Your numbers
These come from your answers. Each one is worked out from your exact amounts and rounded once, as it's printed: money down to the whole dollar, ROAS up, both on the safe side, so a sum redone from the rounded figures here can come out a dollar or 0.1x apart. Your break-even and targets change only when your price, costs, shipping, fees or returns change; your expected cost per purchase is your own A$125 from the last 30 days, and it updates as your results change.
- Today: about A$125 per purchase, from your last 30 days. That's from only 8 purchases, too few to plan from yet, so this check-up uses a starting estimate of A$45–A$75 until you have 10 purchases in 30 days.
- Average order: A$185, including GST. After GST, product cost, shipping and payment fees (taken as about 2%) you keep A$88 of it (47%), before ads.
- GST: You're registered for GST, so 1/11 of every sale is GST, not yours: A$16.82 of an A$185 order. These numbers start from A$168.18, your A$185 order without GST.
- What your ads sell: the A$229 Stonewashed linen sheet set. After GST and costs it keeps A$112 ex GST of each sale before ads, against A$88 on your A$185 average order, so your targets below use it; your lift test uses your A$185 average order, because your sales are every basket.
- Break-even cost per purchase: A$112 ex GST on what your ads sell, the A$229 Stonewashed linen sheet set; A$88 on your A$185 average order. Pay more than this for a first order and you lose money on it.
- Break-even ROAS: 2.1x (rounded up from 2.04x for safety) on what your ads sell, the A$229 Stonewashed linen sheet set, on sales including GST. Each A$1 of ads has to bring back at least this much in sales.
- What Shopify sends Meta: we haven't confirmed whether its purchase value includes GST and shipping. Your 2.1x break-even ROAS (on what your ads sell, the A$229 Stonewashed linen sheet set) is on sales including GST, so check one real order in Events Manager: if Meta's value leaves GST out, compare Meta's ROAS with 1.9x on the Stonewashed linen sheet set, and 1.9x on your A$185 average order, instead.
- Allowable cost for a new customer: A$97, your A$88 break-even on your A$185 average order, because your first orders are any basket, plus credit for repeat orders: 10% of it (half your 20% repeat rate), the most a new customer is worth if they come back as often as your repeat rate says. If most first orders are the Stonewashed linen sheet set, it's A$121 (its A$112 break-even plus the same credit). It's a ceiling, not a target.
- Your targets: A$89 ex GST per purchase on what your ads sell, the A$229 Stonewashed linen sheet set (80% of your break-even, so each sale keeps a margin of safety for Meta's over-counting and the weeks that go badly), a ROAS of 2.6x on sales incl. GST (2.4x if Meta's purchase value leaves GST out).
- How to judge your ads: lift. Your A$1,400 a week of sales already includes your ads, so judge each change on what it adds, from Shopify's own sales incl. GST, after refunds, Australian orders only (filter by shipping country). At today's spend, sales should hold at about A$1,400 a week. Each week, also note Shopify's orders with utm_medium=paid and utm_source meta or facebook and the ones it shows as coming from Facebook or Instagram, with how many are first orders, beside Meta's purchase count: that's the closest read, before your first raise, of how many of Meta's purchases your ads brought in. After a raise, the lift is the rise in sales ÷ the rise in spend: at least 2.1x, your break-even ROAS on your A$185 average order, so your first raise, from A$33 to A$39 a day, which adds A$42 a week of spend, needs sales up at least A$88 a week, or the raise loses money. Judge it on 4 weeks against the 4 before, leaving out sale weeks: one week's sales swing by more than that by chance. MER is a record only.
- Expected cost per purchase: A$45–A$75 (a starting estimate, not yet your own figure: replace it with your own cost per purchase after the later of 4 weeks or 30 purchases)
- MER now: about 6.1x: A$6,105 of Shopify sales ÷ A$1,000 of ads, last 30 days. Meta counts 8 of your 33 orders (24%): those are the purchases it credits to its ads, not everything your pixel sees. Check Events Manager's Purchase count for the same 30 days: near 33 means tracking is fine and the gap is sales Meta didn't drive (repeat, email and organic orders); well under it means purchases are going missing (cause 2.4). MER includes the sales you'd make without ads, so it can't tell you on its own whether the ads pay: it's a record, not a target.
