Two-thirds of the budget, no idea what came back
By Dino Nokic · AUG 2026 · 3 MIN READ
Gartner asked chief supply chain officers two questions worth putting side by side. Where is the digital money going, and did it come back?
67% of supply chain digital investment now goes to AI. And 55% of the officers signing off on it are unclear what the return has been.
Read those together, because separately they're both unremarkable. Two-thirds of the budget committed is a story about conviction. Half the buyers unable to name the return is a story about measurement. Put them in the same sentence and you get the actual condition of the industry: the money is committed, the measurement isn't.
That is not an underinvestment problem. Nobody needs to be talked into spending here. It's a problem of not knowing, at the moment of spending, what you were going to compare the result against.
You can't measure a return you never baselined
I've written before that pilots without a baseline can't succeed, because success was never defined — they can only produce enthusiasm, and enthusiasm has a shelf life of about a quarter. This survey is that failure at industry scale.
Here is how it goes in practice. Something ships. It feels faster. Everyone in the room agrees it feels faster. Then somebody senior asks what it saved, and the honest answer is that nobody counted the before. So the team reaches for whatever number is available — usage, adoption, hours "freed" — and none of it survives contact with a CFO, because none of it was ever tied to the thing the operation actually cares about.
The baseline is not paperwork you do to satisfy finance. It is the only proof you will ever have, and it is only available before you start. Miss that window and it's gone permanently.
Supply chain makes this easy, if you look in the right place
Operations has an advantage here that most functions don't. The work is a chain of handoffs, and handoffs leave evidence. The paperwork that arrived late. The load that sat because a confirmation hadn't been forwarded. The exception that got chased three times by three people who didn't know about each other.
You don't need a modelled business case to baseline that. You need to count what slipped through last month. That number is sitting in your inbox and your exception log right now, and it costs nothing but the afternoon it takes to total it up.
That's also why the seam is the right place to start. Departments already have owners and metrics and people who know the work cold. The seam between them has no incumbent, no defender, and — conveniently — a loss you can count.
What Gartner actually prescribes
Their distinction is a sharp one. A change methodology is the set of steps you use to execute one initiative. A change strategy is how you allocate finite change capacity across all of them. Most organizations have the first and not the second, which is why every AI initiative gets the same generic rollout treatment regardless of what it's worth.
Gartner's prediction: by 2030, organizations that rightsize change management this way will get twice the ROI on AI initiatives compared with those still running legacy methodologies. Their analyst's framing is that the hard part is no longer executing change on a single project — it's deciding where limited change capacity should go.
Which is really an argument about attention. Change capacity is scarce in exactly the way management attention is scarce, and spreading it evenly across every initiative guarantees that the ones that mattered got the same thin slice as the ones that didn't.
Three things before the next line item
Count the before. One metric, one number, one date, written down where somebody else can see it. If you can't produce that number, you are not ready to buy anything.
Point it at a seam, not a department. That's where the loss is already countable and where nobody will defend the status quo.
Decide in advance what would make you stop. A threshold that would cause you to switch it off is worth more than a business case, because it's the only part of the plan that can be falsified.
If you can't say what last month cost you, you can't say what the system saved you. Two-thirds of the budget is already moving. The baseline is the cheapest thing in the project and the first thing skipped.