Contemplative brew today. Hario V60 pour-over, because I only had time for one thing and I wanted it to be good.

Which is roughly the theme of this issue.

On August 24, YouTube redefined a view. A view now counts from the first frame. Shorts, long-form, podcasts, live, all of it. If your thumbnail autoplays for a fraction of a second while someone hovers it, that's a view.

The old definition, the one that required a few seconds of watch time, still exists. It got renamed "Engaged View" and moved into YouTube Analytics under Advanced Mode. Creators can see it. Networks with backend access can see it. You, the advertiser looking at a public channel page, cannot.

So every public view count on YouTube just went up, and not one additional human watched anything.

Glenn and I have been going back and forth on this for a couple of weeks. His position: add a term to our IOs that says YouTube impression guarantees are based on engaged views, not the new public count. “We need to put it in writing and hold the standard,” he said. If a network can't or won't give us access to the backend number, we negotiate it out case by case.

My position: we don't need it.

And I still think that's true. ADOPTER has bought YouTube for years without estimating performance based on views and CPMs. We estimate engaged audience from signals anyone can see; we price against that estimate, and we let conversion data tell us if we were right. The public view count was never our input. So writing a term that demands a number we've been fine without felt, to me, like admitting we need a crutch. It felt weak.

Here's where Glenn is right and I was arguing with my ego instead of my brain.

The term isn't for us. It's for everyone else.

An IO term is a standard you're willing to put your name on. When a partner reads "impressions are counted on engaged views" and signs it, they've agreed on what a human is. When they push back, you've learned something about that partner before a dollar moves. A term you can negotiate away isn't weakness. It's a question you get to ask every network, every time, with a signature attached to the answer.

I'm the chief mistake-maker here at ADOPTER, and the mistake was thinking a standard only matters if you personally need it. Standards matter most for the people who don't have the filter yet.

So we're adding it. The language we're working with:

If you have suggestions on a revised writing, I’m all ears.

Pricing in this industry is backed against CPM. The M (mille: Latin for 1,000) is about to inflate for reasons that have nothing to do with more people watching.

Any network that holds its CPM flat on the new count just got a raise. Not a negotiated raise. A definitional one. And most of them didn't ask for it, which is exactly why it's dangerous. Nobody has to act in bad faith for the market to overpay. The denominator moved on its own.

It also broke the measurement math underneath. Podscribe's own benchmark deck is blunt about this: YouTube attribution is modeled off views, because there's no pixel. The input to that model just changed definition. Every year-over-year YouTube comparison that crosses August 24 is now apples to something else—maybe oranges, but according to Bumper that might still even be too close a comparison.

Some networks I respect have already said they'll report on engaged views only. Great!

That should be the floor, not a differentiator.

Now the part I want to roast. And roast I shall.

Podcast buyers have never had an accurate reach number. Not once. Not in the history of the medium.

A download is not a listen. It never was. A download is a file request from a device that may or may not have a human near it. The industry spent years arguing over IAB 2.0 compliance, Apple's counting, and whether a play means anything, and the answer at the end of it all was: we still don't know exactly how many people heard the ad.

spooky

And we bought anyway. Successfully. Because good buyers never priced off reach. They priced off what happened after.

This is card counting. The dealer never shows you the deck. You track what's already been played, you estimate what's left, and you size your bet on the estimate. Nobody at a blackjack table waits for the casino to publish an accurate card count. The public view count on YouTube is the casino telling you how many cards are in the shoe. Thanks. That was never the number that mattered.

If you've been buying audio, you already have the muscles. YouTube just stopped pretending it was going to hand you the deck.

This is a soapbox I often step onto, so bear with my repetition.

Impressions don't matter. Reach doesn't matter. They're inputs we tolerate because they're the only thing anyone will put on an invoice.

What matters is action. Somebody heard or watched a message delivered through a creator, a programmatic slot, a live read, whatever the modality, and then did something. That something can be a purchase. It can be a search, a signup, a share, a shift in how they feel about the brand, a mention to a friend. The form of the action depends on what you're buying for. The existence of the action is the whole product.

So the question a buyer is actually answering is not "how many people saw this." It's "does this placement have the power to make people do something?" Reach is an input to that question. It has never been the answer.

Stop selling reach. Please. It’s a misunderstanding of the game.

I digress…

And you can reverse the formula. You don't need an accurate public number to estimate whether an audience will act. You need signals that only exist when real humans are paying attention.

Comment-to-view ratio. Tom Alush put this one on my radar years ago. A healthy channel runs roughly 300 views per comment. Comments times 300 gives you a rough floor for the audience that's actually present. The new view count doesn't touch it, because a zero-second view doesn't leave a comment. Podscribe found the same thing independently in their Q2 benchmark: comment-to-view ratio tracks purchase rate on YouTube.

See example: Jesse Michels (American Alchemy)—above average engagement.

The Reddit test. Search the show. Are strangers arguing about it? Recommending it unprompted? That's an audience with a relationship to the content, and relationships convert.

Retention curves, when you can get them. Where people drop off tells you whether the mid-roll is landing on a full room or an empty one.

Advertiser history. Who renewed, who didn't, and at what price. Imperfect, lagging, and still useful.

Your own conversion data. The one signal that ignores every definition YouTube will ever publish.

None of these is the answer alone. I've been wrong with all of them in hand. The 300:1 ratio is a heuristic, not a law, and a channel with a hostile comment section will pass it beautifully. But stack four or five of them, and you have a better read on real influence than any public counter has ever given you, before the change or after it.

If you buy YouTube, five things. Nothing else.

If you buy audio, you've been doing this for years. Keep going.

Engaged View is also YouTube's definition, and they can move it again whenever they want. We're anchoring a contract to a number controlled by a platform that just proved it's willing to redefine its numbers. That's a real weakness in the term, and I don't have a clean fix for it, other than the fact that the alternative is anchoring to a number that's already meaningless.

I digress (again)…

I fought this term because I didn't personally need it. Glenn wanted it because the industry does. He was arguing for the standard. I was arguing for my own competence. Those aren't the same argument, and only one of them is about the client.

Reach was always a stand-in for the thing we actually wanted, which is proof that a message made someone move. YouTube just made the stand-in a little more obviously fake. That's not a crisis. That's a clarification.

If you buy YouTube: has a single network volunteered engaged view reporting to you since August 24 without being asked? I want names, good and bad.

If you sell it: what would make you comfortable putting engaged views in an IO? What's the objection if you wouldn’t?

Hit reply. I read every one.

We’re all benefactors of a better industry with better standards.

Stay caffeinated, Adam