The operating system behind the slide

Pilots with twenty thousand hours still read the checklist out loud before take-off. Nobody thinks less of them for it. In media operations we tend to think the opposite: that the checklist is for juniors and the experts run on instinct. That belief is where most of the expensive errors come from.

Someone who spent five years inside a global holding company recently described what actually runs agency operations, and I have not been able to forget the phrase: "A shocking amount of that stuff, in my estimation, is just Excel macros and prayers." (Quoted by ADOTAT, 15 September 2026.)

I laughed. Then I felt defensive, because I have run large media operations and I know how much real work sits behind the slides. Then I realised the description was half right, and the half it got right is the half nobody writes about. So this is an insider's list. Seven things that are actually there when execution rigour is real, from someone who has built and run them for clients in APAC, the Gulf, Europe, Australia and the US.

Where the operating layer lives now

The real operating layer for global media is increasingly built in India. Omnicom is folding IPG's captive centre into a 7,000-person India engine (Storyboard18, 2026). WPP Media runs its own model here. Publicis has had a global delivery arm in India for years. Western trade press has never described how these operations actually work, which is odd, because they run a large share of the world's campaigns.

None of what follows is exciting. All of it is what separates a 99.9% execution accuracy rate from a 97% one. At scale, that 2.9% is the difference between a renewed contract and a lost one.

The seven things

1. A written SOP per client. One document per account, not a generic playbook. It says how this client names campaigns, which platforms they use, who approves what, what the QA steps are and what "done" means. When the account changes hands, the SOP is what changes hands. Without it the knowledge lives in one person's head and leaves when they do.

2. A QA checklist before anything goes live. Every campaign, every time. Naming, targeting, budget, dates, tracking, creative specs, brand safety, frequency caps. A second person checks. In the best teams I have run, roughly three-quarters of the checklist is automated and a human checks the rest. Like the pilot, the senior person reads it too.

3. An error tracker with a root cause on every entry. "Campaign went live late" is not an entry. The entry says why it went live late, which step failed, and what changed so it will not fail the same way. An error tracker without root causes is a list of apologies.

4. A three-level backup per account. Who covers if the campaign manager is out, who covers if the backup is out, who covers if the whole shift is out. Written down, with names, before it is needed.

5. Follow-the-sun as a hand-off protocol, not a time zone. A 24-hour operation is only as good as the twenty minutes at each shift change. The hand-off is a written note, not a conversation: open Criticals, changes made in the last shift, changes waiting for approval, and who owns each account until the next line.

6. A monthly business review that reads like a contract. What was agreed, who owns what, by when, and what changed since last month. It is numbered and signed off by both sides. This is the document that ends the "but you said" conversation before it starts.

7. Skill metrics per person. Which platforms each team member is certified on, at what level, refreshed when. It sounds bureaucratic, but it is how you know, before a pitch, whether you can actually staff the account.

The workaround for item five

If you run shifts across time zones and the hand-off is still a message that says "all good", do this. Export the platform change history for the last eight hours (every major platform has one) into a shared sheet at the end of each shift. Add three lines at the top: open Criticals, approvals pending, who owns what until the next hand-off. That is the whole protocol. It takes five minutes and it removes the sentence "I did not know that had changed" from your vocabulary.

What the macros-and-prayers version gets wrong

The version in that quote exists, and here is how it goes wrong. The SOP is in someone's head. The QA is "I had a look". Errors are fixed but not recorded, so the same error comes back next quarter. The hand-off is "all good". The monthly review is a deck of screenshots. And the prayer is that nobody senior looks too closely.

The gap between the two versions has little to do with talent. I have worked with brilliant people in both. The gap is whether the rigour lives in the system or in the person. If it lives in the person, it varies by who is on shift. If it lives in the system, it does not.

Why this matters more now, not less

Automation platforms promise account launches 90% faster and budget management with 90% less time. AI agents promise setup in 30 minutes. I believe the numbers. I also notice that none of the seven items above gets easier when things get faster. It gets harder. A faster launch of a campaign with the wrong tracking is a faster way to spend money blind. Speed without the checklist just gives you a bigger blast radius.

Where we stand

We built AdNexa so that the seven items above are not habits people have to remember. They are steps the system does not let you skip. QA runs before anything goes live and the launch is gated on it. Every error gets a root cause, and the pattern is added to the daily checks so the system catches it next time. The hand-off note is generated from the day's actual changes, not typed from memory. And the monthly review is built from the same numbers the client saw every Monday, so there are no surprises in it.