You can sign the retail media deal. Your studio still has to fill it.

Your studio lead will find out in January what got signed in September.

That is the usual order of events, and September is now.

Four thousand retail and consumer goods leaders are at Mandalay Bay this week for Groceryshop, two in three of them director level or above. The show says it will host more than thirty thousand pre-matched one-to-one meetings.

Read the three themes as production, not strategy

The show leads with three: AI on the shelf and in the store, retail media across grocery, drug, mass and club, and the shopper trip across formats.

Translate them into work.

AI on the shelf means more surfaces that display something, and something has to be put on them. Retail media across four channel types means four sets of creative specifications, each with its own sizes, rules and deadlines. The shopper trip across formats means the same promotion rebuilt for every format the shopper passes through.

Each of those is a standing commitment to produce finished assets. Not once. Every week, in every market, for as long as the deal runs.

None of the three themes is about who makes them. That is not a criticism of the themes, it is a description of the whole category. Demand for content gets sold at conferences. Supply gets discovered afterwards, by a studio.

The arithmetic nobody does

Here is the number to start from. In our own research with Savanta, 57% of 256 grocery retail marketing leaders across five markets said they already produce between 50 and 200 variants per campaign. Before anything signed this week.

Now do the multiplication yourself, because no vendor will do it for you.

Take a single promotion. Four markets. Eight places it might have to appear: two retail media networks with different specifications, the shelf edge, the app, in store, the leaflet, the site, and social. That is thirty-two executions before anybody asks for a second size, and sizes are where the real multiplication lives.

Whatever that number is for you, it is not in the business case for any of these deals.

One question, in every vendor conversation you have this quarter

How many finished, approved, localised assets per market per month does this require, and who makes them?

That wording matters. "Assets" on its own is unanswerable, so every vendor answers a different question and none of the answers can be compared. Finished, approved, localised, per market, per month is a unit. It takes nine seconds to ask and it changes the conversation in a way a pricing question never does.

You will get three kinds of answer, and each one tells you something.

Some will say it is minimal, which usually means they have not counted. Some will say their platform generates it, which is the answer to check hardest, because generated and finished are different words. And some will give you a real number, and those are the ones worth your time, because they have thought about the part of the deal that happens after the signature.

If you sell into these retailers rather than being one, the same arithmetic runs against you with an extra term. Their specifications change, and your studio usually finds out after the change has already happened.

The number nobody adds up

Add it up. Every deal you are close to signing, every pilot, every network, every screen.

Then compare that total to what your studio actually shipped last month. Not what it could ship in theory. What it shipped.

If the second number is smaller than the first, you have not found a production problem. You have found a commitment problem, and you found it before signing rather than in January, which is the entire value of asking early.

I am in a stack of those meetings myself this week, and I am going to ask the question in every one of them. Partly because I want the answers, and partly because I have never seen anyone else ask it and I want to know what happens.

So come and find me if you are here. And if you are not, leave your own number in the comments, because thirty-two is what I get from one promotion in four markets and I genuinely do not know whether that is high or low for the rest of you.

Where I am biased

I sell software in this category, so of course I think production capacity is the thing everybody underestimates. Weigh that accordingly.

And there is a reasonable objection. Plenty of these deals do get delivered, because studios and agencies absorb them the way they always have, with overtime and goodwill in November. That is a real answer. It is just not a plan, and it stops working at some volume that nobody has measured because nobody counted.

One last thought

Conferences sell the front of the operation. The decision, the partnership, the launch.

The back of it is not what any show is selling, because it does not photograph well and nobody has ever been promoted for it.

Ask anyway. Nine seconds a meeting.