Detecting Optimization Theater

• Source: Report framework library

Encodes a Revenue AI Report framework so the agent applies the published method instead of improvising one.

1. 1.Copy the SKILL.md text below, or download the raw file.
2. 2.Create a folder named exactly detecting-optimization-theater in your agent's skills directory.
3. 3.Save the file inside that folder as SKILL.md.
4. 4.Ask the agent one of the trigger requests below.
5. 5.Check the output against what you already know before it leaves your desk.

• Our adoption dashboard is green and revenue has not moved in a year. Is this optimization theater?
• We have never turned an AI tool off. Should that worry me?
• The CFO does not believe our AI numbers. Audit the reporting before the board meeting.
• Calculate our Proof Gap for the quarter and tell me which band we are in.

• Score this single AI pilot out of five before we renew it.
• What order should we sequence these AI changes in?
• Find me three AI vendors for call recording.

---
name: detecting-optimization-theater
description: Test whether an AI program is producing reporting artifacts instead of revenue movement, and hand back the counter-pattern rather than a verdict alone.
---

# Detecting Optimization Theater

Test whether an AI program is producing reporting artifacts instead of revenue movement, and hand back the counter-pattern rather than a verdict alone.

## When to use this skill

- AI reporting has improved every quarter and the revenue-linked numbers have not.
- A year of AI spend has passed and nobody can answer what changed.
- No AI tool has been shut off in the reporting period and every initiative is described as a success.
- A CFO or board is challenging the credibility of an AI dashboard before a renewal cycle.
- A program review needs a naming device for the gap between reported activity and business outcome.

## Inputs to collect

- Quarter-over-quarter counts of AI reporting artifacts: pilot counts, adoption charts, time-savings estimates, dashboard views. Source: the AI program tracker and the last four board decks.
- Revenue-linked KPIs for the same quarters: pipeline, revenue, retention, margin. Source: the finance system and CRM, not a program summary.
- The list of AI tools shut off, replaced, or clawed back in the reporting period, with dates. Source: procurement records and the systems owner.
- Total AI spend for the quarter: AI platform contracts, AI seat licenses, AI-specific professional services, and internal headcount hours dedicated to AI enablement. Source: finance system and vendor invoices.
- AI-influenced pipeline for the same quarter, where an AI touch is verifiable in the CRM record and incrementality can be established. Source: CRM reports.
- For each function, whether operators were given time, tools, and permission to change the work. Source: interviews with the operators, not their managers.
- Whether the program has an explicit capability-building runway with a stated end date. Source: the program charter.

## Process

1. **Test signal one.** Determine whether AI reporting has increased quarter over quarter. Record the artifact counts, not the impression.
2. **Test signal two.** Determine whether revenue-linked KPIs are flat or declining over the same quarters. Use the same window for both signals.
3. **Test signal three.** Determine whether any AI tool has been shut off in the reporting period. The absence of reversals is the strongest single tell (https://www.therevenueaireport.com/frameworks/optimization-theater).
4. **Rule out capability building.** Check for an explicit twelve-month capability-building runway with a stated mandate (https://www.therevenueaireport.com/frameworks/optimization-theater). Capability building is not theater.
5. **Run the counter-pattern calculation.** Compute the Proof Gap as total AI spend in the quarter divided by AI-influenced pipeline in the same quarter, and run a Reversal Ledger review of what has been shut off (https://www.therevenueaireport.com/frameworks/proof-gap).
6. **Test the three permissions.** For each function, record whether operators were given the time, the tools, and the permission to change anything.
7. **Write the verdict and the correction.** State present or not present, name which signals fired, and hand back the counter-pattern with a cadence and an owner.

