Case notes · Programmatic & data
An operating model linking setup, inventory quality, pacing and verification into one measurable flow.
Anonymous case note. No client names, no logos, no figures and no reserved terms — only publishable method.
The organisation ran programmatic buying across several platforms, with different teams and suppliers that rarely talked to each other. Every activation began from settings decided case by case, with no shared baseline defining what was acceptable before any spend went out.
They wanted to stop discovering problems once a campaign was over. The question was not "how do we buy more" but "how do we make what we buy predictable and verifiable", linking initial setup, inventory quality, spend rhythm and final control into one repeatable path.
Not a single intervention, but a way of working: written rules, steps in the right order and checkpoints that hold up even when people or platforms change.
Campaign settings lived in the head of whoever launched them. Without a shared baseline, every new activation risked starting from scratch and reintroducing mistakes already seen before.
The inventory being bought wasn't classified consistently: there were no clear criteria to separate what was suitable from what should stay out, nor to explain the reasoning to anyone outside the operational detail.
Spend rhythm was corrected after the fact, once something had already gone wrong. There was no way to notice in time that spend was heading in an unwanted direction.
Checks existed, but arrived late and in a form that never fed back into setup. Verification described the past without correcting the present.
We put the starting rules in writing: what to include, what to exclude and on what criteria. The baseline became the point every activation starts from, instead of an improvised decision.
We defined how the inventory being bought is judged, with reasoning legible beyond the operational team, so a choice can be explained and reproduced the same way.
Spend rhythm was tied to thresholds and signals decided in advance, so intervention happens when direction shifts, not once the damage is done.
Checks were connected to upstream decisions: what verification observes becomes a concrete correction to the starting baseline, closing the loop.
First the baseline and criteria, then rhythm governance, finally verification feeding back into setup: order matters more than speed.
Starting rules no longer live in conversations: they are written, versioned and accessible. Whoever activates a campaign starts from the same point, even if they weren't there at the origin.
Inventory is read through common criteria and explicit reasoning, so a choice stays understandable over time and doesn't depend on who made it.
Spend rhythm has attention points decided in advance: the organisation notices in time when direction drifts away from what was intended.
Control is no longer a closing report but a step that re-enters setup: what is observed becomes an updated rule for the next time.
Moving the criteria to the start of the flow changed the nature of the work: less time spent explaining what went wrong, more time building on a baseline that didn't need renegotiating every time.
Treating pacing as a governed rule rather than a reaction removed surprises and made spend decisions arguable instead of endured.
A system anyone can re-read and reproduce beats a brilliant result that lives in one person's head. Continuity is part of the value.
Why is this case anonymous?
Because we share only the publishable method. The client's name, sector, platforms, suppliers and any economic data stay out: what we share is the way of working, not the identity of whoever adopted it.
What can you reuse from this work?
The framework: a written setup baseline, legible inventory quality criteria, rhythm governance decided in advance and verification that feeds back into decisions. The specific choices always need recalibrating to each context.
Does it only work for large organisations?
No. An orderly flow matters more than volume: even with few campaigns, writing down the rules and connecting verification to setup reduces the mistakes that keep repeating.
How do you know it's working?
When a new activation starts from the same baseline without restarting, when an inventory choice can be explained to someone who wasn't there, and when a check produces a concrete correction instead of a plain summary.