How to build a Competitive Intelligence function from scratch

Most organisations don't have a competitive intelligence function. They have a shared drive nobody updates, a Slack channel where reps occasionally mention a competitor, and a software subscription that half the team has logged into once. Building something that actually changes competitive outcomes requires a specific structure — and an honest decision about which parts to build internally and which to partner for.

Ask ten B2B organisations whether they have a competitive intelligence function and most will say yes. Ask what that function actually does, and the answers converge on something narrower: a monitoring tool, a battlecard template last updated two quarters ago, and an informal expectation that sales reps will "know the competition" through osmosis.

This is not a competitive intelligence function. It is the absence of one, dressed in the language of having built something. A genuine CI function has a defined structure, a specific place in the organisation, a clear division of what is monitored versus what is investigated, and — critically — an honest answer to the question of which capabilities are worth building internally and which are not.

This article sets out that structure: a four-stage maturity model, the organisational placement decision, the five functions a mature programme performs, and a framework — borrowed directly from economic theory on organisational boundaries — for deciding what to build and what to buy.

The four-stage maturity model

Competitive intelligence maturity has been studied extensively in the practitioner and academic literature, notably in Fleisher and Bensoussan's work on competitive analysis capability, which frames CI development as a progression from informal, individually-held knowledge toward a structured, organisationally-embedded capability.1 Most organisations can locate themselves clearly within one of four stages.

STAGE 0

Tribal knowledge

No formal process exists. Competitive information lives in individual reps' heads and scattered conversations. Nothing is captured centrally, nothing compounds over time, and the same intelligence is repeatedly rediscovered — or permanently lost — whenever a rep with tenure leaves.

STAGE 1

Reactive monitoring

A monitoring platform is purchased. Ownership sits, usually part-time, with someone in product marketing. Battlecards exist but are built entirely from public signals and updated inconsistently. This is the stage the largest proportion of B2B organisations plateau at — visible activity, limited strategic impact.

STAGE 2

Structured intelligence function

A dedicated owner, sometimes a small team, runs a defined cadence: regular battlecard refreshes, occasional primary research such as customer win-loss interviews, and a formal intake process for sales requests. Monitoring software handles the surface layer; deeper strategic questions are still typically unanswered.

STAGE 3

Embedded strategic intelligence

CI informs decisions well beyond sales enablement — product roadmap prioritisation, M&A due diligence, pricing strategy, market entry timing. Primary human intelligence is either resourced internally with trained, dedicated capability, or maintained through an ongoing specialist partnership. Intelligence has a functioning feedback loop back into itself.

The jump from Stage 1 to Stage 2 is usually a resourcing decision — someone is given formal ownership and time. The jump from Stage 2 to Stage 3 is a different kind of decision entirely, and it is the one most organisations never make deliberately: it requires acknowledging that monitoring, however well run, cannot answer the questions that determine competitive outcomes, and building or sourcing the capability that can.

Where should the function report?

Organisational placement shapes what a CI function becomes, often more decisively than the skill of whoever runs it. Three placements recur across most B2B organisations, and each carries a predictable failure mode.

SALES ENABLEMENT

Placement here maximises field relevance — battlecards get built for exactly what reps need and are more likely to actually get used in live calls. The failure mode: strategic questions that matter beyond the current quarter's pipeline — should we enter this market, should we acquire this company, is our pricing structurally wrong — rarely reach a function whose mandate is defined by immediate sales support.

PRODUCT MARKETING

Placement here integrates competitive positioning cleanly with messaging and narrative. The failure mode: product marketing teams are typically resourced for content and positioning work, not for primary research, and the function tends to stay permanently at Stage 1 — a well-branded monitoring output with no deeper investigative capability behind it.

CORPORATE STRATEGY

Placement here produces strong integration with high-level decisions — pricing committees, roadmap prioritisation, market entry planning. The failure mode: strategy functions often operate on a slower cadence and longer time horizon than live deal cycles, which means the intelligence produced, however rigorous, rarely reaches a rep before the call where it would have mattered.

RESEARCH CONTEXT

Jan Herring's foundational work on Key Intelligence Topics — a structured process for defining what an organisation actually needs to know, rather than attempting comprehensive coverage of everything about every competitor — remains one of the most practically important contributions to CI programme design.2 A function without clearly defined intelligence priorities will default to whatever its organisational placement makes most visible, regardless of whether that is what leadership actually needs answered.

The practical resolution is less about which box the function sits in on an org chart, and more about whether a mandated, structured intake and output relationship exists with the sales organisation regardless of formal reporting line. A strategy-placed function with a weekly sales sync produces different — and generally better — outcomes than the same function operating in isolation.

"A function without clearly defined intelligence priorities will default to whatever its organisational placement makes most visible."

The five functions a mature programme performs

Regardless of where it sits, a competitive intelligence function that has reached Stage 2 or 3 maturity performs five distinct roles. Skipping any one of them produces a specific, predictable weakness in the finished output.

1. Monitoring

Tracking publicly available signals — website changes, pricing updates, job postings, press activity — for change over time. This is the layer software tools are built for and generally handle well.

2. Primary intelligence gathering

Structured elicitation from human sources — former employees, channel partners, shared customers — to access the tacit knowledge layer that monitoring cannot reach. This is the specialist skill layer, and the one most Stage 1 and Stage 2 programmes never build.

