Pain/gain mapping — one page per role

TL;DR. Turn shadow notes into a one-page map per role with four quadrants: the work, the pain, the gain on offer, the pilot scope. The map is the contract — when the pilot wanders, you point to it. Anything longer than a page is a status doc, not a map.

The format#

A single A4 sheet per role, four quadrants:

• The work (top-left) — bullet list of the recurring tasks in this role.
• The pain (top-right) — where the work hurts: time, errors, judgment calls.
• The gain (bottom-left) — what becomes possible if the pain is removed.
• Pilot scope (bottom-right) — the 1–3 candidates this pilot will address.

Print it. Tape it to a wall. Refer to it.

How to fill it in#

• Pull the work from shadow notes — operator's words where possible.
• Pull the pain from the friction column of those notes.
• Write the gain in business terms (hours, deadlines hit, errors avoided), not feature terms.
• Pick pilot scope on three criteria: highest pain, lowest setup cost, most measurable.

The map's value is its compression. If you cannot fit it on one page, you have not picked yet.

Example: Finance Controller (illustrative)#

A worked example, anonymised:

• Work: month-end close pack, weekly variance commentary, audit queue, board pack inputs.
• Pain: Friday-afternoon variance commentary takes 4 hours; close pack rebuilt every cycle from scratch; audit queue handled inconsistently.
• Gain: variance commentary in 30 minutes; close pack assembled in 1 hour; audit queue with consistent quality.
• Pilot scope: variance commentary first (highest pain × lowest setup cost); close pack second.

That is the entire map. One page, four boxes.

Why one page#

• Sponsors read it.
• Operators recognise themselves in it.
• Engineers know what to build.
• Anything longer is a status doc, not a map.

The validation step#

Within 48 hours, send the map back to the operator with one question: "Did I get this right? Anything missing?" The operator's edits are diagnostic — they reveal gaps in the shadow.

Common mistakes#

• Generic pain ("we want to be more productive"). Push for specifics.
• Tool-shaped gain ("automate it with Cowork"). Push for outcome-shaped gain.
• Over-scoped pilot. Three candidates max; one is fine.
• Skipping the validation step. The whole point is alignment.

Where the map lives after the pilot#

• One copy in the workspace next to CLAUDE.md as pain-gain-map.md.
• One copy in the sponsor's folder.
• Re-reviewed at the six-week ROI checkpoint.

Tinkso's take#

The map is the contract. When a pilot wanders ("can Cowork also handle X?"), we point to the map. Yes, we will add X — in pilot two. No, we will not add it now. The discipline of one page is what makes this conversation possible.

Try this#

Take your highest-leverage role. Without running shadow sessions, fill in the four quadrants from memory. Then run two shadows. Compare. The diffs between the two versions are why the Observe beat exists.