Building an ideal customer profile with intent signals is what turns the ICP from a slide in a board deck into an operating asset that drives weekly action. Per Forrester research, the under-100M-ARR teams that integrate intent signals into the ICP definition outperform their static-ICP peers materially on pipeline efficiency, because the ICP becomes a live filter rather than a quarterly artefact. This guide walks the seven-step build that fuses firmographic, technographic, and behavioural intent into a working ICP that updates weekly.
Full disclosure: Abmatic AI ships an ABM platform that consumes ICP-with-intent definitions to drive campaign and routing decisions. The framework below is platform-agnostic. It works whether your data lives in Salesforce, HubSpot, Snowflake, BigQuery, or a stack of vendor-supplied data sets.
The 30-second answer
Build an ICP with intent signals in seven steps: anchor the firmographic profile on closed-won evidence (industry, size, geo), layer technographic gates (tech-stack indicators tied to product fit), define a fit score from the firmographic-plus-technographic combination, integrate first-party intent signals (web visits, content downloads, demo requests), integrate third-party intent signals (Bombora-style topic surges, G2 buyer-intent), define the merged ICP-plus-intent score with explainable weights, and instrument the CRM with the score and an in-market flag that updates daily. Per public customer reports, ICP-plus-intent definitions produce two to three times the pipeline-efficiency lift of firmographic-only ICPs at the same budget.
See an ICP-plus-intent score driving live campaign and routing decisions, book a demo.
Why firmographic-only ICPs underperform
The recurring failure modes of firmographic-only ICPs, per public customer reports across the under-100M-ARR band:
- Static. The ICP refreshes quarterly at best, and intent shifts week to week. The team is always working a stale list.
- Too broad. Firmographic gates alone produce a universe of 3000 to 8000 accounts, half of which are not in-market for the next 12 months.
- No prioritisation. All accounts that pass firmographic gates are treated as equal, when in reality 5 to 10 percent of them are showing surge intent and warrant immediate action.
- No prediction. Without intent overlay, the ICP cannot tell you which accounts are about to enter market versus which are still 18 months away.
The seven-step framework below addresses each of these.
The seven-step build
| Step | Output | Owner | Time |
|---|---|---|---|
| 1. Anchor firmographic profile on closed-won | Firmographic profile rule | RevOps plus marketing | 3 to 5 days |
| 2. Layer technographic gates | Tech-stack rules tied to product fit | Sales engineering plus RevOps | 3 to 5 days |
| 3. Define a fit score | Combined firmographic-plus-technographic score | RevOps plus analytics | 1 to 2 weeks |
| 4. Integrate first-party intent | Web, content, demo signals tagged to account | Marketing ops | 1 to 2 weeks |
| 5. Integrate third-party intent | Topic surges from Bombora, G2, vendor surfaces | RevOps plus marketing ops | 1 to 2 weeks |
| 6. Define the merged ICP-plus-intent score | Explainable weighted score | RevOps plus analytics | 2 to 3 weeks |
| 7. Instrument CRM with score plus in-market flag | Daily-updating fields | RevOps | 1 week |
Step 1: Anchor firmographic profile on closed-won
Pull the last 24 months of closed-won opportunities, expansions, and renewals. Strip the smallest 20 percent and the largest 5 percent. The remaining 75 percent forms the empirical firmographic profile. Look for industry concentration, employee-band concentration, geography concentration, and trigger-event concentration. For the deeper firmographic build, see how to build an ICP.
Step 2: Layer technographic gates
Technographic gates narrow the universe further. Tech-stack signals that align with product fit (typical examples: cloud platform, CRM system, marketing automation, vertical software). Tech-stack-only filters tend to be too narrow on their own; they work as a layer on top of firmographics.
Step 3: Define a fit score
The fit score combines firmographic and technographic signals into one observable number. A defensible default formula: each signal weighted by its closed-won correlation, normalised to a 0 to 100 score per account. The fit score updates monthly as new closed-won data arrives. For the broader account-fit lens, see account fit score.
Step 4: Integrate first-party intent
First-party intent is the strongest signal. Three sub-categories:
- Direct intent: demo request, pricing-page visit, comparison-page visit. Highest weight.
- Indirect intent: repeat visits to product pages, content downloads, webinar attendance. High weight.
- Engagement signals: email opens, ad clicks, content video views. Medium weight.
For the deeper distinction, see first-party intent data.
Step 5: Integrate third-party intent
Third-party intent extends coverage to accounts not yet engaging your direct properties. Two main sub-categories:
- Topic surges from Bombora-style intent providers. Medium weight.
- Buyer-intent from G2 (or similar review platforms) where the account is actively researching your category. High weight.
For the merge logic, see how to merge first- and third-party intent and signal merge.
