How to use ROAS (or CPA) targets in Google Ads

What happened in Dublin?

I’ve done my fair share of grumbling over the loss of visibility and control in Google Ads over the last few years.

And while I always try to keep an open mind about the benefits of Google’s increased use of automation – and to give positive recommendations for charting the best course through it – still I confess to a certain amount of fruitless whingeing.

So it made a nice change, a couple of weeks ago, to find myself face to face with the product team responsible for smart bidding…

It was one of the breakout sessions following the keynote at the Marketing Live conference in Dublin.

These were valuable sessions – with Google’s participants genuinely receptive to feedback – and quite forthcoming with info on upcoming developments (which we were requested to keep in the room … So what I’m about to share here is – I hope – non-specific enough to uphold the spirit of discretion…)

So, I started with a fairly minor feature request… for the currently missing ability to make ad group-level target changes in bulk (e.g. select multiple ad groups, and set or modify their ROAS/CPA targets as a group.)

The Google guys were sceptical about this, and even suggested the lack of this feature was ‘semi-intentional’ on the grounds that they ‘don’t really encourage ad group level target changes’.

Here I sensed danger…

To me, ad group-level targets are not just a nice-to-have. They’re pretty central to optimisation… And with PMax already operating without asset group-level targets, the removal of ad group-level targets would be entirely in keeping with Google’s direction of travel.

So I quickly jettisoned my (now trivial) request to focus on my new assignment: ‘preserve ad group-level targets’

And when they asked the question: ‘why / when do you use ad-group level targets?’ Luckily, I was prepared… (I‘ve written a course unit on it…)

Why we need ad group-level targets

I explained that when some ad groups are outperforming the desired target (consistently) and have room to spend more, then we want to lower their targets, encouraging more aggressive bidding in order to trade some excess ROAS for unused volume.

Conversely, where we have high-spending ad groups that are underperforming target, they call for a target increase, to drag their ROAS up, and minimise the damage until that happens.


NB These arrows are not definitive recommendations for the ad group targets based on the data in the screenshot – but are the kind of moves I may be looking to make depending on how persistent these performance patterns are, how long the current target has had to take effect, what a genuinely acceptable ROAS is, etc.

Can't the algo handle that?

Of course the overarching idea on Google’s side is that the algorithm can, and should be left to, identify and act on those kinds of performance disparities by itself…

But too often the analysis stops at this point (including it seems by Google’s teams who aren’t in the trenches with these tools).

The algorithm ‘should’ find and exploit performance disparities across the impressive range of variables that Google always tells us about…

But as we see all too often, it doesn’t do this particularly well… or consistently… or quickly… (and of course there may be other considerations beyond ROAS that we want to take into account in weighing up or down a given ad group).

We often see certain segments (keywords, device categories, demographic groups or whatever they may be) consistently underperforming, and yet continuing to take up a substantial portion of budget, with unfavourable ROAS, however patiently we wait for the algorithm to take the preservational action that is so obviously needed.

And when this happens, one of the main levers we have at our disposal is to guide the algorithm’s approach to those segments through target adjustments…

How did that go?

I’m happy to report that Google’s team listened, and took this on board as valuable feedback.

I’ll call that a win for us (and a nice change for me from moaning into the wind about these issues.)

When we went around the table for final thoughts at the end of the session, these were mine:

• Please consider giving us bulk editing controls for ad group-level ROAS targets
• If that’s not getting onto the agenda, at least please don’t get rid of ad group-level ROAS targets
• And I won’t ask for keyword level ROAS targets or I know you won’t take me seriously

They took the last one as a joke, which is how I said it… (but to be honest, if I had my hands on the reins…)

Share this post

Lastest Posts

Google Ads Updates - Q1 2024 Roundup

No slowdown in the pace of change with Google Ads updates in Q1 2024… Here’s…

Aggregation vs Articulation: Competing Virtues of PPC Management

Three ‘A’s Two concepts have been central to the recent evolution of Google Ads: aggregation…