PPC Strategist

17/07/2026

From 17th August, budget-limited campaigns with ROAS or CPA targets will stick to those targets more tightly. That's the promise, anyway - and it's worth thinking about what it can and can't mean.

A target can never say "only spend if you hit this ROAS". The algorithm needs time to establish whether the target is achievable, and it needs to explore auctions - some of which won't pay off - to gather the information that makes optimising towards it possible. Zero tolerance would produce absurd behaviour. So there has to be tolerance for deviation, in both directions, for a certain amount of time. What's changing can only be the width of that tolerance band.

And it has always run wider than is often appreciated. Smarter Ecommerce (smec)'s study of ROAS target changes in practice shows it well: the median result tracks the target faithfully enough, but the spread around it is considerable. The target moves, and actual ROAS follows it - generally, eventually, and with a fair degree of variation. That's the target-performance relationship we've been living with up to now.

https://smarter-ecommerce.com/blog/en/google-ads/target-roas-changes-in-google-ads-your-faq-guide/

Whatever the new tolerance band looks like, it won't be zero. Our job from the 17th onwards is to recalibrate - to redevelop a feel for how much deviation the new regime allows, and for how long.

The campaigns most clearly in the firing line are the ones currently beating their targets. If you want to protect one, lower its target to a ROAS you'd genuinely accept - but not far in advance. The change will have to trigger some form of relearning phase in any case, so there's little to gain by jumping the gun.

The other thing I'll be watching is whether the narrowing is symmetrical. In principle, it cuts both ways - overperformers on ROAS pulled down, underperformers pulled up. For a few reasons, I'm not so sure we'll see as much of the second part:

1. Every example in Google's help page on the change ( https://support.google.com/google-ads/answer/17061251 ) is of an overperformer being brought down to its target.

2. Ginny Marvin has clarified that "this change will not lead to any changes in spend". Likely she meant it purely as a reassurance that Google won't spend more of your money. But taken literally, it could only be true if the change is one-directional. There's no reduction in spend implied by a budget-limited overperformer being made to bid more aggressively - but tightening up an underperformer means less tolerance for auctions with lower expected value, and that has to mean (at least opening the door to) lower spend.

3. Which brings us to the cynical one… pulling performance up towards a ROAS target will entail a reduction in cost. Cost to us = revenue to Google… an equation to which Google is very much alive.

I expect movement in both directions - but perhaps not in equal measure. How unequal is the first thing I'll be measuring after the 17th.

Review campaigns and adjust targets using the Bid Target Adjustment Tool.Starting August 17, 2026, Google will be making changes to its bidding systems, making it easier for you to get more consistent

19/06/2026

Running PMax is a bit like riding an elephant.

🧐 Stay with me…

• The skill of the rider matters, even though you don't have full control.
• It has senses you don't, and it won't tell you what made it bear left.
• An elephant with a feed and one without are not the same beast.
• The same big feet that can take you through difficult ground can flatten what's underneath.
• There are often good reasons not to do it…

If you want to see the analogy stretched to breaking point, my latest blog post takes it the rest of the way!

👉

22/05/2026

Simpson's Paradox. The Denominator Problem. Attribution lag.

They may sound like abstract concepts, but they show up in Google Ads accounts all the time - and they're behind some of the most dangerous analytical mistakes in PPC.

In my latest post, I break down five analytical pitfalls that can easily lead you to the wrong conclusion. Each one involves data that looks straightforward on the surface but has something underneath that subtly distorts it.

A few may be familiar... Probably not all of them!

👉

Five analytical traps that sneakily distort Google Ads data - from Simpson's Paradox to the Denominator Problem - and how to catch them.

19/05/2026

There’s more to Google Ads Editor than most of us appreciate.

I’ve been using it for years, and I only recently started exploring some of the features buried beneath the usual bulk edits and find-and-replace. A few of them are genuinely powerful.

In this blog post, I’ve pulled together five that are worth knowing about. They range from small time-savers (a spell checker you probably didn’t know existed) to some properly useful QA tools - like filter functions that let you query ad groups based on what’s inside them, and custom rules you can build to catch mistakes before you post changes.

If you use Editor regularly, there’s a good chance at least a couple of these will be new to you.

05/05/2026

Most of us run Microsoft Ads by importing from Google. Which means most of us are overlooking the extensions Microsoft has built that Google doesn’t offer at all.

Action Extensions, Filter Links, Flyer Extensions, Video Extensions - all native to Microsoft, all worth exploring.

01/05/2026

There's more to Google Ads Editor than most of us appreciate.

I've been using it for years, and I only recently started exploring some of the features buried beneath the usual bulk edits and find-and-replace. A few of them are genuinely powerful.

In this week's post, I've pulled together five that are worth knowing about. They range from small time-savers (a spell checker you probably didn't know existed) to some properly useful QA tools - like filter functions that let you query ad groups based on what's inside them, and custom rules you can build to catch mistakes before you post changes.

If you use Editor regularly, there's a good chance at least a couple of these will be new to you.

👉

Google Ads Editor has more depth than most of us use. Five features - from formula words to custom rules - that are worth exploring.

28/04/2026

If your remarketing campaign is reaching more people than expected, Optimized Targeting might be on. With it enabled, your audiences aren’t strict targets - they’re signals. Worth checking under ad group settings.

21/04/2026

My latest blog post covers a switch to maximise conversion value that looked, on paper, like an obvious win.

The account had clear, differentiated lead stages and a meaningful value gap between them, so the logic seemed sound.

In practice, CPCs jumped, lower-stage conversions dropped by a third, and the higher-stage conversions we were chasing also declined.

Follow the link in my bio to the post to find out why it went wrong and what to do instead. It comes down to giving Google a better clue. (Did you know ‘clue’ originally meant a ball of string? Ariadne’s, to help Theseus find his way out of the labyrinth.)

21/04/2026

Demographic reports in Google Ads get a fair bit of attention for the known groups. Age brackets, male versus female.

There’s another row that tends to get ignored. ‘Unknown’. It looks like a noise bucket - the leftover users Google couldn’t classify.

But ‘Unknown’ isn’t a random sample. It’s mostly users who weren’t signed into their Google account when they clicked. A specific cohort, which often performs meaningfully differently from the averages of the known groups. If it underperforms, it’s a candidate for exclusion like any other weak segment.

17/04/2026

👀 Value-based bidding is meant to beat max conversions. The logic is hard to argue with: tell the algorithm which conversions matter more, and it'll guide itself better than a flat-value strategy ever could.

I've argued this principle myself, repeatedly.

Then I moved two campaigns in a plateauing account over to Maximise Conversion Value. Strong value differentiation, clean setup, the textbook case for it.

CPCs in the campaigns I'd switched jumped by over 60% in a month, and the high-value conversions I was actually chasing declined... and so did the lower-value conversions that had been doing fine without it.

The mechanism is the interesting part, and it's changed how I think about what to feed smart bidding in general. There's a precondition for VBB that's rarely discussed, and when it isn't met, the theoretical advantage becomes an active handicap.

Full breakdown, the data, and the two ways out in my latest blog post.

👉

Value-based bidding outperforms max conversions in theory. In one plateauing account, it didn't. Here's why, and what the mechanism tells us about smart bidding.