Is your ads agency doing a good job? 10 checks to run yourself
Ten checks you can run in Meta, Google and Shopify without asking anyone, each with what I found auditing Australian ecommerce accounts.
By Mark Tran · Reviewed
The short answer
A good ads agency can show you, inside your own Meta, Google and Shopify accounts, that the purchases Meta and Google report roughly match your real orders, that your return clears break-even once GST is counted, and that your money is going to live products through campaigns built to sell. If they can't show you those three things, the rest of the report doesn't matter.
Most advice on this question is a list of questions to ask the agency. The ten checks below are things you can look at yourself, without anyone's help. Each one comes with what I actually found auditing and running Australian ecommerce ad accounts, mostly between July and October 2026, with the brands' names removed.
1. You own the ad accounts
Where to look: in Meta Business Settings, open Ad accounts and check the owner is your business portfolio, with the agency listed as a partner. In Google Ads, open Admin, then Access and security, and check you have admin access.
What bad looks like: the ad account or the Meta pixel sits in the agency's business portfolio, or you only see the account through their login. If you leave, your history and audiences stay with them.
Google's policy for agencies managing Google Ads requires them to give you your customer ID if you ask for it.
2. Meta's purchases roughly match your Shopify orders
Where to look: pick the same dates in Meta Ads Manager (the Purchases column), in Meta Events Manager (purchases from all traffic) and in Shopify Analytics (orders). Compare the counts, not just revenue.
What bad looks like: Meta credits itself with almost every order the store took, including email, repeat and Google sales.
What I found: the gap doesn't always run the same way, which is why you have to check it rather than assume it.
- A fitness accessories brand: Meta's pixel, counting all traffic rather than just ads, recorded 416 purchases against 410 Shopify orders over the same month, within 1.3%. Its tracking was sound.
- A women's fashion store: Meta claimed 91% of every order the store took over four weeks and reported a 1.64x return. The whole business, on net sales against the same Meta spend, returned 1.44x.
- A clothing brand's promotion: of 10 orders in the promo window, 6 used the promo code, and Meta credited itself with 2. Most came direct or from Google search, where Meta couldn't see an ad click.
A gap is not automatically broken tracking. Read why Meta, Google and Shopify revenue differs before you decide.
3. Every live ad lands on a live product
Where to look: search your brand in the Meta Ad Library with Australia selected, then click through every active ad.
What bad looks like: ads still running to deleted, sold-out or redirected product pages. You pay for the click and the shopper hits a dead end.
What I found: in one women's fashion account, 14 of 31 active ads pointed at 4 product pages that no longer existed. Some of those ads had been running since May.
4. Nobody changed a setting without deciding to
Where to look: in Google Ads, open Recommendations, then Auto-apply. Then check the target ROAS on each campaign against the return it actually achieves.
What bad looks like: Google applying its own recommendations every week with nobody reviewing them, or a target that sits nowhere near real results.
What I found:
- A fashion retailer had 7 auto-apply recommendation types switched on. One quietly moved a Performance Max target ROAS from 768% to 614%.
- A bedding brand's Performance Max target sat at 160% while the account consistently achieved 450 to 500% (the full case study).
- An accessories brand had its Google target ROAS set too high at 440%. The founder caught it, not the agency managing the account.
5. Google reports brand and non-brand searches separately
Where to look: ask for results split between searches that include your brand name and searches that don't. In Google Ads, the search terms report shows which is which.
What bad looks like: one blended ROAS. People searching your name were probably going to buy anyway, so brand searches make the whole account look better than the part that finds new customers.
What I found: in one agency-managed accessories account, brand searches took about 7% of the spend for about 18% of the conversion value. Non-brand searches, the part doing the real work, returned about 3.0 to 3.2x on their own.
6. ROAS is judged against your break-even, with GST
Where to look: work out your break-even ROAS from your own margin, ad spend and management fee, then compare it with what the reports show.
What bad looks like: a report that calls 2x "good" without knowing your margins. If the purchase value your store sends to Meta or Google includes GST, their ROAS reads about 10% higher than the return your business keeps.
The break-even ROAS calculator works out both numbers: the return your costs need, and the return as Meta or Google will show it.
7. A verdict waits for enough purchases
Where to look: before anyone calls a campaign a winner or a failure, count its purchases.
What bad looks like: campaigns switched off or scaled after a week on a handful of sales.
What I found: planning a ten-day launch test for a clothing brand, I expected 2 to 6 purchases. At that count the likely range for ROAS runs from roughly 0.25x to 2.4x, so a failing campaign and a working one look the same. That test was set up to judge click-through rate, cost per click and add-to-cart rate instead, which build up much faster.
8. Budget raises are judged on cost per purchase
Where to look: for each budget increase, compare cost per purchase in the week before and the weeks after.
What bad looks like: budgets that only go up, and an agency whose main question each month is "what budget do you want next month?" rather than a recommendation with a reason.
What I found: this one is from an account I ran. I raised the budget four times in 12 days, from A$80 to A$132 a day. Over those 12 days, cost per purchase rose from A$21 to A$37 and ROAS fell from 4.9x to 3.1x. I then added a duplicate at a higher budget. It looked better over the next three days, but on 28 purchases that is too few to call. A good agency tells you when a raise didn't work.
9. Spend goes to sales campaigns
Where to look: in Meta Ads Manager, sort campaigns by objective and spend. Then check the locations each campaign targets against where you actually sell and ship.
What bad looks like: most of the money in boosted posts or traffic campaigns, which Meta shows to people likely to click or engage rather than buy.
What I found: one swimwear account had 285 campaigns. 178 were Instagram boosts, 57 were traffic campaigns and 41 were sales campaigns. Every one of the account's 12 purchases came from the sales campaigns.
10. Tracking has no gaps, and the events look real
Where to look: in Google Ads, chart conversions by week over the last 12 months and look for flat stretches at zero. In Meta Events Manager, compare checkouts started with what Shopify shows.
What bad looks like: months with no recorded conversions while spend continues, or Meta counting far more checkouts than your store has seen.
What I found:
- A fashion retailer's Google conversion tracking recorded nothing from September 2025 to April 2026, about six and a half months. Someone had said they were looking after it.
- A swimwear store's Meta pixel logged 62 checkouts started in 28 days, while Shopify showed 5 abandoned checkouts in total. Most of the excess came from a single two-day burst.
How to score it
Give your current setup a yes or no on each check.
- 8 to 10 yes: you are in good hands. Keep asking for the Shopify comparison each month.
- 5 to 7 yes: raise the gaps with your agency this week, in writing.
- Under 5: get a second opinion before the next invoice.
Two things sit outside the checklist because you can't see them in an account. Fees should be in writing and simple enough to explain in one sentence. And you should know who actually works on your account, and get a plain-English note each month of what they changed and why. One founder I spoke to got an end-of-month report and nothing in between, and caught two serious faults in her own account that the agency missed.
Want a second pair of eyes?
If you run an Australian ecommerce store and want someone to go through these with you, I do it on a free 15-minute Meta ads audit. Your live campaigns on screen, the one fix to make first. No access needed, no obligation.
Method
Every "What I found" comes from an Australian ecommerce ad account I audited or managed between July and October 2026, in fashion, accessories and swimwear, plus one earlier case study (bedding). Names are removed. Some accounts were run by an agency at the time, some by the founder, and several by me; the agency is named as the cause only where that is on record. These are examples of what to check, from a small number of accounts, not benchmarks for every store.