Andrew Faris challenged my CPMr thinking. Here's what changed.

· Articles

Andrew Faris challenged my CPMr thinking. Here's what changed.

Phil here, Director of Paid Search. I was on Andrew Faris's podcast this week.

We talked about CPMr for an hour. Andrew pushed back hard in the right places. I want to expand on three things I said that didn't get enough airtime.

Phil Kiel Tweet

Phil Kiel Quote Retweet (Why I don’t shut up about CPMr)

CPMr is a flag, not a fix

The most important thing I said in the whole conversation, and I want to make sure it lands, is this: we do not optimise for CPMr. We use it as an indicator.

Three specific uses:

  1. A diagnostic indicator when performance has already dropped.

  2. A leading indicator when performance is about to drop.

  3. A confirmation indicator when you've changed something and want to know whether it worked before spend and revenue confirm it.

The flag analogy matters here. CPMr is not the wind. The wind is everything: the auction, your account structure, your creative, your exclusion hygiene, your product development, your CRO decisions. CPMr is the flag. When the flag moves, something changed. Your job is to find out what.

Andrew asked me whether this is a Meta problem or an advertiser problem. I said both, and I meant it. Meta's algorithm is doing exactly what it's designed to do. It finds warm, already-converting audiences, converts them efficiently, reports a strong ROAS, and gets rewarded with more budget. That is correct behaviour from Meta's perspective.

But Meta monetises impressions. Advertisers need reach. These are not the same incentive. Meta charges the same whether you reach 10,000 people once each or 2,000 people five times each. It has no financial reason to flag the difference. That's not a bug. It's how the system is designed. The error is assuming those incentives are aligned and not measuring the gap yourself.

The example that matters most

I mentioned a brand I started working with about three years ago. UK market. Single purchase product. Six or seven years in business. Very little product development. Heavily overindexed on Facebook in a population of 60 million.

CPMr was the thing that showed up when we compared year-on-year data across every metric in the funnel. Spend, impressions, reach, frequency, CTR, add-to-cart rate, CPA, ROAS. When we laid it out, the metric that had changed the most was CPMr. And that had translated directly into spending more and reaching fewer people.

What I said on the podcast, and what I believe: the media buying work was largely irrelevant in that account. Not because campaign structure and exclusions don't matter, they do, but because the root cause was outside the ad account entirely. They had advertised to most of the reachable UK audience for their product category over seven years. Without new products or meaningful channel diversification, there was no mechanism to reach different people. The flag was pointing at the business, not the account.

This is the thing most agencies won't say. It's easier to sell a campaign restructure or a creative refresh than to tell a founder that the ad account can't solve a product problem. But that's the correct diagnosis when the correct diagnosis is the correct diagnosis.

Phil Kiel Tweet

The Q4 implication almost nobody is running

We touched on this briefly and I want to make it more concrete.

CPMr is a planning tool, not just a diagnostic. If your CPMr is up 50% year-to-date compared to the same period last year, and you want to double your Black Friday revenue, you are going to need more than double the spend. Not to grow. Just to reach the same number of new people you reached last year, before any additional growth on top.

The brands that have this conversation in July are in a different position to the brands that have it in November. In July you have time to fix exclusions, consolidate campaigns, build Partnership Ads volume, and let the account settle. In November you are just spending more into a saturated account at Q4 auction prices.

Year to date versus the same period last year. That comparison, done weekly from now until October, is worth more than any single account change you could make.

Go listen to the full episode. Andrew asked better questions than I expected and pushed me on the places where I was being too vague. The part about retargeting and incrementality is probably the most nuanced section and worth listening to in full.

Links

Watch on episode on Youtube here:https://www.youtube.com/watch?v=aCVhMts9cR8

Listen on Spotify here:https://open.spotify.com/episode/5DIjDZ8aiuPcukVRbgMAUn?si=d362369020a448fd

Read our Q1 CPMr Report:https://taikundigital.com/q1-2026-cpmr-report

Read more about CPMr:https://taikundigital.com/phil-kiel/cpmr-calculation

Book a call with me.

Back to all articles