Beware of ROAS, ROI’s dangerous digital twin

Too great a focus on return on ad spend (ROAS) is leading to short-term thinking and under-investment, which in turn is stifling growth, and it has the potential to be far more damaging than ROI.

Source: Shutterstock

People have been warning about the potential problems with ROI for years. But its digital twin, ROAS (return on ad spend), might actually be doing more damage today. In fact you could call ROI and ROAS allies in marketing’s anti-growth coalition.

If ROI’s dangers can seem quite theoretical, ROAS’s dangers, while similar, are actually a day-to-day reality in significant parts of the digital marketing world.

Before turning to ROAS, let’s quickly recap the core issues it shares with ROI.

ROI’s issues in brief

ROI tends to inversely correlate with profit growth, as due to diminishing returns ROI decreases as you spend more, and increases as you spend less. So the easiest way to increase your ROI is to decrease your media spend. Focusing on increasing ROI would therefore limit growth or even “send you broke” as Byron Sharp says.

Instead you should prioritise the incremental profit or revenue you achieve. ROI is not actually a measure of effectiveness but how efficiently you achieved it. So don’t use ROI as a target, use it to help you check the value for money you’re getting for your media investments. As Les Binet likes to say: “Effectiveness first, efficiency second – net contribution to value first. ROI is a useful metric but not the only one.”

The above is well known to some, but if people need a refresher take a read of my latest blog post ‘Marketing’s most marmite metric‘, or Tim Ambler’s coruscating article from 2004, ‘ROI is dead: now bury it’, a classic that should still be essential reading for marketers and finance people 18 years on.

Despite Ambler’s article, ROI lives on and never was buried. In fact, around the same time he was digging its grave, it actually spawned a new offspring for the digital age which shares the above problems and more besides: ROAS.

Almost half of marketers say brands ‘too focused’ on ROI

So what is ROAS?

ROAS is in very common use by many digital marketers as a buying objective in the real-time optimisation of performance activity across a range of platforms and channels.

The ‘return’ means the sales that occur during a specific time period when your advertising was served by the adtech. If someone is served an ad and if they purchase within a set time frame, those sales are attributed to the activity.

Don’t use ROI as a target, use it to help you check the value for money you’re getting for your media investments.

Whereas ROI is ideally calculated using market mix modelling (MMM, aka econometrics), ROAS is calculated using digital attribution, which can vary from platform to platform. Here’s a helpful article from Google on the different types of attribution modelling and the different ways credit can be attributed.

Digital marketers and finance teams love ROAS. They love the immediacy, predictability, the near instant reaction of the platforms and their algorithms, the confidence that if they plug the number into the adtech, it will buy them a sale.

The issues

Given its very common use as a target set by finance teams and as a buying objective in digital media, ROAS is likely to be leading to a real-time, real world version of the theoretical problems with ROI: short-termism, under-investment, de-prioritising longer-lasting activity and stifling growth.

ROAS isn’t what it says it is

The word ‘return’ in the name creates an illusion of causality that just isn’t true from how it’s calculated. The ‘return’ means the total sales from people who happened to be targeted with ads during a set period of time. If someone buys something at some point soon after they were targeted with an ad, it’s marked down as a ‘win’ for the platform. Or alternatively the last platform to receive a click within the chain might receive all the credit. Ex-Adidas marketer Simon Peel says: “ROAS is a misnomer. It should be called ‘credit for ad spend’.”

ROAS can take credit for other channels’ earlier work

ROAS can foster a sense that channels ‘compete’ rather than work together. Imagine a football manager believing their centre forward is entirely responsible for every goal so ditching their defence and midfield for 10 centre forwards. ROAS is only measuring what happens in the final third of the pitch. As Peel says: “It is a fraction taking credit for the whole.”

Chasing ROAS chases easy sales, not growth

The closer an audience is to the buying decision, the higher the ROAS will likely be. These are people who already know you and are ready to buy. And you can’t rely only on them for your growth, you need to fill your funnel with people who are further away from purchase, so will naturally deliver a lower ROAS. You need a more balanced plan, rather than expecting all activity to have a high ROAS and switching off anything that doesn’t.

Chasing ROAS can mean targeting people who would buy you anyway

A high ROAS may even mean the opposite of an ad ‘causing’ a sale: the algorithms finding people who were going to buy anyway. So it’s not just like targeting low-hanging fruit, but targeting falling fruit. Or like hanging outside a shop and tapping shoppers on the back as they enter then claiming you’ve enticed them in. It can mean inadvertently prioritising channels based more on their technical ability to tag people than their ability to reach or get attention from them.

ROAS may inversely correlate with growth

Brand growth comes disproportionately from light buyers, but focusing on high ROAS can lead to you targeting more and more heavy buyers, so limiting growth. It can make brands inward-looking and too focused on existing customers, rather than on reaching new customers.

So like ROI, ROAS has been presented as a growth metric, when it’s actually anything but. In fact, ROAS appears precision-engineered to keep brands small.

Some real world damage caused by maximising ROAS

Here are just a few real world examples I’ve seen that illustrate some of this:

  • A famous sportswear brand testing some well-known remarketing technology wasted millions of adspend, because while the tech claimed £4 ROAS, only 1-2% of the sales were found to be incremental.
  • A well-known online fashion retailer that’s been targeting a total ROAS across all media for five years has seen its share of search and share of market decline ever since.
  • A respected B2B financial services brand shifted budget to search, its most efficient channel, and subsequently saw net customer acquisition, revenue and market share decline year on year.
  • A sports brand heavily increased ROAS-optimised digital spend and saw share of search decrease and price sensitivity increase year on year.
  • A jewellery brand consolidated spend solely into commercial periods when they saw high ROAS, and saw brand associations with key category entry points decline over time.

