Market And Competitive Intelligence Platforms

On paper this is the category generative AI should have eliminated.

Market and competitive intelligence is, reduced to its mechanics, reading a large volume of sources and summarising what matters. That is the single task large language models are best at, and it became cheap in 2023.

Instead Forrester's criteria count went from twenty four in its Q2 2023 evaluation to thirty one in Q4 2024, and the Leader took maximum scores in seventeen of them.

Understanding why the category survived tells you what you are actually buying, and it is not summarisation.

What these platforms do

The work divides into monitoring, analysis, and distribution.

Monitoring means watching a defined set of sources continuously for anything relevant to your markets, competitors, customers, regulators, and technologies. Not searching when someone remembers to, but standing surveillance.

Analysis means turning that stream into something with meaning: what changed, why it matters, what it implies. Increasingly this includes populating structured frameworks, competitive profiles, SWOT analyses, market maps, rather than producing prose.

Distribution means getting the resulting intelligence to the people who make decisions, in a form they will actually read, at a moment when it is still useful.

The third is where most intelligence functions fail, and it has nothing to do with the quality of the analysis.

Three editions, and a market that reshaped

The Forrester New Wave: Market And Competitive Intelligence Platforms, Q4 2019, authored by Cinny Little, evaluated twelve providers against ten criteria: Bertin IT, Cipher, Comintelli, Crayon, Digimind, Klue, Kompyte, Market Logic, M-Brain, Northern Light, Sharpr, and Wide Narrow.

That it was a New Wave matters. Forrester reserves that format for emerging markets, so as recently as 2019 this was not considered an established category.

The Q2 2023 Wave evaluated fourteen vendors against twenty four criteria across current offering, strategy, and market presence. M-Brain placed as a Leader with the highest ratings in both current offering and strategy, credited for generative AI capability, more than two hundred thousand global data sources, and what Forrester described as the only natural language query capability in the market at that point. AlphaSense was recognised for AI and search capability, vision, innovation, and coverage across proprietary, paywalled, and internal sources.

The Q4 2024 Wave evaluated eleven providers against thirty one criteria, with customer feedback replacing market presence as the third dimension.

Valona Intelligence, the renamed M-Brain, placed as a Leader for the second consecutive evaluation, taking the highest possible score in seventeen of the thirty one criteria including advanced generative AI, other AI capabilities, global reach and delivery, publicly available data sources, proprietary and paywalled sources, and vision.

Twelve providers, then fourteen, then eleven. Several 2019 names have since been acquired or absorbed, which is the normal trajectory for a market that consolidated once buyers started treating it as infrastructure rather than as a tool one team used.

The moat is access, not intelligence

Look at what Valona took maximum scores in and the answer to the opening question is sitting there: publicly available data sources, and proprietary and paywalled sources.

A general-purpose AI assistant can summarise anything you give it. What it cannot do is read a licensed industry publication it has no subscription to, a regional trade journal behind a paywall, a regulatory filing in a language and jurisdiction it does not index, or a research report the publisher sells for four thousand euros.

Valona's own description of its footprint puts the scale of that access at one hundred and fifty countries, one hundred and fifteen languages, more than three hundred paywalled sources, and over three hundred content partners.

Licensing that content, negotiating the rights to process and redistribute summaries of it, and maintaining those relationships across hundreds of publishers is a decade of business development work. It is not replicable by a better model.

The same logic explains AlphaSense's position, built around proprietary and paywalled content alongside a buyer's own internal documents.

So the category's defensibility rests on something unglamorous. Anyone can summarise. Very few can legally read the things worth summarising, at scale, in the languages where they are published.

There is a second-order version of this argument that matters more each year. As general AI assistants become the default way people ask questions, the answers they give are constrained by what they can access, which is overwhelmingly the open web. An intelligence function relying on those tools is systematically blind to exactly the sources that are worth paying for, and it will not notice, because the answers will be fluent.

Two categories wearing one label

The vendor lists across these evaluations contain two genuinely different products, and buyers should know which one they need.

Competitive intelligence for sales enablement is the first. The output is battlecards, competitor profiles, objection handling, and win-loss analysis, delivered into the CRM or the sales enablement platform. The consumer is a seller in a live deal, the time horizon is this week, and the measure is whether win rates against named competitors improve. Klue and Crayon built their businesses here.

Market intelligence for strategy is the second. The output is trend analysis, regulatory tracking, market entry assessment, technology monitoring, and geopolitical risk. The consumer is a strategy, product, or executive function, the time horizon is quarters to years, and the measure is whether the organisation was prepared for something. AlphaSense, Valona, and Northern Light sit here.

Both are legitimate and they need different things. A sales enablement tool needs speed, brevity, and integration into the seller's workflow. A strategic intelligence platform needs depth, source breadth, and analytical rigour, and its output would be useless on a battlecard.

Valona's partnership with a geopolitical consultancy is a signal of which end it plays at. Geopolitical risk analysis is not something a sales team consumes.

The category label covers both, which is why a shortlist assembled from a single Wave without deciding the use case produces vendors that barely compete.

The multilingual dimension

One hundred and fifteen languages is worth pausing on, because it is the same finding that appears in Forrester's intent data research, where non-English signal capture typically accounts for under twenty percent of total volume.

For an organisation operating outside the English-speaking world, or competing against companies that do, coverage in the languages where the relevant information is actually published is not a nice-to-have.

A European manufacturer's competitor announces a plant in a regional German business publication. A regulatory change is signalled in a Finnish ministry consultation. A Chinese competitor's capacity expansion appears in Mandarin trade press months before it reaches Western coverage.

An intelligence platform monitoring English-language sources will find all of these eventually, filtered through whoever decided they mattered enough to translate. Eventually is frequently after the decision window closed.

This is one of the clearest cases in this whole series where a European buyer should weight a criterion differently from an American one, and where a vendor's language coverage is a checkable fact rather than a marketing claim.

The distribution problem

The failure mode in intelligence functions is not analysis quality. It is that good analysis reaches nobody.

The pattern recurs everywhere. A competitive intelligence team produces a weekly digest. It is thorough, accurate, and read by the people who already agree it matters. The executives who most need it receive it, glance at the subject line, and file it. Six months later a decision is made that the digest specifically warned against.

The structural reasons are consistent. Intelligence is pushed on the producer's schedule rather than pulled at the moment of decision. It is generic rather than addressed to a specific question someone has. And the person who needs it does not know it exists at the moment they need it.

This is why the current evaluations weight generative capability and natural language query so heavily. A platform that answers a specific question when someone asks it, in the tool they are already using, changes the distribution model from broadcast to retrieval.

Forrester's note that Valona uses generative AI to populate predefined frameworks like SWOT points at the same shift. A framework someone requested is intelligence they will use. A digest nobody asked for is a newsletter.

The customer quote Forrester recorded, describing an eleven-year partnership, is worth noting alongside this. In a category where the technology is replaceable and the content access is not, relationships run long, and that has implications for negotiating leverage in both directions.

Where this leaves it

Market and competitive intelligence survived the moment that should have commoditised it because the scarce thing was never analysis. It was access to the sources worth analysing, in the languages they appear in, with the rights to use them.

That is a durable position, and it is also a narrow one. If a vendor's value proposition is summarisation and workflow over publicly available content, the case for paying for it weakens every quarter, because that capability is now available in general-purpose tools your organisation already licenses.

The evaluation question follows directly. Ask which sources a platform can access that you cannot reach otherwise, in which languages, under what licensing. That answer is specific, verifiable, and it separates the platforms with a moat from the ones with an interface.

Forrester does not endorse any vendor named here, and tier placement should not be read as a recommendation to buy.