3. Analysis & synthesis

Turning raw findings — from either monitoring or primary sources — into a coherent, decision-relevant conclusion. A list of facts about a competitor is not intelligence until it has been synthesised into an answer to a specific question someone is trying to decide.

4. Dissemination

Delivering intelligence in the specific format, length, and cadence the end user actually needs — a two-page battlecard for a rep, a briefing session for a CRO, a strategic note for a pricing committee. Intelligence that is accurate but undeliverable in a usable form has no operational value.

5. Feedback loop

Validating intelligence against actual field and win-loss outcomes, and correcting course when findings prove inaccurate or stale. Without this, a programme has no mechanism to know whether the intelligence it produces is actually improving competitive outcomes — or simply generating well-formatted documents nobody checks against reality.

The build-versus-buy decision

Once the five functions are defined, the practical question becomes which to resource internally and which to source externally. This is not a new problem — it is a specific application of a well-established question in economic theory. Oliver Williamson's transaction cost economics framework, developed across his 1975 and 1985 work on markets and organisational hierarchies, provides the clearest lens available for making this decision rigorously rather than by instinct.3

Williamson's framework identifies the conditions under which a capability is more efficiently built inside an organisation versus sourced through an external market relationship: internal build is favoured when the capability is needed frequently, when its requirements can be clearly specified in advance, and when the relevant expertise is not highly specialised. External sourcing is favoured when the opposite conditions hold — infrequent need, requirements that are hard to fully specify in advance, and high specialisation that takes years to develop organically.

Applied directly to the five CI functions above, the framework produces a clear, defensible allocation:

The pattern is consistent with what transaction cost theory predicts: the capability requiring the deepest specialisation and the longest organic development timeline — primary human intelligence — is also the one most organisations are least equipped to build internally at the scale most B2B companies actually need it. This is not a failure of ambition. It reflects a genuine, well-documented economic logic about when specialisation is better sourced than grown.

What building HUMINT capability internally actually requires

To be direct about the scale of this specific build decision: internal HUMINT capability requires hiring people with genuine elicitation training or an intelligence background, building and maintaining a source network across the specific competitors that matter — which takes years, not a quarter — and establishing ethical and legal protocols robust enough to withstand scrutiny, discussed at length elsewhere in this series. For organisations running frequent, high-stakes competitive evaluations against a small number of consistent competitors, this investment can be justified. For most B2B organisations, the frequency does not yet justify a full-time internal capability, which is precisely the condition under which Williamson's framework predicts external partnership as the more efficient allocation.

A PATTERN ACROSS EARLY-STAGE PROGRAMMES

A pattern shows up consistently in organisations building a CI function for the first time: budget and attention go disproportionately toward the monitoring layer — the visible, easily procured, easily justified line item — while the primary intelligence layer goes entirely unfunded, either because it was never budgeted separately or because the organisation assumed the monitoring tool would eventually cover it.

The result is a function that looks complete on an org chart and a budget line, and produces almost nothing that changes a competitive outcome — because the layer actually capable of answering "why do we keep losing" was never built at all.

A practical roadmap: the first 90 days

DAYS 1–30

Needs assessment & intelligence priorities

Define a small set of Key Intelligence Topics — the specific questions leadership actually needs answered, not an open-ended mandate to "know everything" about every competitor. Audit existing assets: CRM loss-reason data (with appropriate scepticism about its reliability), current software subscriptions, and informal knowledge scattered across the sales team. Decide organisational placement and, more importantly, define the mandated cadence with sales regardless of where the function formally sits.

DAYS 31–60

Build the monitoring layer & define the gap

Stand up or consolidate monitoring tooling against the priorities defined in the first 30 days. Map the five functions against internal capability honestly, using the build-versus-buy framework above, and identify explicitly where the primary intelligence gap sits. If partnering for HUMINT capability, scope the first engagement using a structured brief — what decision it will inform, what is already known, what format the output needs to take.

DAYS 61–90

First intelligence cycle & feedback loop design

Run one complete cycle end to end: intelligence gathered, synthesised, disseminated as a battlecard or briefing, and used in at least one live competitive situation. Design the feedback mechanism from the outset — a third-party win-loss process, not a CRM dropdown — rather than treating it as a later addition. Review outcomes with stakeholders and revise intelligence priorities for the next cycle based on what was actually useful in the field.

A function built this way reaches Stage 2 maturity within a single quarter. Reaching Stage 3 — where intelligence routinely informs product, pricing, and market entry decisions rather than only sales enablement — is a longer trajectory, typically 12 to 24 months, and depends less on process design than on sustained organisational buy-in that intelligence is worth acting on even when its conclusions are inconvenient.

The structure matters less than the honesty of the build-versus-buy decision inside it. A CI function that monitors well but never invests in the tacit knowledge layer will, regardless of how well it is organised, run into the same ceiling examined throughout this series: it will know everything a competitor publishes, and almost nothing about why deals are actually won and lost.

REFERENCES

1. Fleisher, C.S., & Bensoussan, B.E. (2003). Strategic and Competitive Analysis: Methods and Techniques for Analyzing Business Competition. Prentice Hall.

2. Herring, J.P. (1999). Key intelligence topics: A process to identify and define intelligence needs. Competitive Intelligence Review, 10(2), 4–14.

3. Williamson, O.E. (1975). Markets and Hierarchies: Analysis and Antitrust Implications. Free Press. See also: Williamson, O.E. (1985). The Economic Institutions of Capitalism. Free Press.