The AI bill is bigger than most budgets assumed
New research from Modern Retail, surveying 30 agency professionals in the third quarter of 2026 alongside a focus group of senior media executives from Go Fish Digital, Horizon Media, Novus, Tinuiti, and UM, puts real numbers behind a cost conversation retail brands have mostly been having anecdotally. About 31% of agencies now spend more than $10,000 a month on AI tools, a figure drawn from Ramp research cited in the report. Of those agencies, 31% said the costs came in higher than expected, 45% said about as expected, and 24% said somewhat lower.
That split matters for any retail brand funding an agency relationship, because those AI costs do not stay fully internal to the agency. They show up eventually in retainer renewals, scope negotiations, or quiet reductions in the human hours a brand actually gets for its spend. A brand that has not asked its agency partners directly what their AI tooling costs look like, and how those costs are changing, is negotiating the next contract cycle with less information than the agency has.
The efficiency gains are real, but concentrated
The upside case is also documented: 70% of agencies said AI has saved costs through more efficient task execution, well ahead of the 17% who attributed savings specifically to replacing roles. That gap is worth sitting with. It suggests the dominant pattern so far is AI making existing teams faster rather than AI eliminating headcount outright, a distinction that matters both for workforce planning conversations and for how a retail brand should read an agency's pitch that AI adoption will lower its fees.
The report also flags that agencies are actively tracking token usage per employee and warns that efficiency gains may eventually plateau as the easy automation wins get captured. Agencies are consolidating overlapping internal AI tools that different internal teams built independently, which is the agency-side version of a problem most retail technology organizations are also fighting: AI tool sprawl accumulating faster than anyone is auditing it. That plateauing warning should temper any budget plan that assumes this year's efficiency curve simply continues into next year unchanged.
Budgets are already moving toward AI search
The most consequential number for retail marketing budgets going into 2027 is this: 77% of agencies plan to increase client budgets for AI search and generative engine optimization over the next twelve months. Of those planning increases, 40% expect the money to come from general marketing budgets and 20% from traditional search budgets. Go Fish Digital's Dweck estimated that roughly 60% of clicks have disappeared, concentrated in organic search, forcing advertisers to spend more just to regain the traffic volume they used to get for free.
That reallocation is a direct consequence of the same AI traffic shift reshaping retail checkout and discovery more broadly. If a growing share of product discovery happens inside a conversational AI surface rather than a traditional search results page, the marketing spend has to follow the traffic, which means budgets earmarked for conventional SEO and paid search are the ones under pressure to shrink. Retail brands should expect this conversation to arrive in their own budget planning cycles whether or not their agency has raised it yet.
Agencies are turning AI into a product, not just a tool
59% of surveyed agencies are monetizing AI directly, through new products rather than treating it purely as an internal cost center. Roughly a third offer strategic roadmaps or audits of a client's AI products as a billable service, and 31% charge an explicit add-on fee for AI-enabled work. Named examples include AI-generated avatars, feed optimization tools built for conversational commerce, and synthetic audiences used to simulate focus groups before a campaign launches. Agentic buying itself remains unsettled: Tinuiti is testing AI agents for media planning and audience discovery, but the report describes the category as still unclear rather than mature.
This has a direct implication for how retail brands should structure agency contracts going forward. An AI audit or roadmap that used to be bundled into a standard strategy engagement is increasingly billed separately, and a brand that does not ask explicitly what counts as a billable AI add-on risks discovering the line item after the invoice arrives rather than before the statement of work is signed. Getting that definition in writing during the next renewal cycle costs nothing and avoids a predictable dispute later.
What this means for the brand side of the relationship
Retail CMOs and CTOs reading this data should take two concrete actions. First, ask every agency partner for the same breakdown Modern Retail collected: current monthly AI tool spend, whether that spend is trending above or below internal projections, and which of that cost is being passed through versus absorbed. Second, treat the 77% planning to shift budget toward AI search and GEO as a signal to start that conversation internally now, rather than waiting for an agency to initiate a mid-contract reallocation request once the organic traffic decline shows up in the brand's own analytics.
The deeper lesson sits in the tool-sprawl warning buried in the report. Agencies are consolidating overlapping AI tools their own teams built independently, the same failure mode retail technology organizations face internally when individual teams adopt AI point solutions without central coordination. A retail brand that has not audited its own internal AI tool sprawl is one step behind the exact problem its own agency just admitted to having, and is paying twice for it: once in its own unconsolidated stack, and again in agency fees that reflect the agency's version of the same inefficiency.
