The retail marketing industry is awash with artificial intelligence (AI) hype – much of it vague, speculative or fear-driven.

According to WARC, 59% of marketers now worry about disruption from AI, up from 27% in 2023.

With pressures to balance brand and performance, manage campaigns across multiple channels and prove ROI, retail marketers are being challenged not just to understand AI, but to make it work within existing workflows without slowing operations.

But what does implementation of the tech actually look like for retailers?

From creative development and campaign testing to market research, stock management and decision support, we asked retail marketing leaders and experts to share concrete examples of how AI should be applied, and what measurable impact it has had on efficiency, costs and results.

Workflows that have been automated and the measurable change they have delivered

The biggest AI marketing gains for retailers have come from using the technology as a practical co-pilot.

This much is true for Giulio Beltramo, head of marketing at multinational retailer and brand owner One Retail Group, who trains models to understand what works for his marketing team, why it works and the context in which it works.

“AI has become an extensive valuable companion to bounce ideas off, validate hypotheses against real-world data, research at scale in a cost-effective way and get back a whole lot of information that can prompt further questioning and discovery,” he says.

One Retail Group’s established AI workflows also feed into its creative development teams. While imagination is the limit with their AI stack, Beltramo notes that implementation comes with effort.

“Prompting campaign-level assets or assets we feel comfortable listing on our product pages takes hours of iterations and extensive prompting,” he adds. “Further to this, using two or three tools is often necessary when delivering a final asset.”

The payoff, however, is substantial.

AI allows the team to experiment with lower risk, develop creative variants faster, and ultimately reduce costs. “Over the past few months, we have been able to run variants on creative, try new looks and feels, and truly define the brand and visual direction we want to go in,” Beltramo says.

“This was done at a lower risk, lower cost, and we were able to develop and test more variants quicker, resulting in better performing creatives, reduced cost per click (CPCs) and cost per action (CPAs), and clearer insight into brand direction through metrics like view-through rates and engagement.”

For Fran Bridgewater, marketing director at Drinks Network, which works with a host of wine retailers, the clearest gains in AI implementation are in marketing and communications. AI is saving time, sharpening targeting and helping safeguard authenticity.

“AI drafts product descriptions, social media captions, email copy, and regional campaign variants,” Bridgewater says. “Focused on speed and consistency rather than replacement, these tools cut repetitive work and free small teams for higher-value trade, brand and strategic activities.”

With AI now in the fold, production and campaign turnaround is 30-40% faster for UK boutique wine producers associated with Drinks Network. “It’s enabled more consistent output and emerging conversion gains from clearer messaging and better audience targeting,” says Bridgewater.

According to Phil McMahon, retail and engagement strategist at Really Good Culture, the marketers most likely to benefit from AI are those using agentic tools to remove friction for shoppers navigating AI-powered search.

“At the end of the day, retail marketing is about driving conversion and improving availability and reducing friction are effective ways to do that,” McMahon says.

“These automations also improve availability for shoppers, which is a key driver of satisfaction and loyalty. All of this is of growing importance in an increasingly omnichannel retail landscape,” he says.

The guardrails that are essential for keeping brand risk low

Using AI for research purposes is a must for retail marketers, but Beltramo cautions that generative AI often tells you exactly what you want to hear.

For him and One Retail Group, the research is only as good as the interrogation you do of it.

“Get a tool that shares the sources of its research and actually check them,” Beltramo adds.

“Ask it ‘why have you given me this answer’ or ‘rate this response out of 10 and iterate on your answer’. You have to be vigilant and assume a bias.”

With regards to using AI to aid creative, strong brand and visual guidelines are required to ensure that the vision of the brand does not get lost in the prompting phase.

“When creating using AI you almost have three brains involved,” Beltramo says.

“The brain of the campaign creator, the brain of the person executing or prompting, and then AI’s brain. If you say, ‘create a picture of a dog’, each three of those brains are going to imagine a different dog.”

Bridgewater echoes this sentiment, admitting that AI-assisted outputs must be rigorously reviewed.

“Retailers have to protect brand voice, provenance, and technical accuracy, especially under strict appellation rules,” she says.

“With regards to branding and copywriting, AI doesn’t always deliver, fully automated tasting notes often lack personality and sense of place.”

Human oversight is paramount, agrees McMahon, as AI systems do not guarantee 100% accuracy. Automation should be reserved for flagging issues as opposed to being responsible for judgment calls.

“Despite the hype, it’s still early days for AI in retail and new problems and failures will arise as the technologies scale,” he adds.

“AI is the hammer, not the blacksmith. Human oversight of AI automation will mitigate blind spots and help maximise its benefits, and this should ultimately be reflected in shopper feedback and the retailer’s bottom line.

Lessons learnt along the way

At One Retail Group, Beltramo is clear that his team is still learning how to work with AI and where it can add the most value. He likens AI to being the coach or assistant to a sprinter, not the sprinter itself.

“The marketer in this situation is the runner,” Beltramo adds.

“I would always recommend that you as a marketer still have to do the actual hard work in order to get AI to take your work to the next level. It will get you that much further, that much quicker, but you still need to put in the work.”

Similarly, for Bridgewater the lesson is clear: AI works best as a trusted assistant.

“Start with bottlenecks, prioritise small wins, keep humans in the loop, and track outcomes rigorously,” she says.

“The retailers we work with that are embedding AI across content, data, decision-making and customer experience are already turning experimentation into a tangible competitive advantage.”

Click here to sign up to Retail Gazette‘s free daily email newsletter