- Do your ads pay today? To cover A$1,000 of ads in 30 days, they must bring in about A$2,088 of sales you wouldn't have made without them (A$1,000 ÷ A$88.57 kept per A$185 order, the exact figure). Meta claims 8 purchases, about A$1,480 at your A$185 average order, more than Meta claims, so on Meta's own count your ads don't pay for themselves yet.
- What your budget buys: about 3.8 purchases a week, at A$33 a day now; about 4.5 at your full A$1,200 a month. That's too few purchases for Meta to learn from. The paid plan sets out your choice of optimisation event.
What's holding your Meta ads back
Your ads health score: 5/100 (F)
We checked 12 things on your ads: 3 likely costing you sales, 8 we couldn't confirm are working, 1 looks fine.
Strict scoring: a check only scores when your answers show it's working. Anything we can't confirm scores zero.
How we score: each check is weighted by how much it can cost you. A check that looks fine earns its full weight. Likely, worth checking and not answered earn nothing. Score = earned ÷ total × 100. Grades: A is 90 or more, B 75–89, C 60–74, D 40–59, F under 40.
Your plan shows how to fix each one, in the order that costs you most.
Meta's ad system is an AI. It decides who sees your ad by watching who does the thing you told it to optimise for, then finding more people like them.
When store ads "don't work", the AI is usually doing its job with bad inputs: too little data, the wrong signal, a pixel that miscounts, or a sale that can't make money. Sometimes it's simply taking credit for sales you'd have made anyway.
Before you take anyone's advice, including an AI's: Meta makes its money when you spend yours. The recommendations in Ads Manager, the Opportunity score (0–100) that rises when you apply them, and most of the default settings are built to get more ads running and more budget spent. Meta itself says a high score doesn't mean better results. Ask an AI chatbot how to run Meta ads and you mostly get Meta's own advice back, repeated from its help pages and the blogs built on them. Some of that advice is good. It just isn't measured against your margin. This check-up is.
These are the twelve causes we check first. The ones your answers point to are marked.
You said you "can't tell which ads, if any, are paying for themselves", and that's the right question to ask of any ad spend. From your answers, two things stand out: you're paying for clicks and likes, not sales, and your budget is split across too many ad sets. Below is what each is likely costing you; your plan has the fixes.
2.3 You're paying for clicks and likes, not sales
Applies to you: Likely, from your answers
From your answers: You boost posts.
What's happening. Meta gives you what you ask for. Boosted posts optimise for things like profile visits, website visits or messages (Meta Business Help Center, "Boost an Instagram post"). The Traffic and Engagement objectives work the same way. Meta finds people who tap. Tappers and buyers are different people.
How to spot it.
- Your Results column counts link clicks, landing page views or engagements.
- Cheap clicks, flat sales.
- Your ads came from the Boost button, not Ads Manager.
Your plan shows how to fix this, step by step.
2.2 Your budget is split across too many ad sets
Applies to you: Likely, from your answers
From your answers: You run 2–3 ad sets.
What's happening. Every ad set learns on its own. Four ad sets on one budget means four learners, each with a quarter of the data. One ad set per age group, interest or product is a common chatbot-built setup. On a small budget, none of them ever learns.
How to spot it. Count your active ad sets. Your budget supports 1. Usually one takes most of the spend while the others starve.
From a client's accountA client's Australian online store consolidated several ad sets into one, with ad spend going from A$6,540 to A$6,840. Purchases rose from 131 to 205, or from 30.57 to 47.83 a week on average, and cost per purchase fell from A$49.92 to A$33.37. Attributed revenue went from A$15,982 to A$26,035, and ROAS from 2.44x to 3.81x. That's what happened alongside the change, not proof the change caused it, and one ad set isn't a rule for campaigns that serve different markets or goals.
For you: you run 2–3 ad sets on about 12 adds to cart a week, so each gets about 4–6, far under the about 50 a week each needs to finish learning.
Your plan shows how to fix this, step by step.
2.5 Meta is claiming sales you'd have got anyway
Applies to you: Likely, from your answers
From your answers: Meta counts 8 of your 33 orders (24%), and you said "…I can't tell which ads, if any, are paying for…".
