AI Can Analyse the Campaign. Does Your Junior Understand the Client?
Managing AI output takes more than campaign data. Juniors need to understand the client, and seniors need to bring them into the room.
When a paid media campaign underperforms, there is rarely a shortage of explanations.
The audience was too small. The creative was weak. The platform needed more time to learn. The leads were cheap, but the sales team did not convert them.
Any of those explanations could be true. In media, you can caveat almost anything.
When you are facing a disappointed client, the pressure to explain the result can easily become pressure to defend yourself. The difficult part is knowing whether your explanation actually accounts for what happened.
That takes an understanding of how the campaign fits together, and of the business behind it.
AI can produce a convincing explanation quickly. A junior can arrive with a structured analysis, supporting numbers and recommendations without fully understanding whether those recommendations make sense.
That changes what we need to teach them.
Where does judgment come from now?
Nate B. Jones recently raised a question that stayed with me: if AI takes over the work through which junior people gain experience, how do they develop the judgment needed to become senior?
He explored whether learning to delegate to and supervise agents could become part of that development. In paid media, I think that is already becoming the nature of junior work: checking, challenging and managing AI output.
My own judgment came from a combination of running campaigns, making mistakes, being challenged by seniors and answering difficult client questions. Over time, those experiences helped me understand how the different elements of a campaign connect.
The question for agencies is how to develop that understanding when AI handles more of the analysis.
Consider four familiar explanations for disappointing performance.
“The audience was too small.”
Perhaps it was. But did audience size constrain delivery, or did the budget, exclusions, bidding approach or creative limit what the campaign could achieve? Naming a possible cause is the beginning of the investigation.
“The leads were cheap, but sales did not convert them.”
That might be a sales follow-up problem. It might also be poor lead quality, a mismatch between the advertising promise and the offer, or an optimisation setup that rewards form submissions without regard to what happens next.
Without access to the CRM or reliable client feedback, how much can we actually conclude?
“The platform needs more time to learn.”
Sometimes it does. But additional time will not fix broken tracking or an optimisation event that rewards the wrong behaviour. A junior needs to understand what is being learned and whether the campaign is receiving a useful signal.
“The creative was not strong enough.”
Possibly. But where did performance break down? If people clicked and then dropped out, the landing page, offer or enquiry process also deserves attention.
These are hypothetical examples, but the distinction matters. AI can list the possible reasons. The person managing its output needs to decide which deserve investigation, what evidence would support them and what remains unknown.
The campaign data is only part of the picture
Some of the most important campaign context sits outside the advertising platforms.
A client may take too long to follow up on leads. They may decline to give the agency CRM access, leaving the team with limited visibility into quality and sales outcomes.
In some government or banking engagements, the client may not approve website pixels. That changes what the team can measure and how it can optimise.
The client may also have limited resources to produce creative, or an approval process that moves too slowly to support the testing plan.
An AI recommendation to increase creative testing may be reasonable on paper. Its usefulness depends on whether the client can produce, approve and launch those assets.
This is where client conversations matter. They reveal what the business is struggling with, what it can realistically change and why an apparently straightforward recommendation may be difficult to execute.
We should expect a junior to have relevant context. We should also expect them to recognise when something is missing and ask the account director or client for clarification.
They cannot judge the recommendation properly if they do not understand the conditions under which it must work.
Checking AI requires more than another AI
I want juniors to challenge AI output. Cross-checking with another model can help. So can using a separate checking agent within the same platform.
But agreement between models is not enough. They may be working from the same incomplete information.
The most important check remains: does this make sense, given what we know about the campaign and the client?
A junior should be able to explain what they accepted, what they questioned and why. They should know which conclusions are supported by evidence and which depend on assumptions.
I also want to ask what I think of as “outside the sandbox” questions: questions that step beyond the information and reasoning contained in the AI’s answer.
Would you still recommend this if the sales team could not handle more leads?
What changes if we cannot connect those enquiries to sales?
Can the client execute the proposed creative plan?
And, most importantly, are we still delivering what the client originally asked for?
During a long-running task, AI can lose track of the original objective. Work that began with a brief to generate qualified enquiries can drift toward improving the cost per form submission. Each step may appear reasonable, while the overall analysis moves away from the business need.
Managing AI output means catching that drift before the recommendation reaches the client.
Seniors have to bring juniors into the room
If this is the work we expect juniors to do, senior people have a responsibility to prepare them for it.
We need to bring them into client conversations. They need to hear the questions, understand the pain points and see how business constraints affect campaign decisions.
A dashboard can show what happened within the limits of its measurement. A client conversation may explain why the business could not act on the demand, why an approval stalled or why a different outcome matters more than the metric being reported.
Those conversations should become a deliberate part of junior development.
Seniors can then use the work itself to build judgment: ask the junior to explain the recommendation, challenge an assumption, introduce missing context and discuss how that changes the decision.
That gives us a more useful way to assess their progress. Can they identify what the AI has missed? Can they defend a recommendation with evidence? Do they recognise when they need help? Have they kept the client’s objective in view?
AI can take on more of the campaign analysis. The junior’s responsibility is to make sure the output is sound, relevant and useful.
The senior’s responsibility is to give them access to the business understanding that makes that possible.