Most stalled ad accounts don't have one big problem. They have five or six small ones stacked on top of each other. Meanwhile the team keeps adjusting bids and budgets because those are the levers they can see. If your ROAS has drifted down and stayed there, more tinkering in the ads manager rarely fixes it. Working through the account in the right order usually does.
I've been through this. At Art Republic I rebuilt ROAS from about 3x to 10x+, during a period when I was also overhauling the social and content strategy after an acquisition. What follows is the order I'd work through any stalled account in, whether it's fashion, a luxury product or a lead-gen business.
Decide what ROAS actually needs to be
Before you try to improve ROAS, work out what number the business needs. Plenty of teams chase a target someone picked years ago and nobody has questioned since.
Start with break-even ROAS. The rough formula is one divided by your gross margin after product cost, shipping, payment fees and an allowance for returns. If you keep 50p of every pound, break-even is 2x. If you keep 25p, it's 4x. So a 3x ROAS can be healthy for one brand and quietly losing money for another.
Then separate the numbers that flatter you from the ones that don't. Brand search and retargeting almost always show high ROAS because they pick up people who were going to buy anyway. Prospecting looks worse on paper but is where growth comes from. Report them separately, and keep an eye on blended ROAS (total revenue divided by total ad spend) as the honest check on whether the platforms are telling the truth.
Targets should also move with spend. At Money Clothing, ROAS held consistently at 4x+ and hit 5-6x in sales periods, on £10-20k a month normally and six figures in peak months. As a general rule, every extra pound you push into an account is spent on a slightly less efficient customer than the last one. If you scale budget and expect ROAS to stay flat, you'll be disappointed.
Audit the tracking before you touch the campaigns
I've lost count of the accounts where the "ROAS problem" was partly a tracking problem. Fix this first, because every decision after it depends on the data being right.
Things to check this week:
- Pull the last 30 days of revenue from Google Ads and Meta, add them together and compare with your actual store or CRM revenue. If the platforms claim more than you actually took, they are double counting.
- Check purchase events fire once per order, and that the value being passed is consistent. Including VAT in one place and excluding it in another is a common one.
- Make sure consent mode and server-side tracking (Meta's Conversions API, Google's enhanced conversions) are set up properly. Without them you lose signal, and the algorithms optimise on less information.
- Look at attribution windows. A change in default settings can make an account look like it has collapsed overnight when nothing real has changed.
It's dull work. It's also the cheapest improvement you'll ever make.
Cut the spend that drags the average down
Once you trust the numbers, export every campaign, ad set and product with its spend and return over the last 60 to 90 days. Sort by spend, not by ROAS. You're looking for the big pots of money delivering very little.
A rule of thumb that holds up surprisingly often: the bottom fifth of spend contributes close to nothing, and cutting it lifts the account average straight away. Common culprits include broad match keywords bleeding into irrelevant searches, Meta's Audience Network placements, display inventory nobody chose deliberately, and Performance Max quietly hoovering up branded searches and taking credit for them.
Go through the search terms report line by line and build a proper negative keyword list. Exclude brand terms from PMax where you can and run brand in its own campaign. If something has run for three months without getting near target, pause it. Don't keep nudging it and hoping.
Rebuild the structure around how people actually buy
There are two ways structure goes wrong. Either the account is so fragmented that no campaign gets enough conversions for the algorithm to learn, or everything is lumped together and your best-sellers swallow the whole budget while new products never get a look in.
The fix is to structure around commercial reality rather than habit. Group products by margin and price band, so a low-margin entry product isn't competing against your most profitable lines with the same target. Give hero products their own campaigns. Keep prospecting and retargeting separate so you can see what each is really doing.
On search, a lean structure built around long-tail, high-intent terms usually beats a sprawling one chasing volume. Someone searching for a specific product and size is far closer to buying than someone typing a category word.
And look at your product feed. For Shopping and PMax, the feed is your ad copy. Titles that lead with what people actually search for, clean images and accurate pricing make a bigger difference than most bid strategy changes.
Treat creative as the main lever on social
On Meta and TikTok, targeting has largely been handed to the algorithm. That means creative now does the targeting. The ad decides who stops scrolling, and who stops scrolling decides who the platform shows it to next.
Most stalled social accounts are running the same handful of ads they launched months ago, with small tweaks to the headline. That isn't testing. Fatigue sets in, frequency creeps up, costs rise and ROAS slides.
What works better is a steady flow of genuinely different concepts. Product in use. A customer talking to camera. A founder explaining why the product exists. A straight comparison. As a rule of thumb, aim for three to five new concepts a month, each with a modest test budget, and spend real time on the first three seconds. That's where most ads win or lose.
Set kill rules in advance, so you're not making emotional decisions about ads you liked making. If a concept hasn't beaten your account average after a fair amount of spend, move on.
Fix what happens after the click
ROAS is revenue divided by spend. Most teams only work on the spend side. The revenue side is often easier.
If your cost per purchase stays the same and average order value rises by 20%, ROAS rises by 20%. Bundles, accessories, sensible cross-sells and a free delivery threshold set just above your current average order are some of the simplest ways to do it.
Check the landing pages your ads point to. Slow mobile pages, a product that's out of stock in the popular sizes, or a homepage link where a product page should be will all kill conversion rate, and the ad gets blamed.
Then think about the customers you've already paid for. Proper email capture and a set of automated flows (welcome, browse abandonment, basket abandonment, post-purchase) turn one paid click into several orders over time. That won't show in platform ROAS, but it shows in blended ROAS and in the bank.
Test with discipline, and keep a record
The last piece is rhythm. Stalled accounts often have too many changes happening at once, so nobody knows what worked.
Change one meaningful thing at a time per campaign. Give each test enough budget and time to reach a clear answer, usually at least a couple of weeks for anything conversion-based. Write down what you changed, why, and what happened. A shared spreadsheet is fine.
Review weekly for waste and creative fatigue, and monthly for structure and targets. Where you can, run a simple holdout test (pausing a channel or region for a period) to see what it's genuinely adding. It's the clearest way to find out whether retargeting or brand spend is earning its keep.
FAQ
What is a good ROAS?
There's no universal answer, because it depends on your margins. Work out your break-even ROAS first by dividing one by your gross margin after all variable costs. A good ROAS is comfortably above that, with enough headroom to cover overheads and still leave profit.
How long does it take to improve ROAS?
Cutting waste and fixing tracking can show results within a couple of weeks. Structural changes and new creative take longer, because the platforms need time to learn and you need enough data to judge. Plan on a few months for a proper rebuild.
Should I cut budget to improve ROAS?
Cutting wasted spend almost always helps. Cutting good spend just to push the ROAS figure up can shrink revenue and growth, so be clear whether you're optimising for efficiency or for profit. The best accounts do both: they remove the waste, then reinvest in what's working.
If your paid media has stalled and you'd like a second pair of eyes on it, I take on consulting work, and I'm also open to Head of Marketing and Head of Growth roles, particularly with DTC and consumer brands. You can get in touch through the contact link on tomosborne.me.