An AI opportunity assessment scorecard completes the diagnostic suite this library built across its last two posts — the workshop surfaces candidates, the audit tests the substrate, and the scorecard decides order — and it opens by refusing the genre’s favorite trick, because the refusal is the instrument’s entire integrity: most scoring frameworks are precision theater — seven weighted criteria to two decimal places, producing “Use Case D: 7.83” from inputs that were gut feelings the whole time — and the false precision isn’t harmless decoration; it’s how a preference gets laundered into looking like a measurement. A score is a claim, per the audit framework’s doctrine, and claims need evidence and rubrics; where the evidence is coarse, the scoring must be honest about its coarseness. So the working scorecard runs the other way: two axes, six questions, three-band answers with written rubrics, every band traceable to something observed — an instrument deliberately cruder than the theater and radically more trustworthy, because its output survives the one test that matters: a skeptical operator asking “why is this first?” and receiving an answer made of their own numbers.
The instrument’s position, from the standing frame: prioritization is where the 92/1 gap gets decided in practice — according to McKinsey’s Superagency in the Workplace report (2025), 92% of companies plan to increase their AI investments over the next three years, yet only 1% describe their AI deployment as mature, and the maturity failure is substantially a sequencing failure: organizations that chose the impressive first project over the installable one, the CEO’s pet workflow over the leaking front door, the ambitious analytics build over the data condition that gated it. The scorecard exists to make the first choice boring and right — because per this library’s entire economics, the first install’s success funds, references, and politically enables everything after it, which means sequencing isn’t a planning nicety; it’s the practice’s compounding engine choosing its first gear. (All revenue figures in this post are illustrative business math, not guarantees; individual results vary.)
This guide is the instrument itself: the two axes and their six questions with band rubrics, the scoring session’s mechanics, the portfolio read (what the filled grid actually tells you), the tie-breakers and the overrides (politics, honestly handled), and the honest realities — including the spreadsheet whose decimals decided a quarter nobody can defend.
The Two Axes, Six Questions
Axis one: value evidenced — is the prize real and shown?
Question one: is the pain priced? Bands: priced from measurement (the baseline ran — the call log, the aging report, the touch audit; the annual number has a receipt) / priced from estimate (the workshop’s live arithmetic — real inputs, rough math) / unpriced (adjectives only — “so much time”). The rubric’s teeth: unpriced pains don’t rank until they’re at least estimated, which routes them to the baseline instrument first — the scorecard refusing to score is itself a finding.
Question two: who feels it, and how often? Bands: daily and broad (the leak touches every shift — the front door, the schedule) / regular but contained (monthly close pain, one team’s burden) / episodic (the quarterly fire, the annual event). Frequency is compounding’s raw material: the daily leak fixed pays every day.
Question three: does fixed have a metric? Bands: metric exists and is tracked (the no-show rate, the DSO, the answer rate — the improvement will be visible in a number the client already watches) / metric definable (measurable with the instruments the engagement would install) / unmeasurable as stated (“better morale” — real, and not rankable here). The measurement religion as a scoring gate: the practice’s whole proof economics require the fix to show.
Axis two: implementation reality — will the install actually land?
Question four: is the workflow owned and mapped? Bands: named owner, walkable process (someone runs this and can show you) / owned but foggy (the owner exists; the map lives in heads — the audit’s factor one, making its cameo) / orphaned or contested (nobody owns it, or three people think they do — the political quicksand no tooling survives).
Question five: is the substrate ready? Bands: installable now (the data condition, integrations, and governance posture support it — per the audit’s factors two and four) / one prerequisite (the hygiene remediation, the vendor-term fix — real but boundable) / gated (the install would sit on sand; the prerequisite is the project). The question that stops the analytics dream from outranking the data cleanup it requires.
