Methods & Limitations

The Friction Assessment · storiessystems.com/methods · Instrument v2.1 · Effective July 15, 2026

What this is. The Friction Assessment is a structured self-assessment built from recurring patterns in our systems work with real organizations. It is not a scientifically validated instrument, a personality test, a grade, an audit, or an objective observation of the organization. It summarizes one respondent's report of how work currently moves through a business and returns a dated reading that can be retaken. This page explains how it works and where its limits are.

What it examines. The questions cover six proposed zones of operational flow: Lead & Inquiry Flow, Onboarding & Intake, Repetitive Work, Tools & Data, Visibility & Decisions, and Owner Dependence. Each zone is represented by two core questions everyone receives in the same order. Those twelve responses produce the zone scores and friction map. Deeper adaptive questions refine recommendations and context; they never move the map. The beta will test whether these questions and zones behave as intended; the current structure should not be read as proof that the six zones are distinct or complete constructs.

How answers are scored. Every question offers four behavioral descriptions, ordered from least to most friction, worth 0 to 3 friction points. A zone gets a score only when both of its core questions receive substantive answers — we never extrapolate a zone from one answer, because that would assume the missing answer matched the given one. A zone with incomplete evidence shows "Not enough evidence" on your map, honestly labeled.

When you can't answer. Every question includes "I can't give a substantive answer," with three options: don't know, doesn't apply, and prefer not to answer. None of these ever scores against you, and none is treated as a business failure. They differ in what they mean: don't-know and declined answers lower how confident the reading can be, because evidence is absent; doesn't-apply is information, not uncertainty. Separately, some scored answers describe not-knowing as an operating condition — "We wouldn't even know who's gone quiet" — and those are ordinary answers about how the system behaves, never confused with a skipped question.

When there isn't enough evidence for a reading. If fewer than five zones have complete responses, or more than two core answers are unresolved, the assessment stops before the adaptive questions and shows a Partial Reading: the available map, what the responses can support, and one useful next step. We do not name a Friction Signature when the response coverage does not meet that rule. A partial reading is not a failure, and the assessment does not infer why the information was unavailable.

How the signature is chosen. Your Friction Signature is shorthand for the most prominent pattern in your completed core responses — never a type, a diagnosis, or a certification of organizational health. The rules, in order: a knowledge-concentration condition (both relevant core answers indicating that critical knowledge lives in one person) is checked first, and if present, blocks the Owner Optional result. Owner Optional has the strictest requirements of any result: low total friction, no leaking zones, a smooth Owner Dependence zone with complete responses, and a supporting spot-check. Otherwise the signature maps from the highest-scoring complete zone. One governing rule sits above all of this: a signature must be earnable from responses every participant is invited to provide. Adaptive questions can qualify or reduce confidence in a result; they cannot create eligibility for one.

Exact ties. When two zones tie exactly at the top, we still lead with one name — chosen by a fixed display order (the order the questionnaire asks its zones), which carries no financial or severity meaning — and we say so on your results page: the responses do not rank one above the other. Your map always shows both.

Reading Confidence. The High/Moderate/Limited chip describes the evidence, not the business: how much was answered, whether coverage was complete, and whether the spot-check on your healthiest-looking zone held up. Confidently reporting that your business can't see something does not lower confidence — that's evidence, not its absence.

How the recommended steps are selected and sequenced. Every recommendation is triggered by specific answers you gave — each one shows you the answer that triggered it. Candidates are ranked by the strength and corroboration of their evidence, then ordered by published precedence rules (for example: capture a leak before building the dashboard that would only measure it; map the tool stack before bridging it; document a process before delegating it). Every item is labeled either Fix — a directly supported operating change — or Verify before building — an evidence-gathering test used when your answers show a gap that should be measured before anything gets built. We never fill a slot with an unrelated service or growth recommendation to make the number three. And the limit, stated plainly: the fix sequence is generated from specific reported conditions and predefined precedence rules. It does not establish that one friction zone caused another, and it may be revised when additional operational evidence becomes available. Determining causes, dependencies, and what's safe to change in your business requires evidence this assessment doesn't collect — that's what the Friction Review conversation and any subsequent engagement are for, and it's why the assessment doesn't pretend to do it.

The exposure estimate. When your own answers support it — a directly reported count of unanswered inquiries, and a customer-value range with a defined lower bound (or a figure you enter yourself) — we show a conservative illustrative exposure: count × the low end of your own range, capped for sanity against your reported inquiry volume. It is labeled what it is: opportunity exposed, not revenue lost. It omits close rate, lead quality, capacity, and seasonality on purpose, and you can adjust the assumptions or hide it entirely. If your answers don't support a responsible figure, we show the count, or nothing.

Comparison between readings. Every reading records its date, instrument version, scoring version, and — separately — the version of the logic that chose its recommended steps. The distinction matters: scoring governs what was measured, and recommendation logic governs what we suggested you do about it. The two are versioned independently, so we can improve the advice without disturbing the measurements you compare over time. Same scoring version: changes are calculated and shown. Compatible versions: shown side-by-side with direction only where the version rules permit. Incompatible versions: both readings shown, no arrows, no improvement or decline claims — we won't draw conclusions the versions can't support. Readings taken within a few weeks of each other may reflect normal variation rather than durable change.

What adapts to you, and what doesn't. Wording adapts by lane (business, church, nonprofit sites) and for solo operators — language only. Scoring, thresholds, signatures, and recommendations are identical for everyone. Nothing about your answers changes how prominently we invite you to a sales conversation; there is no hidden lead scoring, and no tier system deciding who sees what.

Validation status, honestly. At Instrument v2.1's beta launch, this is a practitioner-built structured self-assessment, not a validated psychometric instrument. We collect an optional perceived-fit response from participants and may record whether additional evidence reviewed in a Friction Review supported, qualified, or contradicted the reading. Perceived fit is a user-experience signal, not an accuracy or validity statistic. Review data is selection-biased because it exists only for people who book. We will report these signals with their definitions and limitations once sample sizes are meaningful, alongside future checks for response stability, item behavior, internal structure, and performance across business sizes and lanes. Any peer benchmarks will come only from explicitly opted-in, de-identified data, with published sampling, exclusions, and uncertainty. Until then, no comparative claim on a result is a peer claim.

Version history. Changes to questions, thresholds, or structure are versioned and listed here, with a plain statement of what stayed comparable. Instrument v2.1 — Effective July 15, 2026. This is the first beta version. Before beta data collection, six core items were revised to reduce compound answers, replace subjective or humorous anchors with observable behavior or bounded frequency, and clarify ordinal progression. No v2.0 responses were collected for longitudinal comparison.

Recommendation Methodology B1 — 2026-07-13. Recommendation logic has been refined without changing the assessment's scoring, zone states, or signatures. Organizations comparing readings across this update can compare their measured conditions directly. Only the recommended next steps may differ. Three refinements: a recommendation that could never be the strongest reason for itself was retired; two recommendations that were unreachable for reasons of internal wiring rather than evidence can now be reached; and one answer about where repetitive work lands now corroborates existing owner-dependence evidence — it strengthens a recommendation other answers already supported, and can never create one on its own.

Who answers for this. The methodology is owned by David Runnels, Stories & Systems. If a reading was wrong, unclear, or felt harmful, report it: report a wrong or harmful reading. That feedback goes to improving the instrument — not to a sales list.