So how did ROAS get so big?

A conspiracy theorist might think ROAS was created by the adtech world to frame what success looks like on their platforms, to massage the apparent success of campaigns, and make their platforms look like they perform better than they actually do.

Imagine a football manager believing their centre forward is entirely responsible for every goal so ditching their defence and midfield for 10 centre forwards.

Conspiracy or not, ROAS is among the innovations that helped the platforms and adtech become such dominant forces in marketing. Alongside digital’s promises of precision targeting, trackability, attribution and accountability, ROAS has proved to be one of digital’s decisive weapons against ‘traditional’ channels, which weren’t so easily able to demonstrate their value to a new breed of advertisers in real time.

New channels will appear to have an advantage over established channels if people believe wrongly that the higher ROI or ROAS they’re seeing from them is a sign of their superior effectiveness rather than that they’re new and people aren’t spending much on them yet.

Direct Line: Marketers must claim the right to be commercial

ROAS is part of a skewed view of how advertising works

ROAS doesn’t stand up to a lot of scrutiny when seen through the lens of how advertising really works, not how adtech people assume it works.

In adtech’s worldview, people are rational economic actors: someone is targeted with an ad on platform A; they are now aware of product X; they consider it better than product Y; they click and buy; platform A was responsible for the sale.

We should be wary of the platforms’ sales pitch that ‘our platform A will drive more of your sales with a higher ROAS than platform B’ given it’s based on such a simplified fantasy about how people, brands, ads or buying work.

To be fair, it’s a philosophy that’s been around for a century. It’s the ‘strong force’ theory of advertising, and it attributes far more power to advertising or an individual channel or piece of communication than is ever the case.

Advertising mostly influences people’s behaviour by building and reinforcing brand memories before they’re in the market, and when they fall into the market, it can refresh those memories to help the brand be a little more likely to come to mind and be chosen. So rather than driving awareness, consideration and conversion, digital advertising is really part of a rich tapestry of thousands of little touches (including all sorts of stuff that isn’t ‘advertising’ or ‘digital’) that combine together to gently contribute to maintaining or increasing sales.

Ads on a given platform do not exist in a vacuum – with their own distinct ROAS number that can be calculated within that platform alone.

ROAS appears precision-engineered to keep brands small.

Below is a version of the standard ‘awareness-consideration-conversion’ funnel created to try and bring the performance and the brand marketing worldviews together and a little more in line with how brands grow and how advertising works. To build brands in people’s minds, nudge people who are in-market towards choosing your brand, and to help connect people to a brand online who have already decided to purchase. It’s not perfect of course but it appears to resonate with marketers with usually quite different perspectives.

So what now?

Brands are networks of mental associations. Audiences’ use of the platforms is interconnected and fluid. We should create creative campaigns and media strategies that build on and exploit this. Campaigns that bend, flex and scale across platforms. We should plan our media holistically not horizontally, not in separate rows on a spreadsheet and by splitting up available budgets in multiples of £100,000 per channel. And evaluate our campaigns independently from the platforms’ desire to present themselves in a more flattering light than the others.

If the digital advertising ecosystem is going to mature and evolve beyond its origins as a set of short-term direct response channels into the full-funnel brand-building marketing ecosystem it wants to become, the platforms need to become more geared to helping brands genuinely deliver real growth, to be better able to help build demand for brands as well as harvest that demand. So we need them to continue to re-engineer and evolve the way their ad formats, algorithms, measurement and metrics work.

Marketers face ‘tension’ in push to prioritise effectiveness

They are definitely trying to evolve. Many platforms now offer advertisers a much broader set of objectives and metrics than just ROAS or CPA targets, including reach and other brand goals. Some companies like ours have the expertise to bend the algorithms to make them work more effectively through the funnel. Being able to customise the analytics platforms now also means you can explore replacing ROAS as a target with other metrics that are more likely to incentivise actions that generate incremental growth from new customers, such as variable cost per outcome (example below), new SKU sales or rate of growth by territory.

This is not an argument against the use of hard metrics, it’s an argument for a more considered use of genuine growth metrics, not targets that can inadvertently limit it, and an argument for broadening out the metrics we use and never only focusing on maximising one.

But a strong bias towards efficiency metrics such as ROAS remains among many performance marketers. Digital marketers and their finance people will need to change their mindset, philosophy and understanding of how brands really grow and how ads really work. It’s definitely happening – a lot of ‘performance’ marketers really get it – but it will naturally take time.

Crucially, marketing teams will need to change the way they work together and are incentivised. If teams are set up to optimise everything to ROAS as a KPI, they could be unwittingly pushing an agenda that sets them against their colleagues and is to the detriment of their organisation.

But don’t just take my word for it on ROAS. Avinash Kaushik, digital marketing evangelist at Google, said this in his personal blog post ‘Die ROAS, Die’ in 2019: “I profoundly dislike ROAS…It distresses me to no end that companies and agencies all around us use ROAS as currency – inflating its value far, far beyond its minuscule value. It is a navel-gazing advertising-centric metric. It is not a business metric.”

Just imagine how much better digital advertising could be if we could get beyond arguments about how it all works. If we stopped using competing KPIs or metrics that limit rather than promote growth. If we stopped believing we all had to pick a team – team brand or team performance. Then we might be able to ditch our entrenched brand or performance marketing perspectives, for a marketing performance one.