Reference

Why B2B Prospects Fail Scoring: A Taxonomy of Prospect Quality

Most outbound teams measure the wrong failure. A large share of it is decided before a word is written: the prospect should not have been contacted at all. This is a reference taxonomy of nine ways that happens, in four classes — and of the three conditions that look identical from the outside and need opposite remedies.

What prospect scoring is

Prospect scoring is the practice of assigning a comparable, evidence-derived rating to a potential buyer in order to decide who to contact first. A prospect score answers a question of priority. It does not, on its own, answer whether the underlying evidence is sufficient to support that priority.

Prospect quality failure is any condition under which a prospect should not be contacted now, or should not be contacted in the way the score implies. Failure is not a single condition. It is a set of distinct conditions with different causes, different remedies, and different costs.

For how this is applied inside a working system, see prospect scoring.

Every prospect carries two readings

Composite quality. A single comparable rating over eight factors — relevance and business fit in the primary band; response likelihood, contact validity and influence in the supporting band; engagement, audience overlap and relationship proximity in the contextual band. When a factor cannot be resolved it is neutral-filled and flagged, which keeps scores comparable — and means a score built on three resolved factors and the same score built on eight are not the same claim.

Evidence completeness. How much of the model's weight was actually supported by evidence, discounted by how trustworthy that evidence is. Reliability runs, strongest first: verified, provider, inferred, modelled, cold start.

Two readings, two questions: how good does this look? and how much do we actually know? Collapsing them is the root cause of most entries below.

Poor fit, insufficient evidence, and low confidence are not the same thing

This is the core of the resource. Each has a different remedy, and treating them identically is why prospect lists get cut by volume instead of by reason.

Poor fit — the organisation or person is genuinely outside the target profile, and the evidence supporting that conclusion is good. Remedy: exclude. Spending more here is pure waste.
Insufficient evidence — there may be a real opportunity; what exists is not enough to support the judgement the score is making. Remedy: enrich, then re-score. The score is not wrong, it is unfinished.
Low confidence — evidence exists, but its quality, freshness, consistency or attribution is weak: modelled rather than verified, aged past its useful life, or contradicted by another source. Remedy: corroborate or downgrade. This is the most dangerous of the three, because from the outside it looks identical to the other two — a number.

Do not collapse “we don't know” into “bad prospect”. They have opposite remedies: one says stop spending, the other says spend a little more, precisely.

The taxonomy at a glance

Root

Prospect Quality

Two readings: how good does this look, and how much do we actually know.

Fit
Legible, but not who you serve
Evidence
Not enough known to support the claim
Reachability
Qualifies, cannot be reached
Motivation / Timing
No reason to act now

Text equivalent. Prospect Quality divides into four classes. Fit contains ICP mismatch (1.1), surface-match / substance-mismatch (1.2), and right company, wrong unit (1.3). Evidence contains insufficient business evidence (2.1), low evidence confidence (2.2), stale or contradictory information (2.3), and score/evidence divergence (2.4). Reachability contains poor contactability (3.1), incomplete decision-maker identification (3.2), and sending not authorized (3.3), which is a precondition evaluated outside the score rather than a scoring outcome. Motivation / Timing contains weak or ambiguous need signal (4.1), poor timing or priority (4.2), and weak personalization evidence (4.3). No meaning in this diagram is carried by colour alone.

The nine failure modes

Nine modes in four classes. Each entry carries a stable anchor, because people cite one class, not a page. One structural note: 3.3 Sending not authorized is a precondition rather than a reachability failure — it is not a scoring outcome at all. It sits outside the score and overrides it, and it keeps its number for citation stability.

Class 1 — Fit failures

The prospect is legible, and they are not who you serve.

1.1

ICP mismatch

Fit failure

Definition: The business is well-evidenced and falls outside the target profile — wrong segment, wrong size, wrong operating model.
In practice: Business fit reads low on substantiated evidence. Nothing is broken.
Why quality is low: They will not buy, at any message quality.
Observable evidence: Category, size proxies, offering and customer type all resolved, and all pointing away from the profile.
False-positive risk: Low, but non-zero when the ICP definition itself is stale. A profile written for last year's motion will confidently exclude this year's best segment.
Reversing evidence: A change in the account — a segment shift or a new line of business — or a revision to the ICP definition. Not more data about the same account.