What's happening. Broad and Advantage+ campaigns can show ads to your existing customers, your email subscribers and people already on their way to buy. When they buy, Meta counts the sale if they bought within 7 days of clicking an ad, or within 1 day of viewing or engaging with one (Meta's default attribution setting: 7-day click, 1-day view, 1-day engage-through). Retargeting is the strongest case: many cart abandoners come back on their own.
So Meta's ROAS climbs and your bank balance doesn't. That's why this check-up judges your ads on lift: your store's sales above its normal weekly baseline ÷ your ad spend, from Shopify. MER (all store sales ÷ all ad spend) stays a record, never a pass mark: it counts the sales you'd make without ads.
How to spot it.
- Meta's ROAS looks great, but store revenue has barely moved since ads started.
- Most orders in ad weeks come from returning customers.
- Retargeting has the best ROAS in the account and keeps getting more budget.
From a client's accountA client's Australian online store looked beyond Meta's reported ROAS. Returning buyers' share of its prospecting purchases fell from 64% to 31%, and first-time customers rose from 193 to 269. Meta's reported ROAS fell from 4.93x to 4.09x on Meta spend of A$9,360, then A$9,540. Over the same periods, whole-store revenue went from A$59,000 to A$65,940 and blended MER from 3.51x to 3.93x, on total paid-media spend unchanged at A$16,800. The gap doesn't show Meta was wrong: ROAS and MER measure different things. This is what happened alongside the change, not proof of cause.
For you: judge your ads on lift from Shopify: whether sales, Australian orders only (filter by shipping country), hold at A$1,400 a week at today's spend and rise at least 2.1x any extra spend, over 4 weeks, and on new-customer orders. MER (6.1x now) is a record, not a target.
Your plan shows how to fix this, step by step.
2.1 Your budget can't feed the event you're optimising for
Applies to you: Worth checking
What's happening. Meta learns from the one event you choose. It flags an ad set "Learning limited" when it's unlikely to get about 50 of those events in the week after its last significant edit (Meta Business Help Center, "About the learning phase"). Purchases are the rarest event in your store. At your budget there are too few for Meta to learn from, and the step before them, add to cart, comes to about 12 adds to cart a week.
With almost no purchases to learn from, Meta can't tell a buyer from a browser. It drifts toward whoever is cheapest to reach, often people who tap and leave. It looks like "the AI isn't working". It's working on almost no data.
How to spot it.
- The Delivery column says "Learning limited", or "Learning" for weeks.
- Results shows a dash, or one or two purchases a week.
- Most of your spend goes to an age group or placement that doesn't buy from you.
From our own accountsA new Australian jewellery brand spent about A$1,300 optimising for purchases with no purchase history. Meta received zero purchase events to learn from, and delivery drifted to people aged 45+ in the Facebook feed.
Your plan shows how to fix this, step by step.
2.4 Meta can't see your sales properly
Applies to you: Worth checking
What's happening. The pixel and Conversions API tell Meta what shoppers do in your store. Three things go wrong:
- Missing. No pixel, Purchase never fires, or the Conversions API is off, so ad blockers and privacy settings hide part of your sales.
- Doubled. The pixel is installed twice, say by your platform's Meta app plus code in your theme. One sale counts as two. Pixel and Conversions API can both send a purchase only when they share an event ID, which is how Meta removes the copy (Meta for Developers, "Deduplicate pixel and server events").
- Polluted. Test purchases, staff orders and your own add-to-carts land in your live data. Events sent with a test code still flow into Events Manager and are used for targeting and measurement (Meta for Developers, "Using the Conversions API"). On a new store, your team can be most of what Meta learns from, so it looks for more people like them.
How to spot it.
- Compare purchases in Events Manager with orders in Shopify for the same 7 days. Meta higher means doubled. Meta far lower means missing. A small gap is normal.
- Meta Ads Data Advisor (formerly Meta Pixel Helper) shows two pixels on your site.
- Add-to-cart spikes line up with days your team was testing.
From our own accountsIn a new Australian online store's first weeks, the only add-to-cart and checkout events were the team's own testing.
Your plan shows how to fix this, step by step.
2.6 Meta's AI is changing your ads
Applies to you: Worth checking
What's happening. Advantage+ creative lets Meta change your ad automatically: brightness, crops, templates, overlays, music, and where your text appears (Meta Business Help Center, "About Creative Enhancements"). Stores get extras that pull in your catalogue: product cards on your image, a single image turned into a catalogue carousel, and links to other pages on your site.
