlays ÷ 3-second video plays. A ThruPlay is a watch to the end, or to 15 seconds on longer videos.
What it tells you. Whether the body of the ad delivers what the hook promised.
| Hold rate | Weak | Normal | Strong |
|---|---|---|---|
| ThruPlays ÷ 3s plays | Under 30% | 30–50% | 50%+ |
There are several ways to work out hold rate, and they give very different numbers. Some people use 15-second plays ÷ 3-second plays, where 15–25% is normal. Pick one formula, write it down and only ever compare like with like. A high hook rate with a low hold rate almost always means the hook over-promised.
CTR (all)
What it is. Clicks of any kind ÷ impressions. It includes clicks on the profile, “see more”, likes and comments, not just the link.
What it tells you. Whether the ad made people curious. It’s the closest thing statics have to a hook rate, because statics have no video plays.
What good looks like. Under 1.2% is weak, 1.2–2.4% is normal, 2.5% and over is strong. A high CTR (all) with a low link CTR means people are interested in the ad but not the product. Common with funny or striking creative that doesn’t sell.
Creative hit rate
What it is. The share of the ads you launch that go on to take real spend. We count an ad as a winner when it spends many times more than your typical ad.
What it tells you. How good your creative process is, not just one ad. After Andromeda, Meta pours budget into the few ads that work and starves the rest.
What good looks like. About one in twenty launched ads becoming a winner is normal. One in ten is strong. Fewer than one in fifty means the ideas aren’t different enough from each other. Meta treats near-identical ads as one ad, so ten colour swaps of the same idea give you one chance, not ten. This is why creative volume matters: our Meta ad creative calculator works out how many ads your budget needs each month.
05Nurture: is it building intent?
Nurture metrics tell you whether the people the ad stopped are the right people, and whether they’re moving towards a purchase. Most problems that look like “Meta isn’t working” show up here first.
Link CTR (outbound click-through rate)
What it is. Link clicks ÷ impressions. Only clicks that head to your site.
What it tells you. Whether the ad made people want to see the product. This is the click that matters, and the best single creative metric for statics.
What good looks like. Under 0.8% is weak, 0.8–1.5% is normal and 1.5% or more is strong. Retargeting and offer ads run higher. Don’t chase it too hard: clickbait lifts link CTR and then tanks conversion rate.
Cost per link click (CPC)
What it is. Spend ÷ link clicks. It’s CPM and link CTR combined into one number.
What it tells you. How much it costs to get someone to your site. Useful for spotting a change, rarely worth judging on its own.
What good looks like. For most eCommerce brands in Australia it sits roughly between 50 cents and $1.50. If it’s climbing, check which of the two moved: CPM (competition) or link CTR (the creative).
Landing page view rate
What it is. Landing page views ÷ link clicks. The share of people who clicked and actually waited for your page to load.
What it tells you. Whether your site is fast enough, and whether your tracking is working. It’s the most overlooked number in Ads Manager.
What good looks like. Under 70% is weak, 70–85% is normal and 85% or more is strong. Below 70% you’re paying for clicks that never see the page. Check your mobile page speed first, then check the pixel and Conversions API are firing on every page.
Add-to-cart rate
What it is. Adds to cart ÷ landing page views.
What it tells you. Whether the product page and the offer live up to the ad. If people arrive and don’t add to cart, the ad and the page are telling two different stories.
What good looks like. For products under about $100, under 6% is weak, 6–12% is normal and 13% or more is strong. The pricier the product, the lower it runs: for orders over $200, 3–7% is normal. Track the cost per add to cart over time too. It warns you about a weakening ad days before cost per purchase does.
Checkout rate
What it is. Checkouts started ÷ adds to cart.
What it tells you. Whether something in the cart puts people off: shipping costs, delivery times, a surprise at the total.
What good looks like. If fewer than about four in ten people who add to cart start checkout, look at your cart page before you look at your ads. Showing shipping costs early and offering buy now, pay later both help.
Frequency
What it is. Impressions ÷ reach. How many times the average person saw the ad in a period.
What it tells you. Whether the same people are seeing the same ad too often. On its own it isn’t a fatigue score. Meta itself flags fatigue on cost per result, not frequency.
What good looks like. For broad prospecting, a 7-day frequency around 1.5 to 2.5 is normal. Over about 3.5, check the ad. If cost per purchase has risen 30% or more over a week while frequency climbs, the ad is tiring and needs a fresh version, not a new audience.
06Convert: is it selling?
Conversion metrics are where most brands start, and where they should finish. They tell you whether the ads are producing sales, but not whether those sales are profitable. That’s the next chapter.
Conversion rate
What it is. Purchases ÷ landing page views (or sessions, in your store’s analytics).
