A documented ICP ages the moment the market moves, your product ships new value, or a new segment starts closing faster than the one you planned around. An annual ICP refresh is the ritual that keeps the profile honest: re-pull a full year of won, lost, and churned outcomes, test every existing dimension against the fresh data, de-ICP the cohorts that have stopped working, and reset the target lists, lead scoring, and sales plays that depend on it. This guide walks RevOps and product marketing through the full refresh in two to three weeks.
Time: 2-3 weeks
A refresh starts with new evidence, not last year's charts. Export every closed-won, closed-lost, and churned or heavily-contracted account from the last 12 months, snapshotted with the firmographics as they looked at opportunity creation rather than enriched today. Include industry, employee count, revenue band, geography, tech stack, buying committee size, acquisition channel, sales cycle length, ACV, gross margin, and 12-month net retention. Line it up next to last year's pull so cohort sizes and win rates are directly comparable. The three cohorts answer the three questions the refresh needs to settle: who still buys, who no longer buys, and who buys but churns.
Take each dimension from the current ICP and recompute its win-rate lift and retention lift against the new 12-month pull. A dimension that was pulling 1.8x last year and now runs at 1.1x is drifting toward noise. A dimension that still clears 1.5x is a keeper. Score both win-rate lift and retention lift separately so you catch the dimensions that still close fast but churn in month 11. Keep the sample size behind every number visible so a reviewer can tell which lift is signal and which is a small-cell artifact. The output of this step is a single ranked table of existing dimensions by current lift, with the delta against last year called out.
The refresh is not just a scorecard on last year's profile; it is a search for what has started working that was not in the profile at all. Re-cluster the fresh won and expansion cohorts and look for traits with 1.5x lift or higher that are not already in the ICP. Pay particular attention to buyer-title evolution: the champion role that closed deals last year may have been replaced by a different title, a committee may have grown or shrunk, or a new economic buyer may have appeared. Watch for use-case patterns too: a feature shipped mid-year may have unlocked a new vertical that is now punching above its weight. Flag every new candidate dimension with its lift, cohort size, and a one-line hypothesis for why it works.
Markets drift and the shape of your winner drifts with them. Compare this year's winning cohort to last year's on every firmographic axis: are you winning smaller or larger accounts, moving upmarket or downmarket, expanding into new geographies, winning in verticals you did not previously target, or losing ground in a stronghold vertical? Firmographic shifts rarely show up as a single trait change; they usually appear as a composite. A size band that grew by 20 percent of your wins paired with a new vertical mix and a longer sales cycle is a signal that your motion itself is shifting, not just the profile. Document each shift with the delta, the suspected cause, and whether it should harden into the next ICP or stay on watch for another quarter.
The hardest part of a refresh is removing dimensions and segments that were in last year's ICP. A de-ICP decision is a formal retirement of a dimension, vertical, size band, or buyer motion that no longer earns its spot. Pull every cohort that fell below 1.0x on both win rate and retention lift and write a short retirement memo for each: what it was, why it worked once, what changed, and what the team should do with the pipeline that still sits inside it. Add the retired cohort to the anti-ICP with a clear disqualification line so BDRs stop sourcing it, marketing stops targeting it, and CS stops staffing new onboarding plays for it. De-ICPs protect capacity for the dimensions that still earn their keep.
A refresh that lands as a Slack message disappears by Monday. Publish the new ICP as a one-page artifact with the full year-over-year changelog beside it: dimensions kept, dimensions retired, new dimensions added, buyer-title changes, firmographic shifts, and the anti-ICP additions. Show the lift numbers behind every call so a skeptical rep or marketer can audit the math. Version the document so anyone can diff this year against last year. Link it from every GTM template that depends on the ICP: the sales qualification script, the marketing brief, the demand-gen target list, the CS onboarding plan, and the territory design doc. If a new hire cannot read the artifact in 10 minutes and tell you what changed, the changelog is too long.
The refresh only matters if it changes what shows up in a rep's queue on Monday. Rebuild the target account list against the refreshed dimensions, both inside the CRM and in whatever outbound tooling the SDR team runs. Reweight the lead scoring model: dimensions that gained lift get more points, retired dimensions move to zero or negative, and new buyer titles enter the scoring table. Update sales plays so the discovery, qualification, and disqualification language reflects the new buyer evolution and the de-ICP cohorts. Marketing rebuilds segments, paid-media audiences, and nurture tracks against the same list. CS updates onboarding plays and health-score inputs against the retention lift findings. Lock the next quarter plans to the refreshed ICP so nothing in-flight quietly runs against the old profile.
A refreshed ICP that does not get retrained against is a document, not a program. Run a working session with each GTM function the week the artifact ships so sales, marketing, demand gen, SDR, and CS all walk through the changelog, the new dimensions, and the retired cohorts out loud. BDRs practice the new disqualification lines for the de-ICPs so the cuts stick on live calls. Marketing presents the new target lists back to sales for sign-off. CS presents the new onboarding plays back to the account team. Then lock the next refresh date on the calendar: a quarterly lift re-check and a full annual refresh with named owners for each. Treat the ICP like production code: versioned, reviewed, owned, and backtested every release.
A new ICP build is a first-time extraction with no prior profile to compare against; it leans on customer interviews and the first full year of data to form a hypothesis. An annual refresh assumes a documented ICP already exists and tests it against a fresh 12 months of outcomes. The refresh is faster, more quantitative, and produces a year-over-year changelog with explicit de-ICPs rather than a new artifact.
RevOps owns the data and lift calculations, and product marketing co-owns the narrative, use cases, and buyer-title evolution. Sales, demand gen, and customer success leaders sign off on dimensions, retired cohorts, and the new target lists. An ICP refresh owned only by one function tends to miss either the market story or the retention story.
Run a lighter quarterly lift check: re-pull trailing-quarter won, lost, and churned data, recompute lift on each dimension, and flag anything that drops below 1.3x lift or crosses 1.5x. The quarterly check surfaces drift early; the annual refresh is where you formally retire, add, and republish dimensions with a full changelog.
A major product launch that changes the use case, a pricing or packaging change that shifts the economic buyer, entry into a new segment or geography, or two consecutive quarters where multiple dimensions fall below 1.3x lift. Any of those is enough to pull the next refresh forward rather than wait for the calendar anniversary.
Pair every de-ICP with evidence and a sunset plan. Show the win-rate and retention-lift numbers so the retirement is not a judgment call, decide whether in-flight pipeline runs to close or gets reassigned, and offer a six-month sunset where the cohort is allowed in existing pipeline but gets no new SDR capacity, campaigns, or target-list seats. Reps accept a cut backed by numbers and a transition plan; they resist a cut that feels like a surprise.
A one-page refreshed ICP with 5-7 dimensions and thresholds, a short anti-ICP list, a year-over-year changelog of what was kept, retired, and added, the lift numbers behind every call, two or three named example accounts per quadrant, and the next review date with named owners. Link to the raw evidence sheet so anyone can audit the math behind each decision.
Strkr brings target lists, lead scoring, and pipeline into one platform so every refreshed ICP dimension shows up where sales, marketing, and CS actually work.
No sales call, no migration consultant, no four-month implementation. Enter your card, get 14 days of the full Pro tier, cancel any time before day 14 with zero charge. Spin up a workspace, import your CSV, and have something useful before lunch.
