No — AI does not replace media planners. It replaces the assembly parts of the planning job and leaves the judgement parts. Drafting plans, mapping budgets across channels, pulling the weekly performance view, and formatting the deck are largely automatable now. Deciding which constraint actually matters, what to do about it, and staking a recommendation on that call is not. The net effect is that the role concentrates: fewer planners, doing more consequential work.
I have been on both sides of this question for a long time. In 2011 I wrote that media planners would lose their jobs because of Google AdPlanner. I got the direction right and the timing very wrong, and the reason is the whole answer to this question. AdPlanner automated the assembly and left the judgement untouched. That is exactly what today's models do, only with more of the job in scope.
The honest short answer
The question strips a job down to a binary it was never built as. A media planner is not one activity; it is a bundle of them, and AI hits the bundle unevenly. When people ask whether AI replaces planners, they are usually imagining the whole role flipping at once. That is not how automation has ever worked in this trade.
What happens instead is a re-sorting. The activities that are rule-based, repetitive, and reversible get absorbed by the model. The activities that require context, taste, accountability, and a point of view stay with a person. A planner who is mostly the first set is in trouble. A planner who is mostly the second set becomes more valuable, because the machine has removed the busywork that used to dilute them.
So the answer is not "yes" or "no." It is: AI replaces the assembly layer of the role, and the planner's future depends on which layer they actually live in. If you cannot name what you do that a model cannot, that is the finding — not a forecast.
What was actually assembly
Assembly is any part of planning that takes a defined input and produces a defined output by following a process. Channel budget splits across a fixed total. Pacing calculations. Reach and frequency estimates from a planning tool. Formatting the plan into a client-ready shape. Building the competitive landscape summary.
None of this is trivial, and for years it was a genuine skill to do it quickly and without errors. But it is skill of the assembly kind: learnable, repeatable, and checkable. A model with the brief, the constraints, and the historical data can now produce a competent first pass of every one of those outputs. It will not be perfect. It does not need to be, because the planner was never shipping the first draft unedited either.
The tell is reversibility. If the work can be regenerated cheaply and checked against rules, it is assembly. Channel splits, pacing tables, and formatted decks all pass that test. That is why they are the first parts of the job to move to the machine.
What was actually judgement
Judgement is the part where the input is ambiguous and the output is a call rather than a computation. Which of three plausible objectives is the real one this quarter. Whether the client's stated budget is actually a decision or an opening position. What the number means when the result comes in below plan but the brand metric moved.
These do not have a right answer hiding in the data. They have a defensible answer that a person has to own. A model can lay out options, name the trade-offs, and even recommend. It cannot sit in the room and be accountable for the choice — and accountability is a large part of what senior planning has always been. The more the machine handles the assembly, the more the whole job becomes this: choosing among good options and standing behind the choice.
Why the 2011 prediction didn't land
When I wrote that first post, the logic looked clean. Google had a tool that could generate a media plan, so planners should have been nervous. Some were. But the jobs did not disappear, and the reason is instructive.
AdPlanner automated the assembly and nothing else. It could produce a plan; it could not decide which plan was right for a specific client in a specific market with a specific history. Planners absorbed the tool, used it for the mechanical half, and kept the judgement half. The prediction was directionally right — the assembly work did get automated — but it mistook the assembly for the job. That is the same mistake people make about AI now, just with a bigger tool.
What's different this time
What is different now is the breadth and the ease. AdPlanner did one narrow thing. A modern model can do research synthesis, plan drafting, copy generation, reporting, and localization in one workflow, and it can be pointed at a company's own context instead of a generic planning tool. The share of the job that counts as assembly is simply larger than it was in 2011.
That widens who feels the change. It is not only junior planners doing the mechanical work; it is the mid-level planner whose week is full of first drafts and formatted updates. The judgement requirement has not changed, but the bar for what counts as "value-add" has risen, because the machine now does a credible version of what used to be a full week.
I came back to this exact fear myself later, in a more personal post about losing my own job to machine learning. The lesson there holds: the anxiety is real, and the response is not to defend the assembly but to move up the stack deliberately.
Which planners survive
The planners who stay valuable are the ones who do three things. They take responsibility for the call, not just the document. They can explain why a plan is right in a way a client trusts, which is a communication and reasoning skill, not a production skill. And they can direct the machine — set the constraints, judge the output, and catch the plausible-but-wrong draft that models produce.
The planners at risk are the ones whose entire value is producing the artifact. If your day is generating plans and updates that a senior person signs off on, you are standing where the automation lands. That is not a moral failing; it is a description of the workflow. The move is to get closer to the decision — to be the person whose judgement the first pass is serving.
What to do on Monday
Start by auditing your own week against the assembly/judgement line. List what you did, mark each item as regenerate-and-check or own-the-call, and look at the ratio. If the list is mostly regenerate-and-check, that is the signal to redesign the role, not to wait.
Then pick one judgement activity and get better at it deliberately: write the recommendation before you write the plan, defend a choice to a colleague, or take a small budget decision end to end. The Strategy guide maps where planning sits in the wider operating model, and the course FAQ works through the questions planners ask first.
The honest answer, after nearly two decades in this trade, is that AI does not replace planners. It replaces the version of the job that was mostly assembly, and it makes the judgement version more valuable than it has been in a long time — provided you are willing to do the harder half.
FAQ
Will there be fewer media planners in five years?
Probably fewer, and more senior on average. The assembly layer that once supported large junior benches is exactly what AI absorbs, so teams will be smaller and the floor will be higher. Fewer people, each doing more consequential work, is the shape I would plan for.
Is planning more exposed to AI than buying?
They are exposed in the same way but at different speeds. Planning produces documents and options, which are easy to generate, so the assembly share is high. Buying involves live decisions and relationships under time pressure, which are stickier. Plan for both to move, with planning first.
What skill should a junior planner build now?
The ability to make and defend a call. Learn the mechanics, but treat the mechanical work as the entry ticket, not the job. The juniors who thrive will be the ones who can explain why a recommendation is right, direct a model to produce a better draft, and take responsibility for the result.
Should planners learn to build AI workflows?
Enough to direct them, not necessarily to engineer them. You need to know what the model can see, where it fails, and how to judge its output. That literacy lets you design your own role around the judgement instead of waiting for someone to redesign it for you.
The short version
AI does not replace media planners; it replaces the assembly half of the job and leaves the judgement. That makes the role concentrate toward fewer, more senior people who own the call. The planner's move is to get closer to the decision and further from the artifact. If you want the full operating picture, start with the Strategy guide.
I got the timing wrong in 2011 and I am still recalibrating in public. If you are a planner working this out on your own team, I would like to hear which half of your week you would hand over first.
Cheers, Chandler