Question six: what’s the blast radius if it wobbles? Bands: contained (after-hours, additive, reversible — the standing wedge shapes) / visible but recoverable (touches live workflows with rollback) / high-stakes surface (the anchor clients, the compliance clocks, the safety-adjacent flows — installable, per this cluster’s architectures, but never first). The question that encodes the whole library’s pilot doctrine: first installs are chosen for survivable failure as much as visible success.
Mechanics, the Portfolio Read, and the Overrides
The scoring session. Run with the client, never for them — the workshop’s parking lot (or the audit’s findings) as the candidate list, each use case walked through the six questions aloud, bands assigned with the evidence named (“priced from measurement — the two-week call log”), disagreements resolved by going to the artifact or downgrading the band (uncertain scores round down by rule — the conservative-attribution religion, applied to prioritization). Thirty to sixty minutes for a typical lot; the filled grid fits on one page, which is the point: an instrument the room can hold, argue with, and own.
The portfolio read. The grid’s quadrants tell the story without arithmetic: high-value / high-reality — the first-install shortlist (usually two or three candidates; the tie-breakers below choose); high-value / gated — the roadmap’s second chapter, with the prerequisites scoped as their own engagements (the audit-to-install funnel, formalized); low-value / easy — the quick wins deployed as adoption momentum where the change-scar inventory suggests the organization needs an early visible victory (a deliberate, named exception — not a default); low-value / hard — declined, in writing, which per the standing declination economics buys the credibility the shortlist spends. The tie-breakers, in order: blast radius (safer first), metric visibility (the fix the client’s existing dashboard will display wins), and momentum fit (whose workflow, politically, makes the best first champion — the change-management post’s stakeholder map getting its vote openly). And the override, honestly handled: sometimes the owner simply wants a particular project first — and the scorecard’s job is not to forbid the politics but to document them: the override is recorded as an override, the shortlist’s top candidate is noted as the instrument’s recommendation, and the divergence becomes data both parties can revisit when results arrive. A scorecard that pretends politics doesn’t exist gets quietly ignored; one that names it keeps its authority. We do not build the AI. We implement it — and the scorecard decides where implementing starts, on evidence the room can point to.
Why Honest Coarseness Wins
The structural recommendation: run prioritization on bands, rubrics, and named evidence — two axes, six questions, uncertainty rounding down — and refuse the weighted-decimal theater, because a ranking’s entire value is its defensibility, and defensibility comes from traceable claims, not from arithmetic performed on guesses.
The reasoning is structural:
- The precision illusion inverts the error bars: gut-feel inputs carry wide uncertainty, and averaging seven of them to two decimals doesn’t reduce the uncertainty — it hides it, producing rankings whose confidence is cosmetic; the banded instrument keeps the coarseness visible, which is what lets the room trust the parts that are actually firm.
- The rubrics are the anti-capture mechanism: written bands with evidence requirements resist the meeting’s loudest voice (the executive’s pet project can’t band “priced from measurement” without a measurement), which is precisely the protection a prioritization instrument exists to provide — the recruiting post’s transparency doctrine, applied to the roadmap.
- The six questions are the practice’s methodology, compressed: pricing (the leak check), frequency (the compounding logic), metrics (the measurement religion), ownership and substrate (the audit’s factors), blast radius (the pilot doctrine) — the scorecard teaches the client how the practice thinks while ranking their lot, which converts the instrument into the relationship’s operating language.
- And the suite closes by design: workshop → scorecard → audit-scoped prerequisites → the standing pilot playbook — every diagnostic exit lands on the practice’s map, every ranking’s first pick enters the four-gate pilot machinery, and the first install’s success (chosen, by the instrument, for exactly that) starts the compounding this whole library is built on.
I graduated from Vanderbilt. Almost went straight into investment banking. I spent years at Vanderbilt University reading the same labor reports and McKinsey decks that documented the trends now defining 2026 — and I came away with one inescapable conclusion: a salary has a ceiling. Inflation doesn’t.