Link to 1.1

1.2

Surface-match, substance-mismatch

Fit failure

Definition: Relevance reads high on topical and keyword shape while business fit reads low.
In practice: Right vocabulary, right category page, wrong underlying business — a directory that lists your ICP, a publisher writing about them, a consultancy serving them.
Why quality is low: Relevance is a primary-band factor, so a high composite can be driven almost entirely by surface language. The score is confident and hollow.
Observable evidence: Divergence between the two primary factors, with relevance carrying the composite.
False-positive risk: Moderate. Genuinely in-profile businesses with thin websites can present the same divergence for the opposite reason — low business-fit evidence rather than low business fit. Check the evidence reading before excluding.
Reversing evidence: Any first-party or provider-verified fact about what the business actually sells, and to whom.

Link to 1.2

1.3

Right company, wrong unit

Fit failure

Definition: The organisation qualifies; the specific entity discovered does not — a franchise location, a regional subsidiary, a holding shell, a dormant brand.
In practice: Fit evidence is strong, but it attaches to a parent that is not the entity you found.
Why quality is low: The buying decision does not live at the level you are addressing.
Observable evidence: Corporate-structure signals, shared infrastructure across many named entities, location-page patterns, ownership disclosures.
False-positive risk: High. Many multi-location businesses buy locally. Franchise does not imply no autonomy; this is the entry most often applied too aggressively.
Reversing evidence: Evidence of local purchasing authority — a named local decision maker with budget, or locally distinct vendor relationships.

Link to 1.3

Class 2 — Evidence failures

You do not know enough to make the claim the score is making.

2.1

Insufficient business evidence

Evidence failure

Definition: Too few factors resolved; most of the model's weight was neutral-filled.
In practice: A number that reads mediocre and is actually an absence of information, or one that reads high and is actually an absence of contradiction.
Why quality is low: The composite is not describing the prospect. It is describing the default.
Observable evidence: The evidence reading is thin or speculative while the composite reads normal.
False-positive risk: Low as a diagnosis, high as a rejection. This is a reason to enrich, and treating it as a reason to drop is the expensive mistake.
Reversing evidence: Almost anything — one verified business fact typically moves the reading.

Link to 2.1

2.2

Low evidence confidence

Evidence failure

Definition: Enough factors resolved, but predominantly from modelled or cold-start sources rather than verified ones.
In practice: The model has an opinion and very little fact. Characteristic of new workspaces, where response likelihood has no send history to learn from and falls back to a proxy.
Why quality is low: The score is structurally least trustworthy for the first prospects a team ever scores — exactly when people trust it most.
Observable evidence: Reliability mix skewed to modelled and cold start, while coverage looks fine.
False-positive risk: Moderate. Modelled evidence is not worthless; it is unverified. Treat it as a discount, not a disqualification.
Reversing evidence: Corroboration from an independent reliability class — a provider assertion or a first-party observation agreeing with the model.

Link to 2.2

2.3

Stale or contradictory information

Evidence failure

Definition: Evidence exists but has aged past the point its source intended, or two sources disagree.
In practice: A role that changed, a site that was rebuilt, a business that moved, two providers reporting different headcounts.
Why quality is low: Contradiction is worse than absence — it produces confident wrongness, and the operator gets no cue that anything is wrong.
Observable evidence: Source timestamps beyond their useful life; disagreement across providers. Decay rates differ sharply by factor: contact validity degrades fast, business fit slowly.
False-positive risk: Moderate. Slow-moving facts flagged by a uniform freshness rule generate noise; freshness policy has to be per-factor.
Reversing evidence: A recent, higher-reliability observation that resolves the conflict.