From 28 October 2026, Meta is adding in-app product browsing to existing ads, so shoppers browse your products inside Facebook or Instagram (Meta's notice to advertisers, reported by PPC Land, 15 Oct 2026). It can be switched off per ad in the Advantage+ creative and destination settings.
Why it matters: a cropped catalogue packshot can replace the image you chose, and in-app browsing skips your website, where your free-shipping offer, product details and returns policy live.
How to spot it. Open each ad's Advanced preview and look for carousel or collection versions you didn't make. Then find the ad on your phone in the real feed. A correct-looking editor isn't proof.
From our own accountsA new Australian jewellery brand's single-image ads were being shown as roughly 40-card catalogue carousels. About A$59 and 1,600 impressions went out before it was caught.
Your plan shows how to fix this, step by step.
2.7 Your catalogue isn't connected, or your feed has errors
Applies to you: Worth checking
What's happening. Your catalogue is Meta's copy of your products. Catalogue ads (Meta calls them Advantage+ catalog ads) need it to show each shopper the products they looked at (Meta Business Help Center, "About Facebook Advantage+ catalog ads"). A broken feed means missing products, wrong prices or images, and ads pointing at things you can't sell.
How to spot it. In Commerce Manager, open your catalogue. Does the item count roughly match your live products? Does the Issues tab show errors (Meta Business Help Center, "How to use the Issues tab in your catalog")? Are prices in AUD?
Your plan shows how to fix this, step by step.
2.8 Meta's AI doesn't have enough creative to work with
Applies to you: Worth checking
What's happening. Meta finds buyers partly by testing your different ads on different people. One or two ads give it nothing to choose between. The same people see the same image, stop noticing it, and your cost climbs.
Variety means different angles, not one photo with three captions: the product alone, in a real room, the material up close, in use, a customer's own photo, the offer. And don't assume video wins. Shoppers study a good still, especially products they inspect closely. Run stills and video as separate ads in the same ad set, and judge each one by your stop rule.
How to spot it.
- Fewer than 3 active ads, or every ad is the same format.
- Frequency rising, click-through rate falling, cost climbing together.
From our own accountsAn Australian jewellery brand's video-only ad set spent about A$133 in 7 days for no add-to-carts.
Your plan shows how to fix this, step by step.
2.9 You're selling below break-even
Applies to you: Worth checking
What's happening. Ads can't fix the maths. Each sale your ads bring in earns you A$112 before ads, on what your ads sell, the A$229 Stonewashed linen sheet set. Through Meta, each purchase Meta counts is expected to cost A$45–A$75. Your cost per new customer from Meta is your Meta spend ÷ the new-customer orders in Shopify with utm_medium=paid and utm_source meta or facebook, or, while some of your live ads' links carry no tags, all the new-customer orders Shopify shows as coming from Facebook or Instagram. Those first orders are any basket, so compare it with A$88, the break-even on your A$185 average order: above it, every extra first order loses money, and better targeting won't change that. Repeat orders can bring a new customer up to about A$97, but treat that as a ceiling, not a target. If most first orders are the Stonewashed linen sheet set, it's A$121, and its break-even is A$112.
How to spot it. Work out your cost per new customer for the last 30 days (Meta spend ÷ the new-customer orders in Shopify with utm_medium=paid and utm_source meta or facebook, or, while some of your live ads' links carry no tags, all the new-customer orders Shopify shows as coming from Facebook or Instagram) and compare it with A$88 on your A$185 average order (A$112 when most of those orders are the Stonewashed linen sheet set), and check whether you only get sales when you discount.
From a client's accountA client's Australian online store improved its offer before increasing spend: Meta spend stayed at A$5,660 and total paid spend at A$7,330. Average order value rose from A$85 to A$104 and purchases from 156 to 182, while cost per purchase fell from A$36.28 to A$31.10. Attributed revenue went from A$13,260 to A$18,928, ROAS from 2.34x to 3.34x, whole-store revenue from A$28,220 to A$35,680 and MER from 3.85x to 4.87x. That's what happened alongside the new offer, not proof it caused it, and revenue isn't profit: check what the offer costs you in margin.
