The real work of in-housing

When internal teams start to think about in-housing they generally start by focusing on what areas of digital media go where. But, as AI and AI-led media are now so pervasive, they should be thinking about what work is required, and who should do it. Does it sit on the agency side or do we bring certain parts into the business, and what does that mean in terms of headcount and savings against how we currently operate?

While this sets the stage for current industry shifts, it overlooks the most critical transformation: the very nature of paid media has evolved, and the responsibilities and practices surrounding it must change as well.

Digital media execution today is largely handled by AI-led systems, with campaign types that sit within Meta, TikTok and Google platforms handling the majority of retailers’ marketing capital.

These platforms continually deploy, optimise and pace campaigns using real-time signals. With the right operating model, they can align themselves to real-time customer demand, but they no longer benefit from the old-school manual optimisation approach.

10 years ago when I worked in-house in retail organisations, my confidence in agencies grew with the more hours, people and experience they had - under the belief that my media performance and sophistication would benefit from the same economies of scale. But today with AI taking on the manual work, an increase in practitioners doesn’t correlate with better results, it only means more execution costs.

The new paid media operating model

What’s important now is decision quality.

As media execution and expertise becomes fully driven by data and technology, the responsibility should move upwards within the retail organisation. The important questions have changed from “how are we going to alter campaigns this week”, or “have we got the right categories and the products in campaigns to support a promotional period”, to “who is involved in setting objectives”,” who defines when there is acceptable risk”, and “who owns the performance guardrails that the media behaves within”.

Measurement has to be the first area that needs an overhaul. Today, digital media directly affects revenue, margin and profit, so confidence in measurement and performance are vital components to managing the business. Evidence of media outcomes need to connect and calibrate platform signals, alignment between Google Ads, GA4 and MMM, to ensure alignment with commercial performance, and have to be understood consistently across senior levels in marketing, trading, finance and the board. Reporting alone is not sufficient, it is an absolute requirement to collaborate through governance across these teams.

AI-led media certainly doesn’t remove the need for oversight, however it does change its focus. Instead of frequent optimisations, retailers need a layer that translates commercial intent into system inputs and reviews outcomes in the context of demand, pricing, retail practices and market conditions. Without this framework, media execution continues aimlessly while leadership debates the results. This lack of consensus stalls decisions, the business remains trapped by old behaviors and outcomes suffer.

The core media team becomes leaner as operational tasks are automated by AI. Instead of manual campaign optimisation, teams shift their focus to higher-value work: setting objectives, navigating business and market constraints, and defining the retailer’s appetite for risk.

These new media teams have some interesting challenges to focus on, such as what gets prioritised in media deployment so that the business can learn, scale and hit its objectives. When it comes to prioritisation they can focus on whether margin is prioritised over topline growth, whether returning customers should be the focus or if the business needs to acquire new customers. They also need to consider whether there are parts of the range that are simply not performing well and no longer receive ongoing support because they are declining in their lifecycle and attractiveness in the market. The business needs these types of insights just as much as it needs to know that newer categories are going to see greater visibility with an incremental learning budget.

A team’s value is no longer measured by manual output within platforms like Google and Meta. Instead, the focus of an in-house function must shift toward deciding what the business wants media to prioritise, how success is measured and when corrective action should be taken. In-housing is not just about replicating agency structures internally. It is about recognising that as technology absorbs more of the “doing”, the organisation must strengthen its capability around the decisions it's going to make. It often falters when responsibility is relocated without being clarified; the execution changes hands, but ownership of outcomes remains blurred. Progress accelerates when retailers accept that systems deploy media, and they own the strategy.

This also means analytics shifts from just reporting to evidencing what will be used to make commercial decisions. That means answering questions such as: what commercial outcome are we trying to influence? What would success look like in margin terms, not just ROAS? What level of performance volatility is acceptable? What is incremental versus what would have happened anyway?

Over time, this changes how the retailer behaves. Media performance moves from an 'after-the-fact' report to an active part of daily operations. Once that responsibility is clear, the next step is purely practical: deciding what actually needs to be moved in-house and what doesn’t.

This is a lot of change for media management, so what does the paid media team now need to look like? See the next in our series