Link to 2.3

2.4

Score/evidence divergence

Evidence failure — diagnostic

Definition: The composite reads well and the evidence reading is thin.
In practice: Not a separate defect so much as the instrument that surfaces 2.1 through 2.3.
Why quality is low: It is the single most useful thing to put in front of an operator before they approve a send, and almost no tool shows it.
Observable evidence: The two readings, side by side, on the same row.
False-positive risk: Low. It is a diagnostic, not a verdict — it prompts a look, not a drop.
Reversing evidence: Enrichment that lifts the evidence reading while the composite holds.

Link to 2.4

Class 3 — Reachability failures

The prospect qualifies and you cannot actually reach them.

3.1

Poor contactability

Reachability failure

Definition: No deliverable address, or one that fails verification.
In practice: Contact validity is a supporting contribution deliberately, because contactability is better treated as a gate than as a contribution.
Why quality is low: A prospect you cannot reach is not a slightly worse prospect; in this channel they are not a prospect. Averaging that away inside a composite is a modelling error made constantly by outbound tooling.
Observable evidence: Verification result, deliverability signals, absence of any resolvable address.
False-positive risk: Moderate. Verification providers disagree, and catch-all domains are routinely misreported as invalid.
Reversing evidence: A verified address, or a different channel where the same person is reachable. This is channel-specific, not absolute.

Link to 3.1

3.2

Incomplete decision-maker identification

Reachability failure

Definition: A contact exists, but their relationship to the decision is unknown or wrong.
In practice: A generic inbox, a role that cannot buy, a name with no title resolution. Influence reads neutral not because influence is average but because it is unmeasured.
Why quality is low: The message will be well-targeted at someone who cannot act.
Observable evidence: Missing or unresolved title; a role-based rather than personal address; no seniority signal.
False-positive risk: High in small businesses, where a generic inbox is frequently read by the owner and is the correct path.
Reversing evidence: Title resolution, org-structure evidence, or an observed reply from an authorised person.

Link to 3.2

3.3

Sending not authorized

Precondition — not a scoring outcome

Definition: Suppression, unsubscribe state, jurisdictional constraint, workspace safety limit, or an unverified sending identity.
In practice: A hard gate evaluated outside the score, able to override any score.
Why quality is low: It is not a quality question. A well-scored prospect who has opted out is not a decision, it is a violation.
Observable evidence: Suppression state, consent record, sending-identity verification state.
False-positive risk: Irrelevant by design. This gate is deliberately over-inclusive: a false block costs one prospect, a false pass costs the sending domain.
Reversing evidence: Only an authoritative change in consent or authorisation state. Never a score.

Link to 3.3

Class 4 — Motivation and timing failures

Fit and reachability are fine; there is no reason to act now.

4.1

Weak or ambiguous need signal

Timing / motivation failure

Definition: Nothing observable suggests the problem you solve is present or acute.
In practice: Everything qualifies and nothing points at a problem.
Why quality is low: This is a question of priority, not viability. Note the asymmetry: absence of a need signal is not evidence of no need, it is evidence of no observation.
Observable evidence: Absence of the signals your category normally leaves behind.
False-positive risk: High. Under-observed prospects are indistinguishable from unmotivated ones without checking the evidence reading first.
Reversing evidence: Any observed trigger, or enrichment revealing that the signal was there and unread.

Link to 4.1

4.2

Poor timing or priority

Timing / motivation failure

Definition: The need is real; the trigger has passed or has not arrived.
In practice: They just bought, just rebuilt, just hired for it, or are mid-cycle somewhere else.
Why quality is low: The most perishable dimension in the taxonomy — and the one most often absent from scoring models entirely.
Observable evidence: Recent-change signals, contract-cycle proxies, hiring patterns.
False-positive risk: Moderate. “Just bought” often means “will re-evaluate on a known date”, which is a scheduling fact, not a rejection.
Reversing evidence: A dated trigger, or simply the passage of time plus a re-score.

