Marissa Jarratt distinguishes between AI in general, which her marketing technology has used for years across customer data, loyalty, CRM and media optimization, and generative AI, which the team has been getting a handle on more recently.
Generative AI is being applied in three areas of marketing: insight, creative development and production, and media planning and execution.
Key findings
- On the insight side, the company runs a proprietary consumer research panel of about 300,000 customers who take part in qualitative and quantitative research and provide their own data on attitudes and beliefs.
- The value of generative AI there is scale rather than novelty. Open-ended questions can be put to thousands of customers rather than ten, and the tool synthesizes the responses into themes.
- In creative work, generative AI is used for ideation rather than finished output. A team might build a mood board for a product design or explore campaign messages, and the result is not creative that gets used.
- Production is where the volume problem sits. Prices and product ranges differ by market, so many versions of each asset are needed, and the question is the right combination of offshore partners and generative AI.
- A third production case is in-store radio across 13,000 stores, part of the retail media network, where a partner platform generates the spots that play in store.
- Media planning is the gap. Execution already has various AI solutions in place, while on the planning side she would like to feed historical plans, current goals and budget into a tool and have it prepare scenarios for the experts to refine.
Who marketing works with on this
The first partner is the internal team focused on generative AI solutions suitable for enterprise use, which have been through an evaluation process that allows proprietary data to be put into them.
The second is legal, privacy and compliance, who oversee risk and governance of company assets including data. Marketing works with them to explain a new tool and agree how to run a proof of concept.
External partners are treated as a way to get smarter, and that learning often happens through working alongside each other rather than in workshops.
Building the capability
Capability building starts in the annual planning cycle. The leadership team sets business objectives and then asks what new capabilities the department needs to achieve them, with generative AI a large part of that.
The requirement ranges from the basic to the strategic: how to write a prompt, how to think about AI as part of a workflow, where to look for opportunities to integrate it, and what questions to ask.
One approach is broad: everyone needs to be conversant in the basics, supported by training, shared reading, and a culture of information sharing on the team’s internal channels.
The other is specific. Each leader identifies the use cases and pilots in their own area, and the person chosen to lead one is sometimes picked as a development opportunity rather than because it falls in their job.
The team volunteers for company-wide pilots as well, including an enterprise chatbot and tests run by the data and AI team.
Why marketing volunteers
The reason she gives is a view of the function’s role. Marketing sees itself as a change agent not only for consumer behavior but inside the company, and wants to be a thought partner and implementation partner on large initiatives.
The specific contribution she names is empathy. Marketing understands it deeply and applies it to consumers, but the same skill can be applied to employee experience and other audiences, helping partners think more broadly about what they are building.
Marketing to the assistant, not just the customer
The question occupying the department is who marketing is actually addressing: the end consumer, or the AI platforms that will shape what that consumer is recommended.
Her expectation is that both matter. Consumers still need to be marketed to, but AI assistants and platforms become the influencer in between, and a strategy for reaching them is needed.
The behavior she anticipates is simple to picture: instead of searching for a restaurant or working out where to stop on the way home, a person asks their assistant and takes the recommendation.
What the technology cannot do
Her caution about generative AI is that it is only as good as the information inside it, which makes it a backward-looking way of creating something for the future, and it is logical and rules-based by nature.
Marketing’s magic, in her account, is creativity that is sometimes not logical: taking two entirely unrelated ideas and putting them together to make something nobody has thought of before.
Her position is not a rejection. It is a case for using AI where it is strong while keeping human guidance, and she is looking forward to being freed from the detail of spreadsheets and data checking to think more creatively and strategically.
The question she wants answered
The open problem is money. She does not have a clear view of how to allocate budget for generative AI solutions, tools and external partners, or how that should change from year to year.
The complication is that generative AI is not free. It requires investment in technology, in data and in the people to manage and use it, so the offset against existing spend is unclear.
The same question applies to agencies. She expects a change in their labor and business model within roughly a two year window, and wants to know what the right path looks like, along with how legacy benchmarks such as working against non-working spend should change.
Her observation about agencies is that maturity varies, and that the larger groups are mostly building generative AI consulting and technology as add-ons to existing scopes of work rather than using it to change how they deliver. Her view is that it should be both.
Scope and limitations
The source is an automatically produced meeting transcript, so some passages are garbled, and the account describes one company’s practice at one point in time.
Source
This page summarises Transcript: Interview with Marissa Jarratt on AI in Marketing at 7-Eleven, by Mariana Peneva, Marissa Jarratt, December 19, 2024.