Rakuten Advertising pushes probabilistic measurement and LLM citations

By Amara Osei

Seven months after asking performance marketers at DealMaker in March to assess Rakuten Advertising more holistically, EMEA SVP Nick Fletcher said the company was working with more publishers that drive performance inside large language models.

The interview offers no baseline publisher count, client roster or measured lift. It gives no spend, conversion, EPC, commission, payout, fee, cookie-window, attribution-window or sample-size figures. Treat its claims as vendor assertions, not results.

What changed after the March DealMaker call?

Fletcher’s March appeal centered on probabilistic measurement and the channel’s wider impact. He said progress followed, including on the brand side. Brands working with Rakuten were changing how they measure. The source gives no year for the event.

“The brands that we work with and we talk to are evolving their measurement,” Fletcher said. “I think they understand that deterministic measurement isn't necessarily good enough for the channel and undervalues the channel a lot of the time.”

He expects advanced measurement to unlock more opportunities and spend. Fletcher also claimed the channel’s measurement arsenal is arguably wider than that of any other media channel. The excerpt supplies no comparison set, methodology or forecast behind either claim.

How are LLMs changing partnerships?

Fletcher said AI creates an opportunity for Rakuten, including from the brand side. More brands now approach the channel to understand how they can become visible inside LLMs.

The problem, as he described it, is control. Brands have invested significant budgets in media they control; they do not control what an LLM cites or what people say about them. According to Fletcher, LLMs increasingly trust partner content over brands’ own statements about their products and services.

That is a consequential assertion, not a disclosed test. The source names no LLM, prompt set, geography, citation metric or study design. It also does not show whether an LLM mention leads to a site visit, lead or sale.

What is Rakuten doing with publisher data?

Rakuten now works with more publishers that Fletcher says drive performance within LLMs. The company is identifying which publishers drive that performance and citations, feeding the data back to clients and publishers, and starting to incentivize publishers to produce more visibility and the desired behavior.

He characterized partnership marketing’s opportunity as bigger than ever and rapidly expanding. The account does not define “performance within LLMs.” It could refer to citations, referral traffic or an outcome attributed elsewhere; the source does not say.

Is this still a lower-funnel channel?

Fletcher described the belief that Rakuten is lower funnel as a persistent misconception. His defense rests on the mix of publishers and media, plus measurement beyond last-click optimization.

“I think a well-run performance marketing program will always be full funnel and the evolution of measurement in the space is allowing us more than ever to bring different types of publishers, different types of media into the space working on a performance basis but not necessarily always optimising towards a last-click-wins basis,” he said.

Fletcher concluded that Rakuten is a full-funnel marketing channel and that delivering that role would leave the industry positioned to keep growing. His statement does not quantify any upper-funnel outcome, brand lift or conversion change.

What evidence and terms are missing?

Rakuten presents the route from publisher behavior to LLM visibility to client spending, but the source stops before the commercial mechanics:

• No publisher or client names identify scale.
• No vertical or client segment shows where the approach applies.
• No CPL, CPA, rev-share or hybrid terms explain the publisher incentives.
• No fee structure, cookie window or attribution window links a citation to a payable outcome.
• No sample size, benchmark or lift demonstrates that probabilistic measurement improved results.

The item also leaves two compliance questions open: how publisher content meets FTC disclosure obligations, and whether LLM citation activity falls within existing program terms. Neither is answered here.

The article gives no starting publisher count or later count, so readers cannot test the scale claim. The next test is whether Rakuten can identify participating publishers, define LLM performance, disclose attribution and commercial terms, then show whether AI visibility changes brand outcomes or conversion economics. Until then, seven months of claimed progress marks a direction of travel, not proof.