The Competitor Profile Readiness Standard: Publish, Patch, or Hold

This is the grading rubric an analyst applies to a finished single-competitor profile before it goes to sales and leadership. It is for competitive intelligence analysts, strategy and research teams, and operators who own a competitor library and need publish, patch, and hold verdicts that hold up across analysts and revisions. The library already covers how to build a teardown and how a mixed briefing gets cleared to ship; this sets the per-competitor definition of done, so a verdict is reproducible instead of a matter of taste.

The standard rests on three axes: coverage (are the required sections there and substantive), sourcing (is every claim attributed and genuinely corroborated), and freshness (is every field dated and inside its shelf life). A profile clears only when all three clear. The most valuable part of this document is the failure-mode section, because that is where "looks finished" and "is finished" diverge.

What a complete competitor profile must contain

A complete profile carries a fixed set of sections, each holding at least one dated, sourced claim rather than filler. Public sources agree on the core and diverge at the edges, so the standard names both.

Sources converge on identity, products and services, pricing, positioning, and a strengths-and-weaknesses read. The most detailed public framework, Wikipedia's competitor analysis entry, extends this to background, finances, products, markets, facilities, personnel, and strategies, with background broken into offices and online presence, history, and ownership and governance. Strategy sources add a predictive response profile - the competitor's objectives, assumptions, strategy, and capabilities compiled into likely offensive and defensive moves. SaaS templates add ICP, value proposition, and go-to-market motion.

Where sources diverge is whether financials, facilities and personnel, and a predictive response profile are mandatory or optional. The standard treats them as required-if-relevant: mandatory when the competitive question depends on them, optional otherwise, and never left blank without a note saying why.

A section marked "TBD" is not a partial pass; it fails coverage until it holds a real claim or a documented reason for absence.

The load-bearing criterion is freshness, not coverage

Freshness is the axis that fails a profile most often and most quietly, because a fully populated profile can be wrong within weeks. Coverage tells you the section exists; freshness tells you whether what it says is still true.

The numbers make the case. B2B contact data decays at 2.1% per month, compounding to 22.5% per year, per the MarketingSherpa research behind HubSpot's Database Decay Simulation. Job title is the single biggest driver of record decay; mobile phone is the most stable. Firmographic data - the "stable" category - still goes obsolete at 20 to 30% per year per Dun and Bradstreet, and direct phone numbers decay at an estimated 25 to 35% per year. A profile you called complete last quarter has silently degraded on its most decision-relevant fields.

So the standard splits fields into volatile and stable and grades each against a shelf life. Volatile fields - pricing, headcount, leadership, titles - track the monthly rate and must carry a last-verified date under 90 days. Stable fields - founding, HQ, ownership - rarely change but still need a date; flag them past 12 months.

The 90-day line is not arbitrary. Two independent bodies of practice land on it: the recommended CRM contact re-verify interval is every 90 days, and the battlecard staleness threshold holds that a section dated older than 90 days needs a refresh. When two unrelated disciplines converge on the same number, you have a defensible bar rather than a house preference.

Field-specific shelf life is not formally established for CI profiles in public sources. The decay figures above are the closest defensible proxy. If your market moves faster than the averages, tighten the volatile-field window locally and record the reason.

The sourcing rule that separates fact from inference

Grade the source and the information separately, and make both visible on every material claim. This single discipline, borrowed from intelligence tradecraft, is what lets two analysts agree on whether a claim is solid.

The named implementation is the NATO/Admiralty Code: a two-character rating combining a source reliability grade from A to F and an information credibility score from 1 to 6. You do not need the full alphabet to use the principle. Rate the source (who said it, how reliable are they), rate the information (how well corroborated is the claim itself), and never collapse the two into a single "sourced/unsourced" tick.

The trap this prevents is fact-versus-judgment blur. A note that a competitor is "pulling back from enterprise, as indicated by a well-placed contact" hides whether the pullback assessment belongs to the contact or the analyst. Attribute the assessment explicitly to one or the other. If it is the analyst's inference, label it inference.

The second trap is false corroboration through circular reporting - when the same information reaches you through multiple channels and you mistake it for independent confirmation. Genuine corroboration requires that the confirming sources had independent access to the underlying facts. Trace each report back to its origin to tell whether you have five sources or one source counted five times.

The most common sourcing failure in practice is the profile built entirely from the competitor's own marketing site. That is a single source dressed as many. Any material claim needs at least one independent corroborator, and the corroboration must survive the origin trace.

When you need a person to run the reproducibility check in the procedure below, or to benchmark how mature teams staff the reviewer seat, Refolk resolves that search directly from the request. In Refolk's index, that talent is thin on the ground, which is exactly why a written rubric has to carry the weight a second reviewer would.

Refresh cadence: full republication versus interim scans

Freshness is maintained by a tiered cadence, not a single refresh date. The profile is a living artifact fed by fast signal scans and reset by slower full refreshes.

