You’ve probably seen this kind of thing before…
Google helpfully recommending that we ‘get more conversion value’ by decreasing our ROAS targets.
OK… Aim for more volume at a lower ROAS. That often makes sense. Does it make sense here?
Well, what impact does Google expect from its recommended change?
An additional €292 in conversion value for the bargain price of… €370.
That would be a ‘no’.
But while the negative impact of this move (a ROAS of less than 1 on the incremental spend) is easy to spot here – and shows the ‘recommended’ action to be plainly unwise – incremental ROAS isn’t always so obvious…
And its lack of transparency often allows unprofitable moves to slip under the radar.
We often judge our activity in the bigger picture of overall / blended ROAS.
The idea of maximising revenue ‘within a given ROAS’ is a useful heuristic. It allows us to follow a simple algorithm like – ‘push spend and bidding up to the point where blended ROAS falls to 400% but no further’.
We can then do what it takes to increase revenue at the price of lower ROAS, as long as overall ROAS remains above that target….
And this approach to balancing volume with cost effectiveness has plenty going for it… but it also leaves a lot of room for detrimental moves (albeit within a generally sensible framework)
(we all need room to make detrimental moves, don’t get me wrong… but let’s explore the paradigm for now)
Take the following scenario.
Mid-sized e-com account, typically ROAS-focussed, with a floor-target of 400% ROAS, and a flexible budget above that.
Running the account the best way you know how, balancing the low-hanging fruit with sensible expansion, you achieve:
• Overall spend: $10,000
• Revenue: $50,000
• Blended ROAS: 500%
Natural move within the overall objective: Increase spend with higher budgets and more aggressive bidding until ROAS drops much closer to 400%…
Activity expands to increase spend by $3000.
This leads to an additional $4000 in revenue.
• Total ad spend: $13,000
• Total conversion value: $54,000
• Blended ROAS: 415%
So, making this move increases revenue (✅) and preserves ROAS above the overall target of 4 (✅)
But the contribution of the additional spend ($3000 that could have been used elsewhere) is a ROAS of 1.33.
We’re still within – and closer to – our overall target, but was that $3000 put to good use?
(1.33 may not be quite as bad as Google’s recommendation with its incremental ROAS of 0.79… but it’s not likely to helpful in light of an overall target of 4)
Isn't that a bit... short-termist?
Of course, high-ROAS revenue – with its short-term implications – is not the only aim of marketing spend.
We continually want to be seeding the upper parts of the funnel, and sowing while we reap…
But that requires upper-funnel focussed campaigns, not simply a low-ROAS instance of our ROAS-oriented activity.
(Tempting as it may be, a vague hand-wave towards the upper funnel doesn’t justify under-target spend).
What’s the alternative?
ROAS of any number always consists of some activity under that level, and some activity above it…
If you zoom in close enough, the majority of clicks represent zero-ROAS spend, but we don’t call for them to be eliminated.
So we have to be careful not to overplay the idea of cutting ‘under-ROAS activity’.
The problem arises in degrees as we zoom out from that micro level, to meaningful decisions about expansion, campaign budgets and bidding aggression.
The aim of decreasing ROAS from 5 towards 4 in order to increase revenue is perfectly sound.
But if we can only drop ROAS to 4 by throwing in a wave of clearly-unprofitable spend, then that initial guideline needs to be put aside in order to avoid causing harm.
If on the other hand we can bring ROAS in to land more gently at 4 – by finding a larger pool of activity with an incremental ROAS of, say, 3, then our final, blended ROAS may not be any better… However, we’ll have reached the limits of what our guidelines allow, with less avoidable waste (and we’ll have a greater volume of above-target revenue to show for it).
In other words, we can factor into our thinking – however loosely – a ROAS floor not just on overall, blended activity, but on incremental spend, whatever our current ROAS may be.
All of this relates to the concept of diminishing returns or – in our field specifically – ‘marginal opportunity’.
Once the marginal return reaches a certain gradient, continuing to trade ROAS for volume simply becomes a bad deal.
Keeping sight of incrementality
We tend to think of incrementality when viewing our marketing mix through a wide-angle lens.
What are the genuinely incremental gains benefits of each marketing channel?
It’s a valuable question, and not well answered by looking at attributed conversions alone…
Less well appreciated is the importance of incrementality within a given account or campaign.
Within the framework of account-level targets, it’s not always easy to keep a good line of sight to the genuinely incremental gain generated by specific campaigns or spend increases.
And it’s increasingly hard to keep sight of those incremental effects in the era of aggregated campaign types like – in particular – PMax, which is voracious and opaque, and operates fluidly (and again, stealthily) up and down the funnel, in both prospecting and remarketing.
What’s more, Google in their messaging will often focus on the ‘opportunity to drive revenue’ without much of a thought for profitability:
• “Get more conversion value by adjusting your ROAS targets”
• “Drive more sales with Performance Max”
• “You missed on revenue opportunity in Google Ads”
The question of the cost of generating that revenue is often conspicuously absent…
So – in the face of this cacophony of ‘more’ – we for our part need to be disciplined in how we think about ‘adding’ to an account a campaign…
And try not to let the big picture mask the effects of individual actions… The changes to the input and output of your PPC activity specifically caused by the move you’re thinking about making.
To ignore those ‘incremental segments’ by dilution is just the kind of thing we complain about when platforms do it to us by blending inventory in ways that make it hard to unpick…
So let’s try to avoid doing that to ourselves!