Adobe

Marketing operations and reporting at enterprise scale, across approximately $6B in marketing operations.

At that level of spend, reporting is the instrument the operation steers by, and its value drains away with every day that passes between the event and the answer.

Reporting turnaround limited how useful it could be for operating decisions.

The numbers were right, they just arrived after the window in which they could have changed anything. An organization in that position keeps making decisions while it waits, working from memory and instinct, and the correct answer turns up later to confirm or regret a choice that’s already been made, a bit like getting the weather forecast the evening after the picnic. The reporting function ends up fully staffed, fully accurate, and structurally unable to affect the operation.

The constraint was assembly, not analysis.

Most of the elapsed time between a question and its answer went into moving, reconciling, and reformatting data that already existed, in a sequence nobody had designed, one that had accumulated over years the way a path gets worn across a lawn one shortcut at a time. Each step had a reason when it was added, and none had been checked since against the decision the report was supposed to serve. A sequence like this survives a platform purchase more reliably than almost anything else: the new tooling inherits it, and the organization ends up buying a faster version of the same delay.

A restructured reporting and data flow, designed around the decisions it supported and working backwards from there to the sources it drew on.

Each report was rebuilt from the decision backwards, starting from what has to be true for someone to act, how fresh that has to be to still be actionable, and what the shortest path is from the system of record to that answer. Steps that had survived only because they’d always been there didn’t make it past those questions, and manual assembly was designed out of the flow entirely.

An 86% reduction is out of reach for efficiency alone: people working faster don’t produce it, and neither does better tooling applied to the same process. A change of that size means most of the elapsed time wasn’t doing anything at all.

Assembly time is usually exactly that, the cost of the route the answer has to travel before anyone can use it, and a route can be shortened, sometimes to almost nothing, while the analysis itself takes the time it takes.

86% reduction in reporting turnaround.

Reporting moved inside the decision window, which means it now shapes what the organization is able to decide as well as what it’s able to see.