The composite of practitioner guidance is a weekly signal scan feeding a monthly-to-quarterly full refresh, with two exceptions that move faster. Field intel - something a rep or partner captured in the wild - should be published within 24 to 48 hours of capture. Pricing is event-driven: verify it before every competitive deal, regardless of the last-verified date on the card.

The failure to watch for is cadence theater: a team that says "we refresh quarterly" while the market moves weekly. Quarterly-only means a rep can be working from three-month-old intelligence in a fast market, and a three-month-old battlecard can actively mislead a rep rather than help them win. The weekly scan is what makes quarterly refresh safe. If there is no weekly signal scan feeding the profile, the freshness axis fails even when every field is technically inside its window.

The grading procedure

Grade the profile in a fixed order so the verdict is reproducible: scope, then coverage, then sourcing, then freshness, then failure-mode screen, then verdict, then a second-analyst check. The whole pass runs about four to six hours for one competitor.

One ordering note where sources disagree: battlecard practitioners lead with freshness and last-verified dates, while intelligence-tradecraft sources lead with source rating. The order above puts sourcing before freshness because a fresh but unsourced claim is still ungradeable, but if your team leans on field intel, swapping steps three and four is defensible as long as both are complete before the verdict.

Mapping grades to a verdict

The verdict is mechanical once the grades are in. Publish means all three axes clear. Patch means the gaps are fixable in a bounded edit - a missing citation, an undated field, one thin section - without changing the profile's conclusions. Hold means a failure that makes the profile actively misleading, so a rep would lose ground by trusting it.

How this goes wrong: the failure modes that trigger a hold

Most profiles that look finished fail on one of a small set of documented failure modes. Coverage failures usually warrant a patch; sourcing and freshness failures usually warrant a hold, because they make the document misleading rather than merely incomplete.

The single most common hold trigger is feature-parity padding, because it is the path of least resistance. Profiles get assembled from the competitor's own website, which inflates coverage while adding zero differentiation. A list of "we have API, they have API" is a checklist, not a wedge, and buyers do not care about parity. Each product line must state a wedge or it does not count as coverage.

Here are the failure modes, what each looks like, and the verdict each drives.

A word on the win/loss floor. Win/loss evidence outranks field anecdote, but it has a volume threshold: statistical viability needs 10 or more closed-lost deals per quarter. Below that, an individual deal dominates the "pattern," so a profile citing buyer evidence must also pass a sample-size check. Cite the count, not just the conclusion.

The subtlest failure is the coverage completeness false positive: every section filled, so the profile passes a naive checklist, but the sections are padded rather than substantive. This is why coverage grading asks not "is the section present" but "does the section contain a dated, sourced claim." A profile written for completeness instead of execution passes the wrong test.

The readiness checklist

Run this before you record a verdict. It is the operational form of the three axes plus the failure-mode screen, and it is the artifact a second analyst uses to check your work.

Keeping the standard current and reproducible

The standard is only worth as much as its reproducibility, so the last step is a second analyst regrading a sample and reconciling verdicts. When two graders reach the same publish, patch, or hold call on the same profile, the rubric is working; when they diverge, the fix is a tighter bar, not a longer argument.

Two-grader review is a staffing problem for most teams, and the data explains why. In Refolk's index, 223 people in the US hold Competitive Intelligence Analyst or Manager titles, against just 20 in the UK - an 11-to-1 concentration.

Both counts come from Refolk's index of profiles titled "Competitive Intelligence Analyst" and "Competitive Intelligence Manager"; the share and multiple are derived from the two counts. In the US sample, top employers include ADP, Apple, PwC, IBM, and Boomi; UK CI roles concentrate in London, with employers including Frazer-Nash Consultancy, Sopra Steria, and Lockheed Martin. The practical reading: outside a handful of large US employers, most teams cannot staff a dedicated second reviewer, which is exactly why a written, mechanical rubric substitutes for headcount.

Keep the standard current the same way you keep a profile current. Re-check the freshness intervals against your own market speed each quarter - if your volatile fields decay faster than the 90-day line assumes, tighten locally and record why. Re-check the failure-mode list against the holds you actually issued; if a recurring failure is not on the list, add it with its verdict. And when you need to recruit or benchmark the reviewer role the rubric depends on, Refolk resolves the search in plain English against the same index the numbers above come from.

Competitor: [name]
Graded by: [analyst]  Date: [date]
Coverage: pass / partial / fail - [note]
Sourcing: pass / partial / fail - [note]
Freshness: pass / partial / fail - oldest volatile field: [date]
Failure-mode flags: [cleared / logged - list]
Win/loss basis: [n closed-lost deals this quarter]
VERDICT: PUBLISH / PATCH / HOLD
Reasons: [one to three sentences]
Next refresh due: [date]  Reviewer reconciled: yes / no

Adopt the checklist and the verdict record as team policy, run the two-grader reconciliation until verdicts stop diverging, and the publish, patch, or hold call stops being a matter of taste and becomes a repeatable grade.