WHY REMI
Marketing intelligence,
not general intelligence
What REMI is, what it's trained on and how it works.
01 — The problem
Why generalist LLMs get marketing advice wrong
General-purpose LLMs are trained on public content about marketing - awards submissions, HBR case studies, blog commentary. They don’t have access to end-to-end marketing data, which is normally held across many agencies (strategy, brand, market research, advertising, sponsorship, media, digital, shopper marketing and PR to name a few), not to mention the walled-garden Google, Meta, TikTok ecosystems at the performance end of the spectrum.
A true marketing LLM would need to see all this data to understand how a decision at one end makes an impact on a commercial outcome at the other. In its absence, their advice is as generic as it is confident. The evidence on which it is built just isn’t up to the job.
02 — the solution
MSE: the AI-ready marketing approach
Marketing System Efficiency - or MSE - is the 27-lever framework that scores how efficiently a business is running its marketing system. REMI codifies every aspect of marketing, and then runs a diagnosis across any named competitor-set to create thousands of data points. It reveals how efficiently a business is deploying its human and financial marketing capital and reports not just for one company, but benchmarks against their competitors.
REMI has found the average marketing efficiency score to be circa 70% (lower for B2B, higher for B2C) which means that on average 30% of an organisation's marketing is wasted. MSE is the Marketing OS for everything REMI does: every diagnostic, product and intelligence feed runs on the same underlying framework and allows REMI to think with the collective intelligence of a thousand CMOs.
03 — training data
REMI training
REMI wasn't trained on a scrape of the public internet. It was built and trained by marketers and technologists with the collective experience of hundreds of years, on the same 27-lever MSE framework.
REMI was built on data from 40 business turnarounds, and then trained and tested on more than 5,000 businesses owned by Private Equity across the UK and US. PE is an audience unusually well versed in numbers and evidence, not something than can traditionally be said for the entire marketing universe. REMI was designed to produce investment-grade marketing diagnostics, and a disproportionate amount of time and attention is devoted to ensuring the fidelity of the data. When REMI outputs are used to inform investment decisions or to drive huge scale MarTech campaigns, being right matters.
04 — data sources
5,500 usable points of data
REMI doesn’t use internal company data, and was conceived that way to avoid the traditional self-comparison or assumed success rate of old approaches. Marketing as a practice is designed to be seen from outside an organisation, and indeed awareness is normally at the top of any CMO's KPI list. REMI interrogates 153 different data sources containing around 5,500 usable data points. Each piece of REMI diagnostic analysis will apply anywhere between 2,000 and 4,000 of those to each company, depending on the type of business.
To that MSE data set, REMI can then import client proprietary data to enhance the effectiveness of marketing products. For example, a REMI Market Segmentation can be transformed into a Customer Segmentation in a matter of minutes with the addition of sales data.
05 — marketing archetypes
B2B, B2C, and a dozen points in between
In addition to the 27-lever framework, the 5,500 datapoints, the complex scoring algorithms and the contextual framework, REMI recognises that different businesses run different marketing models that suit their commercial outcome. Beneath the simplistic approach of B2B versus B2C lie a dozen distinct marketing archetypes, each running marketing systems with important distinctions. In many cases, the marketing objectives for two digital businesses, one selling to consumers and another to business, may have more in common in how they Go-To-Market than two enterprise-facing companies.
This intelligence is baked into REMI. Without it, analysis and recommendations around growth prospects would be dangerously generic. Making any decision based on REMI intelligence will always be a real-world marketing recommendation, not a generic AI hunch.
06 — Proof
Fact-based marketing for the AI age
Part of REMI's testing phase involved rigorous linkage between the 27 levers of growth and the real world outcomes that they predicted or instructed. Much of that was proprietary data which is used at a training layer, but can’t be published. REMI ran MSE across ten US businesses that publish their marketing data as part of the SEC submissions. Using retrospective MSE analysis we were able to track marketing-led growth MSE with a coefficient of 0.78%. At the other end of the scale, early tests creating B2C-style video content with and without REMI intelligence led to an improvement in engagement levels of more than 100% versus the controls. Between these two extremes REMI has been proven time and time again against a whole host of old-state marketing tools, and outperforms all General AI in marketing experiments.
REMI welcomes project work with PE and brand owners to quantify the business advantages of adopting MSE as part of a value creation strategy. Its sister organisation Bridge F61 regularly undertakes such consultancy assignments.
REMI Intelligence: get in touch
Building on Jasper, Copy AI, Optimizely or your own MarTech stack? Talk to us about licensing the REMI Intelligence layer.
Get in touch
© 2026 Bridge F61