1. The numbers that justify the change of direction
The Northwestern University and Minty study covered 150 e-commerce marketing decision-makers and shows a level of consensus rarely seen today. Out of every additional dollar spent on e-commerce marketing, 63 cents will go to AI shopping tools and to cashback and savings apps. Classic digital advertising — display, classic performance campaigns — is losing ground to channels closer to the purchase process.
The scale of the shift is best seen in the planned mix: the share of cashback and savings apps in the budget is expected to grow from 18% to 30% according to the study. This means merchants treat savings mechanics not as a niche but as a central element of customer contact — especially when AI systems actively search out financial benefits for the customer and bring them into the decision process.
An important methodological caveat: this is a declarative survey, not hard data on fund flows. Yet its strength lies in consensus. When 150 people responsible for budgets say the same thing, the risk that this is „just a fad” is small. The additional money — that is, the money that decides growth rather than maintaining current presence — is planned to go where a real purchase decision is created.
2. Why this is happening: AI becomes the gateway to the customer
The central driver is a changed search-and-purchase behaviour. Customers increasingly use AI systems that look for discounts, cashback offers and cheaper alternatives. In such a model the question „will the brand be visible” loses meaning — it is replaced by the question, whether the brand will be visible in the assisted purchase process: in price comparison, in verifying savings potential, in product selection.
According to the study, two thirds of the 150 surveyed decision-makers see AI as the most important „gateway to the customer” in the coming year. Significantly, more than half of that group explicitly links the use of AI to savings apps. This is evidence that automated product recommendations and the search for price benefits are increasingly being thought of together. For e-commerce marketing this shifts the emphasis: relevance is created where purchase intent, price interest and digital assistance meet in one moment.
The change also concerns the role of the brand. In the classic model the brand built visibility by reaching the broadest possible audiences and repeating the message. In a model where an AI assistant filters, compares and recommends, the brand must be recognisable and unambiguously described in the data — in a different way it „appears” in the answer to a customer query. It is not enough to be present in awareness; you have to be structurally correct in a system that makes the decision for the customer.
3. What this means for budgets and for the structure of spending
The consequences for planning are concrete. The additional money in e-commerce will no longer be used to „be visible to everyone” in the classic media model. Instead, it goes to technologies that give access to the customer exactly when there is a concrete interest in a product, a price or an alternative. The contest is about visibility in digital assistants, discount environments and cashback offers — not about a spot on a publisher's page.
Around the same time, the importance of transactional data is growing. Savings apps and AI shopping tools run on basket and intent data. A retailer that owns first-party data and can plug it into these ecosystems gains an advantage that classic advertising cannot copy. Those who do not organise their data will become dependent on intermediaries controlling access to the customer in the new channel.
In Poland an additional factor is the economics of margins. Retail chains — especially grocery — operate on net margins of about 2–4%. Such a model does not withstand media costs that in classic performance have been rising for years. Cashback and AI-shopping channels work on a different assumption: you pay when the decision really forms, and the savings mechanic can be translated into the cost of acquiring a specific basket. For chains with low margins this is the difference between a „certain” spend and a spend settled against the effect. All the more so because a cashback customer arrives with clear intent — and the customer-with-intent type is always the most expensive to acquire classically and the cheapest in a model where the decision forms close to the purchase.
4. Concrete guidance for e-commerce executives
The change confirmed by the study can be translated into concrete actions. The point is not to radically replace the entire mix overnight but to deliberately shift the centre of gravity to where the purchase decision truly forms — and to prepare the organisation before the market saturates with competition. Here are five recommendations I give executives in this budget year.
- Redesign budget allocation around purchase-close channels. Plan for 2027 a share of savings and AI-shopping mechanics at 25–30% of marginalised spend — matching the 18→30 percent direction from the study. Presence in cashback and AI assistants is no longer a test but a condition of being available in the purchase process.
- Build your own, organised data layer. Before you choose a shopping or savings platform, make sure you have a unified customer profile combining transactional, behavioural and loyalty data. Without it you pay intermediaries for access to your own customers.
- Treat presence in AI shopping as a new performance channel. Define how your brand is seen in assistants that plan purchases and compare prices. This requires structured data on product, prices and promotion — digital merchandising, not just another campaign.
- Measure share in the assisted process, not just impressions and CTR. Classic media metrics fail in a channel where the decision is created in dialogue with AI. Measure availability, share of shelf in assistant results and conversion from recommendations, not just clicks.
- Prepare the organisation for agentic commerce. 63% of incremental budget going to AI shopping is a sign that agent-led purchases will stop being a niche. The sooner your catalogues, promotions and pricing data are understandable to systems, the lower the risk that your assortment drops out of automated selection.
A separate point concerns the organisation. Shifting budget towards AI shopping and cashback is not just a change to the media plan — it is a change of capabilities. It requires a team that understands product data, the terms of integration with savings platforms and the logic of assistants. In practice this means that part of the budget that today goes to campaign execution will move to working on offer structure, data quality and technology partnerships. A retailer that prepares the organisation for this new role in 2026 will enter 2027 with an advantage that cannot be built with budget alone.
Finally: the point is not to abandon classic performance, but to stop treating it as the default destination of every additional euro. In e-commerce, growth is decided today by channels where the decision forms closest to the purchase. The study shows that the industry already knows this — and the data stands behind it. The question is whether your network does too.
5. Conclusion: not a fight for space, but for a place in the process
The Northwestern University and Minty study result is not an anecdote — it is consistent with the broader trends we see in consulting projects. Budget flows where the decision is actually made. And in 2026 and 2027 the decision is increasingly made in dialogue with an AI system that looks for the best price and the largest saving for the customer.
Companies that conclude „if money stops going into classic advertising, there is no point investing” make the opposite mistake. The money does not disappear — it changes channel. It goes to tools that give access to the customer at the most valuable moment of the journey. Whoever organises data, offer structure and presence in assistants and cashback will capture the stream of spend before competitors do.
If you are preparing next year's marketing budget and wondering how to realistically shift spend from classic advertising to purchase-close and AI channels — I help executives design such a transformation. Let's talk.