## Decision rules

- Declare Optimization Theater present only when all three signals coexist: AI reporting has increased quarter over quarter, revenue-linked KPIs are flat or declining, and no AI tool has been shut off in the reporting period (https://www.therevenueaireport.com/frameworks/optimization-theater). Two of three is a warning, not a diagnosis.
- Weight the reversal signal heaviest. Every large AI deployment produces tools and workflows that do not work, so a program that reports only success and never shuts anything off is reporting the theater, not the outcome (https://www.therevenueaireport.com/frameworks/optimization-theater).
- Override the dashboard when the ratio and the reversal signal agree. If the Proof Gap is above 4.0x and no tool has been shut off in the period, Optimization Theater is present regardless of what the adoption dashboard says (https://www.therevenueaireport.com/frameworks/optimization-theater).
- Use the published Proof Gap bands as stated. Under 1.0x is defensible to a board, 1.0x to 4.0x is a watch state requiring quarterly review, and over 4.0x on a rolling four-quarter basis triggers a Reversal Ledger review (https://www.therevenueaireport.com/frameworks/proof-gap).
- Calculate the Proof Gap on a rolling four-quarter view. Single-quarter results are noisy, and the rolling view is what a board should see (https://www.therevenueaireport.com/frameworks/proof-gap).
- Exclude time-savings estimates from the pipeline side. AI-influenced pipeline requires a verifiable AI touch in the CRM record and established incrementality (https://www.therevenueaireport.com/frameworks/proof-gap).
- Do not apply the label to a genuine capability-building phase with an explicit twelve-month runway. Reporting adoption as revenue is theater; capability building is not (https://www.therevenueaireport.com/frameworks/optimization-theater).
- Do not confuse this with AI washing. AI washing is a marketing pattern of describing existing capability as AI to raise valuation or price. Optimization Theater is an operating pattern of running real AI work and reporting the activity as outcome (https://www.therevenueaireport.com/frameworks/optimization-theater).
- Name the three permission failures when the pattern is present. It persists because nobody was given the time to change anything, nobody was given the tools to change anything, and nobody was given the permission to change anything, so nothing changed but the reporting did (https://www.therevenueaireport.com/frameworks/optimization-theater).
- Expect the activity-to-outcome distance to be large, and use it to calibrate rather than to accuse. In a single sample of GTM leaders, sales reported 87 percent more selling time against 13 percent higher quota attainment, sales development 76 percent more outbound time against 27 percent higher quota achievement, and marketing 92 percent higher productivity against 19 percent better lead-to-opportunity conversion (Scale Venture Partners, n=278, https://www.therevenueaireport.com/research/proof-gap).
- Do not read a zero-reversal record as good engineering. In a 10-country sample of 2,527 senior decision makers, 74 percent had already rolled back or shut down a deployed AI customer communications agent, and enterprises describing their guardrails as mature rolled back more often, not less, which is a detection effect (Sinch, The AI Production Paradox, fielded January to February 2026, vendor research, https://www.therevenueaireport.com/research/rollback).
- Record every case you find in the Optimization Theater Watch schema, and only where the activity gain and the flat or falling revenue metric are measured over the same window, by the same team, and both documented. Cases where the revenue metric was never measured are excluded, because an unmeasured outcome is not evidence of theater (https://www.therevenueaireport.com/data/optimization-theater-watch).

## Output requirements

Deliver the three-signal test, then the case rows using the published Watch schema so the finding is comparable across teams.

| Signal | Result | Evidence and window |
|---|---|---|
| AI reporting increased quarter over quarter | Yes or no | |
| Revenue-linked KPIs flat or declining | Yes or no | |
| No AI tool shut off in the period | Yes or no | |

| quarter | seat | activity_metric | activity_delta | revenue_metric | revenue_delta | constraint_missed | source_name |
|---|---|---|---|---|---|---|---|
| | | | | | | | |

Also deliver: the Proof Gap for the quarter and the rolling four-quarter view with the band it falls in, the three permissions answered per function, the verdict of present or not present with the signals that fired, and the counter-pattern with a named owner and a quarterly cadence.

## Verification loop

Validate the diagnosis before the label is used anywhere.

1. Confirm the activity window and the revenue window are identical. A gap measured across two different windows is an artifact, not a finding.
2. Confirm the revenue metric was actually measured. If it was never measured, remove the row, because an unmeasured outcome is not evidence of theater (https://www.therevenueaireport.com/data/optimization-theater-watch).
3. Confirm the Proof Gap inputs are finance-verified spend and CRM-verified pipeline, with no time-savings estimates substituted on either side.
4. Confirm the capability-building exclusion was tested and the result recorded, so a legitimate runway is not mislabelled.
5. Fix any failure and repeat checks 1 through 4 in full, because dropping a row usually changes the pipeline denominator and therefore the band.

Only proceed to state the verdict when all four checks pass on the same version of the analysis and all three signals have an evidenced yes or no. If one signal cannot be evidenced, report two of three as a watch state and name the missing evidence, rather than completing the diagnosis on an assumption. Flag the verdict for human review before it is used in a board setting, because the label carries a named accusation about the program's reporting.