What it tells you. Whether the site turns visits into sales. It’s as much a website metric as an ads metric.
What good looks like. For traffic from Meta, under 1% is weak, 1–2% is normal and 2.5% or more is strong. Store-wide rates run higher because they include returning customers and email. Conversion rates from Meta traffic have fallen over the past year as ads reach more people earlier in their decision, so don’t compare this year to two years ago. The cheapest way to make every other number on this page better is to improve this one. Our conversion rate calculator shows what a lift is worth.
Cost per purchase (CPA)
What it is. Spend ÷ purchases.
What it tells you. What each sale cost in ads. The single best number for comparing ads against each other.
What good looks like. There’s no useful outside benchmark for CPA. A $40 sale is a great result for a $150 jacket and a disaster for a $35 candle. The number to aim for comes from your own maths, and there are two of them.
| Number | How to work it out | What it means |
|---|---|---|
| Break-even CPA | Average order value × your margin after product, shipping and fees | The most a sale can cost before you lose money on it |
| Target CPA | Break-even CPA minus the profit you want from each order | The number to run the account to |
A worked example (made-up numbers)
A $90 average order with a 55% margin after product, shipping and fees leaves $49.50. That’s the break-even CPA: pay more than $49.50 for a sale and you lose money on it. Want $15 profit from each order? Your target CPA is $34.50. If customers come back and buy again, you can afford to go closer to break-even on the first order. If they don’t, you can’t.
Our CPA calculator works out both from your own figures.
ROAS (return on ad spend)
What it is. Revenue Meta attributes to your ads ÷ spend on those ads.
What it tells you. Which ads and campaigns Meta thinks are driving sales. It’s Meta’s view, built on Meta’s attribution, and it includes sales from existing customers who might have bought anyway.
What good looks like. This is the most asked question in Meta ads, and the honest answer is that it depends entirely on your business. A 2x ROAS makes good money for a brand with high margins and loses money for one with thin margins. Industry averages tell you nothing about either. So, like CPA, the number to aim for is one you work out.
Work out your break-even and target ROAS
Break-even ROAS = 1 ÷ your margin after product, shipping and fees. At a 55% margin that’s 1.8x; at 30% it’s 3.3x. Below that line, every sale loses money before you’ve paid for anything else. Your target ROAS is the same sum with the profit you want taken out of the margin first: at a 55% margin, keeping 15% for profit means aiming for 1 ÷ 0.40, or 2.5x. Our ROAS calculator and break-even calculator do the maths for you.
Break-even ROAS and break-even CPA are the same line seen from two sides. Hit one and you’ve hit the other. Use whichever your team finds easier to watch.
Average order value (AOV)
What it is. Revenue ÷ orders.
What it tells you. How much each sale is worth. The average Australian online order is a little under $100. Raising AOV lifts your ROAS and lowers your break-even without changing a single ad, which is why bundles, thresholds for free shipping and add-ons are some of the best levers on this page. Try the AOV calculator.
07Profit: is it making money?
This is the level that decides. Platform metrics can all look healthy while the business goes backwards. These numbers come from your store and your accounts, not from Ads Manager, and they’re how we judge whether Meta is working for our clients.
MER (marketing efficiency ratio)
What it is. Total store revenue ÷ total marketing spend, across every channel. Also called blended ROAS.
What it tells you. Whether all your marketing together is paying its way. It can’t be fooled by attribution, because it doesn’t use any: it’s just what came in against what went out.
What good looks like. Most eCommerce brands sit somewhere around 3 to 5x. Under 3x is usually a warning. Subscription brands and brands with strong repeat purchase run higher. Like ROAS, the real benchmark is your own break-even MER, which our MER calculator works out with your fixed costs included.
aMER and first-order ROAS
What it is. Revenue from new customers only ÷ total marketing spend. Sometimes called acquisition MER.
What it tells you. Whether your ads are actually growing the customer base, or mostly collecting sales from people who already buy from you. If MER is steady but aMER is falling, you’re spending more to reach the same customers.
What good looks like. At least your first-order break-even: 1 ÷ your margin on a first order before marketing. Brands that make money on the first order can scale hard. Brands that don’t need strong repeat purchase to cover the gap, and should know exactly how long that takes.
New customer acquisition cost (nCAC)
What it is. Total marketing spend ÷ new customers. Not purchases, new customers.
What it tells you. What it really costs to win a customer. It’s almost always higher than Meta’s cost per purchase, because Meta counts every purchase, including repeat ones.
What good looks like. Lower than what that customer is worth to you (next two sections). Watch the trend month to month: a rising nCAC with a flat CPA means Meta is finding more of your existing customers.
Share of new customers
What it is. The share of Meta’s purchases that come from people who have never bought from you.