Link to 4.2

4.3

Weak personalization evidence

Timing / motivation failure — message level

Definition: Everything qualifies, and there is nothing specific enough to say.
In practice: The prospect is defensible and the message cannot be.
Why quality is low: Produces the failure that looks like a copy problem and is an evidence problem. Generic outreach is usually the symptom of a prospect you did not learn enough about.
Observable evidence: No citable, specific, current fact about this business.
False-positive risk: Low. If you cannot find something to say, you cannot say it.
Reversing evidence: One concrete, current, verifiable observation.

Link to 4.3

Reading the taxonomy as a decision

The nine conditions do not carry equal consequence.

Disposition for each failure class, with the reason for that disposition
Class	Disposition	Why
Authorization (3.3)	Hard stop	Never overridable by score
Contactability (3.1)	Gate	Not a candidate in this channel
Confident ICP mismatch (1.1)	Exclude	Working as intended; do not spend more
Substance mismatch (1.2), wrong unit (1.3)	Verify, then decide	Score is confident and possibly hollow
Evidence failures (2.x)	Enrich, then re-score	The cheapest interventions live here
Decision-maker gaps (3.2)	Enrich	Recoverable
Timing and need (4.1, 4.2)	Defer and watch	Re-scoring later is the point
Personalization gap (4.3)	Enrich or downgrade	Do not send generically to compensate

The useful reframe: most “bad prospects” are not bad, they are unfinished. Class 2 and 3.2 are information problems with a known price. Only 1.1 and 3.3 are true terminations. Class 4 is a scheduling problem misfiled as a quality problem.

One number cannot tell you whether to drop a prospect, buy more data about them, or come back in a quarter. That is the argument this taxonomy exists to make.

Glossary

Composite score: A single comparable rating derived from weighted factors.
Evidence completeness: The share of a model's weight supported by real evidence, discounted by reliability.
Reliability class: The provenance grade of a single piece of evidence: verified, provider, inferred, modelled, or cold start.
Neutral fill: Substituting a mid-range value for an unresolved factor to keep scores comparable.
Cold start: The state in which a model has no observed history and substitutes a proxy.
Freshness decay: The declining trustworthiness of evidence with age; the rate varies by factor.
Gate: A binary precondition evaluated outside the score.
Suppression: A record marking a contact or domain as ineligible to receive outreach.
Entity resolution: Determining which legal or operating entity a discovered record refers to.
Trigger: A dated, observable event raising the probability that a need is currently active.

Methodology and limitations

What this is. A conceptual taxonomy derived from StrykePoint's scoring architecture and from general outbound reasoning. Every claim here is a claim about model design and about the logic of evidence, not a measured finding about the world.

What this is not. Not a study, not an empirical analysis, not benchmarked. It reports no frequencies, rates or percentages, and it should not be cited as evidence that any failure mode is common or rare.

Single-vendor perspective: derived from one scoring architecture.
Category-shaped: written from B2B outbound to SMB and mid-market service businesses.
Non-exhaustive: nine modes are the ones this architecture can distinguish, not all that exist.
Untested empirically: reproducibility of the classes by independent operators is unmeasured.
Evolving: the model changes, and the taxonomy will lag it.

No customer data, workspace data, prospect records, or telemetry of any kind were used in producing this resource. No scoring weights or thresholds are disclosed.

Version, revisions, and citation

Version 2.0Last updated 2026-08-12Last reviewed 2026-08-12

Published by StrykePoint. Versioning is semantic: a major revision when a class or entry is added, removed or re-typed; a minor revision when an entry's content changes materially. Reviewed on a six-month cadence whether or not anything changes.

Revision history
Version	Date	Change
1.0	2026-08-12	Initial internal draft: nine modes in four classes.
2.0	2026-08-12	Per-entry structure added; 3.3 re-typed as a precondition; the three-concept distinction, glossary, methodology and citation sections added.

Suggested citation

StrykePoint. (2026). Why B2B Prospects Fail Scoring: A Taxonomy of Prospect Quality (Version 2.0). https://strykepoint.ai/resources/prospect-scoring-failure-taxonomy

Two readings, on every prospect

StrykePoint reports a composite score and an evidence completeness reading side by side, so an operator can tell the difference between a weak prospect and an unfinished one before approving a send.

How scoring works Talk to sales