I decided not to try and outrun inflation with a salary. I replaced my corporate salary by implementing pre-built AI tools we leverage — Intercom AI, Helios AI, and n8n at the core, plus the broader implementation stack — for service businesses with operational gaps they can’t fix on their own.
What Most Articles Won’t Tell You About Opportunity Scoring
A few honest realities:
The failure mode with your name on it is the Precision Illusion. It’s the spreadsheet that decided a quarter — seven criteria, custom weights, conditional formatting, “Initiative C: 7.83 vs Initiative F: 7.41” — presented to leadership as analysis and assembled, on inspection, from guesses: the “strategic alignment: 8” someone typed in a hurry, the impact score that was one person’s optimism, the weights themselves chosen to make the answer come out right (the theater’s open secret). The illusion’s damage is double and delayed: the organization sequences its AI program on cosmetic confidence — the 7.83 stalls on the substrate nobody scored honestly, the 7.41 that would have landed waits in the backlog — and when the stall arrives, the post-mortem finds a ranking nobody can defend, which discredits not just the spreadsheet but prioritization itself, leaving the next round to pure politics. The tell is any score whose evidence can’t be named in one sentence; the cure is the instrument this post builds — bands with rubrics, evidence spoken aloud at scoring, uncertainty rounding down, overrides documented as overrides — plus the sentence installed where the spreadsheet ambition reads it: a decimal point is a claim of knowledge; if we don’t have the knowledge, we don’t get the decimal.
The scorecard re-runs — it doesn’t retire. The quarterly re-score (new baselines arrived, prerequisites cleared, the first install’s results in) is the roadmap’s living cadence and the account’s expansion engine — the parking lot from the workshop, graduating band by band.
Rounding down is the instrument’s spine — defend it. Every scoring session pressures uncertain bands upward (“it’s probably priced-ish”); the conservative rule holds because the shortlist it produces must survive contact with the pilot’s four gates, and optimistic bands write checks the install has to cash.
The declined quadrant is a deliverable, not a leftover. The low-value/hard list, delivered in writing with reasons, is the engagement’s credibility ballast — per the standing declination economics — and the document the client forwards when someone upstream asks “did we consider X?” The standing arithmetic (3-5 clients = full-time corporate-equivalent income working a few hours a week once implementations stabilize) holds with the scorecard choosing each account’s first gear. You learn a skill instead of buying into a business model — and in prioritization, the skill’s signature is the ranking a skeptic interrogated and left agreeing with. (Illustrative math throughout; results vary.)
According to McKinsey’s Superagency in the Workplace report (2025), 92% of companies plan to increase their AI investments over the next three years, yet only 1% describe their AI deployment as mature. The consultants who own the prioritization category in 2026 are not the ones with the most sophisticated weights. They’re the ones whose rankings traced to evidence — and let the boring, right, defensible first install start the compounding the decimal theater never could.
Build the One-Page Grid This Week
The action sequence for ai opportunity assessment scorecard:
This week: The instrument drafted — two axes, six questions, band rubrics written out, the one-page grid formatted; the theater’s spreadsheet, retired.
This month: The first scoring session run — the workshop’s parking lot as the lot, bands assigned aloud with evidence named, the quadrant read delivered with the declined list in writing.
Per engagement: Uncertainty rounds down; overrides documented; the shortlist’s top pick enters the four-gate pilot; prerequisites scoped as engagements; the quarterly re-score calendared.
Ongoing: The grids accumulating into the practice’s cross-client pattern file (anonymized, per the standing rules); the rubrics sharpened; the decimals declined forever. (Illustrative trajectories; results vary.)
Prioritization is the 92/1 gap choosing its own fate — so rank with an instrument the skeptic can interrogate. Two axes. Six questions. Bands with receipts. Uncertainty rounds down. Politics named, never laundered.
The first install, chosen boring and right, is the compounding engine’s first gear — and the one-page grid is how the room agrees to shift into it.
Pick the industry. Take the first step. If you want to see the playbook fully in action – tap here to start.