What it tells you. Whether your budget is finding new buyers or retargeting old ones. Set up Meta’s audience segments (new, engaged and existing customers) in your ad account settings and you can see this split in every sales campaign.
What good looks like. Under 70% new is weak, 70–85% is normal and 85% or more is strong. If it’s low, your best ads may be quietly selling to people on your email list.
LTV and LTV:CAC
What it is. LTV (lifetime value) is how much a customer is worth over time. LTV:CAC is that value ÷ what it cost to acquire them.
What it tells you. How much you can afford to pay for a new customer, and whether your growth is sustainable.
What good looks like. The common rule of thumb is 3:1. Two things make it useful rather than flattering. First, work LTV out on margin, not revenue: a 3:1 ratio on revenue can be below 1:1 once you take out product costs. Second, use a set window, like 12 months, not a forever guess. On that basis, under 2:1 is weak, about 3:1 is normal and 4:1 or more is strong. Our CAC and LTV calculator works it out.
CAC payback
What it is. How long it takes for a new customer’s margin to pay back what you spent to acquire them.
What it tells you. How much cash you need to grow. A long payback means every new customer ties up money for months.
What good looks like. Paid back on the first order is strong. Three to six months is normal for brands with repeat purchase. Over 12 months is risky, because you’re relying on predictions about customers who haven’t come back yet. Don’t justify losing money on the first order with an LTV you hope for. Use the one you’ve seen. The customer retention calculator helps here.
Contribution margin after ads
What it is. Revenue minus product costs, shipping, payment fees and ad spend, as a share of revenue. It’s what each sale leaves over to pay for the team, the rent and the profit.
What it tells you. Whether your growth is profitable. This is the number we’d put above every other on this page.
What good looks like. Under 15% is a danger zone, 20–35% is healthy and 35% or more is strong. Food and drink run lower, supplements and beauty higher. Our profit margin calculator shows yours.
Is Meta actually causing the sales?
Every number above Meta’s own reports assumes the ads caused the sale. Some of those people would have bought anyway. Meta now offers an incremental attribution setting that optimises for the sales it predicts the ads caused, and its Conversion Lift studies test it properly with a holdout group. For brands spending seriously, we’d run a lift test at least once a year, and test incremental attribution against the standard setting rather than assuming either is better.
08Reading the numbers together
No metric means much on its own. The useful question is always: which number broke first? Start at the top of the list and work down until you find it.
| What you see | What it usually means | What to change |
|---|---|---|
| Low hook rate | The opening doesn’t stop anyone | New first frames or headlines, same body |
| Good hook, low hold | The hook over-promised or the body is slow | Pay off the hook in the next five seconds |
| High CTR (all), low link CTR | People like the ad, not the product | Make the product and the reason to buy clearer |
| Good link CTR, low page views | A slow site or broken tracking | Mobile page speed, then pixel and Conversions API |
| Good page views, low add to cart | The page doesn’t match the ad | Land on the product in the ad, repeat its promise |
| Good add to cart, few purchases | A surprise in the cart or checkout | Shipping costs, delivery times, payment options |
| CPA rising with frequency | Creative fatigue | A genuinely new version of the idea |
| CPM rising on every ad | The market, not your ads | Hold your nerve, lean on your best converters |
| Good ROAS, flat MER | Ads claiming sales that would have happened anyway | Check the new-customer share, exclude recent buyers |
| Good MER, falling aMER | Growth is coming from existing customers | More discovery creative for new buyers |
09When to look at what
Checking every number every day leads to bad decisions. Meta needs time to learn, and most metrics are noise over a day or two. This is the rhythm we use.
Daily
Spend, cost per purchase and anything broken: rejected ads, tracking gaps, a sudden drop in page views. Look, don’t touch.
Weekly
Creative: hook and hold rates, link CTR, cost per add to cart and CPA by ad. Turn off losers, brief new ads, move winners into new formats.
Monthly
The business: MER, aMER, new-customer share, nCAC and contribution margin. Set next month’s budget from these, not from ROAS.
Quarterly
LTV, LTV:CAC and payback by the month customers were acquired. Plus a lift test or holdout when spend justifies it.
Give it time before you judge
A new ad needs enough spend to be judged fairly. Hook rate settles within a few thousand impressions. Cost per purchase needs dozens of purchases. Turning ads off after a bad day is one of the most common reasons accounts never find a winner.
10Metrics checklist
Work through these once and your reporting will tell you the truth. Your ticks are saved on this device.
0 of 10 done
Thanks for reading. The bands in this guide are where we start, not where we finish: every brand’s margins, prices and customers move them. If you’d like a second pair of eyes on your numbers, or a team that reports on profit and not just ROAS, that’s what our paid advertising service does. Book a call and we’ll take a look.
Sam Despo
Founder & Director, Pulp Digital