Step 6: Define the merged ICP-plus-intent score
The merged score combines fit and intent. A defensible default: the score is a multiplicative combination (fit score times intent score), so an account with high fit but no intent is medium-priority, and an account with low fit but high intent is also medium-priority. The combination of high fit plus high intent is the action priority. Alternative: weighted sum with explicit weights, easier to explain to stakeholders. Choose based on whether the team values action-readiness or fit-confidence more.
Step 7: Instrument CRM with score plus in-market flag
Three CRM fields:
- ICP-plus-intent score: the merged 0 to 100 score, daily refresh.
- In-market flag: boolean, true when the score crosses a defined threshold (typical default: above 70).
- Last-signal timestamp: for freshness reasoning at the rep level.
Without these three fields in CRM, every downstream tool rebuilds the score and the scores drift.
The framework: fit, intent, merge, action
- Fit layer defines who could buy (firmographic plus technographic).
- Intent layer defines who is in-market now (first-party plus third-party).
- Merge layer combines them into one score.
- Action layer uses the score for routing, prioritisation, and campaign decisions.
The four-layer structure produces an ICP that updates daily and drives weekly action, rather than a static slide.
Skip the manual work
Abmatic AI runs targets, sequences, ads, meetings, and attribution autonomously. One platform replaces 9 tools.
See the demo →What to measure
Three metrics, in order of importance. First, in-market account count: number of accounts crossing the in-market threshold each week. Material drops or spikes indicate ICP or intent calibration drift. Second, in-market-to-meeting conversion rate: of accounts that crossed the in-market threshold, what percentage produced a meeting in 30 days. Target band: 5 to 15 percent at the under-100M-ARR band, per public customer reports. Third, ICP-plus-intent versus firmographic-only lift: cohort comparison of the two scoring methods, validating that intent overlay is doing real work.
Common traps
Trap 1: Static firmographic-only ICP
The ICP refreshes quarterly, intent shifts week to week. Without intent overlay, the team always works a stale list.
Trap 2: Treating all intent sources equally
Demo request and Bombora topic surge are not the same signal. Source weighting is non-negotiable.
Trap 3: No freshness decay
An intent signal from 30 days ago is noise. Linear decay over 14 days is the defensible default for most sources.
Trap 4: No CRM instrumentation
Without the three CRM fields, every downstream tool rebuilds the score, the scores drift, and reps see different numbers.
Trap 5: No explainability
Black-box scores collapse trust. Every score should be explainable in three to five top contributing signals at the account level.
How this connects to the rest of the ABM stack
ICP-plus-intent is the upstream definition that feeds everything downstream. Outputs flow into the target account list, the intent routing pipeline, the mixed-signal prioritisation framework, and ultimately the buying-committee orchestration.
Related ICP and intent guides: how to build an ICP from scratch in 2026, how to use intent data, and predictive intent data.
FAQ
How is an ICP-with-intent different from a traditional firmographic ICP?
The traditional ICP is static (firmographic and technographic gates that refresh quarterly). The ICP-with-intent overlays in-market signals so the score reflects who is buying now, not just who could buy. The traditional ICP defines the universe; ICP-with-intent prioritises the action queue.
What weight should fit and intent each carry?
Roughly equal in a defensible default merge. If your sales cycle is short (under 90 days), weight intent slightly higher because freshness matters more. If your sales cycle is long (above 12 months), weight fit higher because intent decays before the deal closes.
How often should the score refresh?
Daily for intent inputs, weekly to monthly for fit inputs. Reps should see a refreshed score every morning; analytics should see refreshed inputs every week.
What is the right in-market threshold?
Calibrate to volume. A 0 to 100 score with a threshold at 70 typically produces 5 to 15 percent of the universe in-market at any time, which is the manageable band for most under-100M-ARR teams.
Should the score be a black-box ML model or an explainable weighted score?
Explainable weighted score for the first 12 months. Black-box ML can earn its place after the team has trust in the rule-based score and enough labelled data to validate the model. Skipping straight to black-box collapses adoption.
How does ICP-with-intent connect to predictive intent data?
Predictive intent extends the framework with model-derived in-market scoring (often vendor-supplied). Use predictive intent as one of the third-party intent inputs, weighted alongside Bombora-style topic surges and G2 buyer-intent. Predictive intent is most useful when calibrated to your closed-won data; vendor defaults out of the box are weaker.
Building an ICP with intent signals is the difference between a quarterly slide and a working asset that drives weekly action. Seven steps, three CRM fields, four layers. The teams that build it have a live, daily-updating prioritisation system; the teams that ship a static firmographic ICP rebuild it every two quarters and never see the lift.
See an ICP-plus-intent score driving live campaign and routing decisions, book a demo.