## Quality checks

- All three signals answered with evidence and the same window.
- The Proof Gap shown as a quarterly figure and as a rolling four-quarter figure.
- No time-savings estimate appears on either side of the ratio.
- Every Watch row has both an activity metric and a measured revenue metric.
- The capability-building exclusion is explicitly tested, not assumed away.
- The three permissions are answered per function, with operator input.

## Limitations

- The diagnosis names a reporting pattern. It does not prove that the underlying AI work was worthless, only that the reported activity did not convert.
- Incrementality is hard to establish. Where it cannot be established, the Proof Gap denominator is soft and the band should be reported with that caveat.
- The absence of reversals is a strong tell rather than proof. A very young program has had less opportunity to shut anything off.
- The Optimization Theater Watch dataset publishes a method and column definitions. No observations have been published yet, so there is no benchmark median to compare a case against (https://www.therevenueaireport.com/data/optimization-theater-watch).

## Example input

A 900-person company, four quarters of AI reporting. Pilot count rose from 3 to 11. Adoption charts show 78 percent weekly active seats. Pipeline, net revenue retention, and gross margin are flat over the same four quarters. No AI tool has been shut off. AI spend in the quarter is 1.4 million dollars. Verified AI-influenced net-new pipeline in the quarter is 0.3 million dollars. Illustrative and synthetic, provided to show output shape.

## Example output

Signal one: yes. Pilot count and dashboard reporting both rose across four quarters.

Signal two: yes. Pipeline, net revenue retention, and gross margin are flat over the identical window.

Signal three: yes. No tool shut off in the period.

Proof Gap: 1.4 divided by 0.3 is 4.67x for the quarter, which is above the published 4.0x line that triggers a Reversal Ledger review (https://www.therevenueaireport.com/frameworks/proof-gap). Rolling four-quarter view is required before this goes to the board.

Verdict: Optimization Theater is present. All three signals coexist and the ratio is above the published 4.0x line with no reversals, which the framework states overrides the adoption dashboard (https://www.therevenueaireport.com/frameworks/optimization-theater).

Permissions: sales operators report no protected time to redesign the workflow, marketing reports tools but no authority to change the brief, and customer success reports neither. Three permission failures across three functions.

Counter-pattern: stand up a quarterly Proof Gap calculation and a Reversal Ledger review in the same meeting, owned by the VP of Revenue Finance, with the first review dated before the next renewal cycle. Flagging for human review before the label is used in board material.

## Rules of conduct

- Write for a Director, VP, or operator. Short sentences. Explain uncommon terms.
- Separate facts from assumptions. Never hide uncertainty.
- Do not invent numbers, benchmarks, quotes, or customer names.
- Do not send messages, change CRM records, or publish anything unless the user explicitly asks.
- Flag when a decision needs human review.

## Evidence

- https://www.therevenueaireport.com/frameworks/optimization-theater
- https://www.therevenueaireport.com/frameworks/proof-gap
- https://www.therevenueaireport.com/frameworks/proof
- https://www.therevenueaireport.com/frameworks/reversal-ledger
- https://www.therevenueaireport.com/data/optimization-theater-watch
- https://www.therevenueaireport.com/data/reversal-ledger
- https://www.therevenueaireport.com/research/proof-gap
- https://www.therevenueaireport.com/research/rollback
- https://www.therevenueaireport.com/research/spend-vs-attribution
- https://www.therevenueaireport.com/methodology/editorial-standards

## Cite this framework

Kvarfordt, Jonathan. "Optimization Theater." The Revenue AI Report. https://www.therevenueaireport.com/frameworks/optimization-theater

What does the Detecting Optimization Theater skill do?

Test whether an AI program is producing reporting artifacts instead of revenue movement, and hand back the counter-pattern rather than a verdict alone.

Where does the Detecting Optimization Theater skill come from?

Report framework library. It was written by The Revenue AI Report against a 12 criterion quality rubric and graded in an independent scoring pass.

Why was the Detecting Optimization Theater skill chosen for this library?

Encodes a Revenue AI Report framework so the agent applies the published method instead of improvising one.

When should the Detecting Optimization Theater skill not be used?

Do not use it for: Score this single AI pilot out of five before we renew it. Or: What order should we sequence these AI changes in? Or: Find me three AI vendors for call recording.

How do I install the Detecting Optimization Theater SKILL.md file?

Download the file, create a folder named exactly detecting-optimization-theater inside your agent's skills directory, and save the file inside it as SKILL.md. The agent loads it when a request matches the description.