Boots Turns the Ad Library Into a Learning System

Boots’ latest measurement work is more interesting than another claim that AI can predict a winning advert.

The retailer has analysed more than 3,500 creative assets used across Meta and YouTube with CreativeX and DAIVID, connecting creative-quality, attention and emotional-response signals with campaign outcomes. Alongside a newly launched brand platform, *Give It Some Boots*, it offers a useful mini case study in a marketing problem that is usually hidden by topline reporting: not every asset in a campaign is doing an equal job.

The important shift is from evaluating a campaign after the fact to creating a practical feedback system for the large, uneven library of work that modern media plans require.

The average campaign result conceals too much

A full-funnel campaign can contain films, cut-downs, creator content, retail activation, social adaptations and multiple versions designed for different placements. A respectable aggregate result may therefore conceal a great deal of waste: weak executions can be carried by a small number of strong ones, while useful creative lessons disappear into a blended dashboard.

Boots’ pilot attempts to give teams a common way to inspect that variation. DAIVID says its creative-intelligence signals showed a 0.89 correlation with campaign outcomes in the Boots analysis. Correlation is not proof that a particular visual device, emotional cue or branding treatment caused a result. Media conditions, audiences, frequency and offer all remain material.

But that is not the only value of the exercise. It gives creative and media teams a more disciplined starting point for asking why one asset received attention, was remembered, or converted an exposure into a useful commercial response while another did not.

That matters because digital optimisation has often been better at shifting budget between placements than improving the work placed inside them.

A brand platform needs asset-level evidence

The timing is notable. Boots launched *Give It Some Boots* in September as a broad new platform spanning health, beauty, wellness and pharmacy, with TV, cinema, digital out of home, social, stores and online activity. Its ambition is expansive: to make the brand feel present in life’s more joyful and difficult moments.

That kind of proposition cannot rely on a strong hero film alone. It needs to retain its meaning when translated into shorter, more functional or more targeted work. The risk is familiar: the campaign idea is warmly received at launch, while its numerous adaptations become generic retail communications that could belong to almost anyone.

A creative audit can help detect this drift. It can show where branding is too faint, where an idea is not clear enough at a given length, or where an execution is attracting attention without creating a distinctive Boots memory. Used properly, those findings should inform the next brief and edit, rather than simply rank assets after money has been spent.

Treat scores as prompts, not verdicts

There is a danger in making any creative score too authoritative. The most effective work will not always resemble past winners, and an optimisation model can easily reward familiar patterns at the expense of originality. Nor should a system built to diagnose digital video dictate a brand’s broader creative ambition.

The more useful operating model is modest. First, establish whether each asset makes the intended brand and message recognisable. Then connect its characteristics to outcomes in the specific media context where it ran. Finally, turn the finding into a testable creative hypothesis: not “make everything like the top scorer”, but “does clearer early branding improve recall here without weakening the story?”

Boots’ approach suggests that creative effectiveness is becoming less of an annual awards-case exercise and more of an everyday learning discipline. For brands producing hundreds of assets, that is a more commercially useful ambition than simply making more content faster.