For you: your break-even is A$112 per purchase on what your ads sell, the A$229 Stonewashed linen sheet set, and A$88 on your A$185 average order. Work out any bundle or threshold at its new margin before you raise spend.
Your plan shows how to fix this, step by step.
2.10 Your product pages, cart and checkout lose the sale
Applies to you: Worth checking
What's happening. Meta's job ends at the click. Your store closes the sale: the product page, then the cart and checkout. Slow pages lose people before they load. Weak pages lose them after: no shipping cost or delivery time, no returns policy, few photos, no reviews, no contact details, or a business name that doesn't match across your terms and contact page. The cart and checkout lose people who had already chosen: a shipping cost they first see at checkout, a slow cart or checkout, or no way to pay that they trust.
How to spot it.
- In Ads Manager, compare link clicks with landing page views. A big gap means a slow page or accidental taps.
- In Shopify's reports, compare sessions that added to cart with sessions that reached checkout. A big drop means the cart or checkout loses them.
- On your phone, send yourself a product page link in an Instagram message, open it there and tap Add to cart once: it shows whether the cart works inside Instagram's browser, where most ad clicks open. Stop at the cart: never a test checkout or a test purchase on your live pixel.
- Our check of your site: your Stonewashed linen sheet set page, the one your ads link to: it took 3.4 seconds to show its main image on a phone (Google rates 2.5 seconds or less as good); its price, shipping and returns information were on the page.
Your plan shows how to fix this, step by step.
2.12 You're taking Meta's recommendations on trust
Applies to you: Worth checking
What's happening. Ads Manager shows recommendations and an opportunity score (0 to 100, in Account overview and above the Campaigns table) that goes up as you apply them (Meta Business Help Center, "About opportunity score"). The score tracks how many of Meta's suggestions you've adopted, not whether your ads make you money: Meta says it doesn't measure performance. Recommendations can also be set to apply themselves (auto-apply, in Account overview), so changes can land in your account without you choosing them. Typical suggestions: raise your budget, broaden your audience, add placements, switch on more Advantage+ features. Any of them can be right for you. All of them mean more spend, or more of your decisions handed to Meta, and some feed the problem in cause 2.5: credit for sales you'd have made anyway.
The same advice reaches you second-hand. AI chatbots and most "how to run Facebook ads" content repeat Meta's guidance, so following an AI's setup steps usually means following Meta's recommendations without seeing the score.
How to spot it.
- You've accepted recommendations to lift the score.
- Your spend grew after a recommendation, but your store's revenue (your MER) didn't move.
- Your campaigns were set up by following an AI chatbot's steps.
- Auto-apply is on in Account overview.
From a client's accountA client's Australian online store acted on recommendations that raised Meta spend from A$7,560 to A$11,340, and total paid-media spend from A$9,450 to A$13,230. Purchases rose from 186 to 221, but cost per purchase climbed from A$40.65 to A$51.31, Meta-attributed ROAS fell from 3.27x to 2.59x and blended MER from 5.22x to 3.87x, while whole-store revenue moved only from A$49,320 to A$51,150. After a reset to A$8,190 of Meta spend (A$10,080 of total paid-media spend), it recorded 213 purchases at A$38.45 each, ROAS of 3.46x, and MER of 5.23x on A$52,740 of store revenue. Read the three periods together: they show what moved alongside each change, not which change caused it.
For you: judge every change on lift, not on the score: sales holding at about A$1,400 a week, and rising at least A$88 a week for your first raise, from A$33 to A$39 a day, which adds A$42 a week of spend.
Your plan shows how to fix this, step by step.
2.11 You keep changing things, so Meta keeps starting over
Applies to you: Unlikely, from your answers
What's happening. Pausing an ad set or changing its optimisation event, audience or creative is a significant edit. Big budget changes can be too (Meta Business Help Center, "Last significant edit"). Edited ads can also go back through review. Tweak a headline on a live ad and you can lose the momentum you just found.
How you start matters too. A new ad set on a daily budget that starts at 7pm tries to spend the whole day's budget in the hours left.
How to spot it.
- Your ad sets almost always say "Learning".
- Ads go "In review" after small fixes.
- Day-one spend is bunched into a few hours.
Your plan shows how to fix this, step by step.