Solved: Inconsistent tROAS, what causes erratic ad spend?

Hi there,

Thanks for reaching out! I'm happy to give you some initial thoughts on the tROAS situation you've described. It's a common headache, and tbh, the way most people are told to use it is a recipe for disaster. You're seeing exactly what happens when the theory of automated bidding meets the messy reality of a live ad account – underspending, unpredictable results, and the feeling that you're fighting the algorithm instead of working with it.

The short answer is you're probably focusing on the wrong lever. Trying to force a higher tROAS on an unstable foundation is like trying to tune a car engine while it's falling out of the chassis. We need to look at the underlying structure and data first. Let's get into it.

We'll need to look at why your tROAS is so unpredictable...

Right, first things first. Let's debunk a major myth. Target ROAS isn't a simple instruction you give to Google. It’s not a thermostat where you set the temperature and the machine just delivers it. It's a constraint you place on a predictive algorithm. You're telling it, "Don't enter any auction where you don't believe, based on hundreds of signals, you can hit this return." When you set the target too high, or when the algorithm doesn't have enough data to be confident, it simply stops bidding. That’s your spend plummeting right there.

The erratic performance you're seeing – some campaigns hitting 280% while others slump at 180% on the same 220% target – points to one massive culprit: data sparsity. For tROAS to work properly, a campaign needs a steady, consistent flow of conversions. The general rule of thumb is at least 30-50 conversions per month, per campaign. If your campaigns are getting fewer than that, the algorithm is essentially guessing. Some guesses will be lucky (280% ROAS), and some will be terrible (180% ROAS). It's not optimising; it's gambling with your money.

You asked how common it is for actual ROAS to be higher than your target. It's fairly common, especially with reporting lag. A user might click an ad today, but not actually make a purchase for 3-4 days. Google's system might have bid based on a lower predicted value, but the final purchase was larger. So, the reported ROAS for that conversion looks great. The problem is, you can't build a scalable strategy on these happy accidents. Relying on them is a fast track to inconsistent results.

Here's a simple way to think about diagnosing the problem. Follow this logic for each of your campaigns.

I'd say your breakeven point is the real killer here...

Let's be brutally honest for a second. A 240% ROAS for breakeven is incredibly tight. It means for every £2.40 in revenue you generate, you're spending £1 on ads. That leaves just £1.40 to cover the cost of the product itself, payment processing fees, shipping, staff, overheads... everything. Your margin for profit is wafer-thin, and your margin for error in advertising is basically zero.

This is a business model problem as much as it is an advertising problem. When margins are this low, you are completely at the mercy of the ad platforms. Any small fluctuation in CPCs or conversion rates can immediately push you into the red. You can't afford to have a bad week, and you certainly can't afford to invest in testing and learning, which is essential for scaling.

I'm reminded of an eCommerce client we worked with who sold cleaning products. They were convinced they needed a 500% ROAS to be profitable. However, when we actually dug into their numbers, their true breakeven was closer to 300%. This insight was a game-changer. It allowed us to set a more realistic tROAS of 350-400%, which unlocked far more campaign volume and, paradoxically, led to a much higher actual ROAS of 633% because the algorithm finally had room to breathe and find buyers. They had been strangling their own growth by aiming for a number they didn't actually need.

You need to understand your numbers inside and out. Use this calculator to see how your gross margin impacts the ROAS you need just to break even, let alone turn a decent profit. You might be surprised at how much pressure your margins are putting on your ad campaigns.

You probably should stop trying to 'force' the tROAS...

Your strategy of nudging the tROAS up by 10% weekly is, with the best will in the world, one of the most common ways to kill a perfectly good campaign. Every time you change that target, you're essentially telling the algorithm: "Scrap what you learned last week, the goalposts have moved. Go find me conversions at this new, harder target."

This sends the campaign straight back into the 'learning phase'. During this period, performance is expected to be erratic. If you're doing this every single week, your campaigns never leave the learning phase. They are in a permanent state of confusion, which is why spend becomes volatile and then dries up completely as the target becomes unreachable.

A much better approach is to find a stable, profitable tROAS and let it run for at least 1-2 weeks without touching it. Let the algorithm stabilise and build up a reliable performance history. If you then want to increase profitability, you should only ever make changes of around 15-20% at a time, and then leave it alone again for another couple of weeks to see how the system adapts. Frequent, small tweaks are poison to automated bidding.

Think about it like this: your current approach is like constantly poking a sleeping bear. You're never letting it settle down and do its job. The better way is to give it clear instructions and then step back to let it work.

You'll need a better campaign structure to feed the algorithm...

This brings us to the most important, actionable advice I can give you: fix your campaign structure. If the root cause of the problem is data sparsity, the solution is to pool your data together. It's highly likely that your account is over-segmented, with too many campaigns and ad groups, each with its own little budget, all fighting for a handful of conversions. This is a very common mistake.

You need to consolidate. Group similar products or categories with similar ROAS potential into fewer, bigger campaigns. For instance, instead of having separate campaigns for "Men's T-Shirts," "Men's Hoodies," and "Men's Jackets," you might create one single "Men's Apparel" campaign. Yes, you lose some granular control, but you gain something far more valuable: data density. By feeding all the conversions for men's apparel into one campaign, you give the tROAS algorithm a rich, consistent stream of data to learn from. It can then make much more intelligent decisions about which auctions to enter and how much to bid.

I've seen this consolidation principle work firsthand. For example, one of our outdoor equipment clients came to us with dozens of small, sputtering campaigns. We consolidated them into just five core product category campaigns. The effect was immediate: by pooling the conversion data, we gave the algorithm the fuel it needed. Their overall account ROAS climbed, and we were able to increase spend by 50% while maintaining profitability. That new structure drove over 18k website visitors, simply because it allowed the bidding algorithm to finally work as intended.

Here's what that structural shift looks like in practice:

This is the main advice I have for you:

So, to bring this all together, you need to stop tinkering with the tROAS percentage and start rebuilding the foundation of your account. The goal is to create an enviroment where automated bidding can actually succeed. It involves a phased approach, moving from diagnosis to restructuring, and only then to strategic scaling. It takes patience, but it's the only way to achieve stable, profitable growth.

As you can see, it's a methodical process. The problem you're facing isn't a quick fix you can solve by changing one setting. It requires a strategic rethink of how you're structuring your account to work with the algorithms, not against them. It also requires a deep understanding of your own business's profit margins to set goals that are actually achievable.

This is where expert help can make a significant difference. An experienced eye can audit an account like yours, quickly identify the structural flaws and data gaps, and implement a robust new structure based on years of seeing what works across hundreds of accounts. It's about moving from constantly reacting to daily performance swings to proactively building a system for predictable growth.

If you'd like to go through your account together, we offer a completely free, no-obligation 20-minute strategy session where we can screen-share and pinpoint exactly where these issues lie and what a new structure could look like for you. It's often the fastest way to get clarity on the path forward.

Regards,

Team @ Lukas Holschuh