For consultantsTemplates

Scope a job before you start, for consultants

List everything a job could involve, decide what is in and out, say what done looks like in a way you can check by looking, and set when to stop, all on one page.

This is the version for consultants. The steps and checks are the same as in the general template; each prompt carries one added line about your setting, and the examples are from your kind of work.

Sample results: runs and scores on this page are illustrations while testing is set up. How we test

  • Who it is for: Independent consultants and small-firm partners who get a loose request from a client and want a one-page boundary before they quote or start.
  • Time: 15 minutes

Download it, in the format you use

  • Prompt chain

    Copy each step into ChatGPT or Claude, one after another.

  • Checklist

    Follow the steps and check off the checks by hand.

  • Claude skill

    A folder Claude loads, so it follows the steps by itself.

  • n8n file

    Import into n8n and connect your own AI account.

Using a download:How to use this in n8nHow to install a Claude skill

How it runs

Select a part to see its prompt, input, output and check.

The same job, with and without the template

Sample

One plain prompt

Help me plan the work for this request from a client. (The same request and limits are pasted below.)

Here is a plan for reviewing the company's driver onboarding:

  1. Review the onboarding binder and the 3-week schedule.
  2. Analyze onboarding records for the drivers hired this year.
  3. Interview about 15 drivers and the HR team.
  4. Run a short survey of all current drivers.
  5. Benchmark the process against industry practice.
  6. Assess the outside safety training.
  7. Write a report with findings and recommendations.
  8. Build an implementation roadmap.
  9. Present the results at the board meeting on November 20.

Timeline: about 6 weeks of full-time work. Deliverables: a comprehensive report and a presentation. The review is complete when the client is satisfied with the findings.

Problem:It adds a driver survey, benchmarking and an implementation roadmap nobody asked for, never says what is out, defines done as the client being satisfied, and plans 6 weeks of full-time work against 9 billable days.

With this template

The same input, run through the steps.

The job

Review how the company onboards new drivers and describe what is wrong, in a paper for the board meeting on November 20.

In scope

  • The onboarding binder and the 3-week schedule.
  • The onboarding records for the drivers hired this year, about 60, pulled by the HR coordinator in up to 4 hours.
  • The paperwork new drivers sign.
  • A board paper that describes the problems, due November 13.

Not included

  • Recommendations for changes (later, if the board asks).
  • Talking to drivers (later, if access can be arranged).
  • A review of the outside safety training.
  • Changing the binder or the paperwork.
  • Presenting at the board meeting on November 20.

Done means

  • Binder and schedule: every section is marked as read in my notes. Check: open the notes and see each heading ticked.
  • Records: every record sent is logged with the hire date and the day onboarding finished. Check: count the log rows against the records sent.
  • Paperwork: every form a new driver signs is listed once. Check: count the list against the forms in the folder.
  • Board paper: each problem is named with the records that show it. Check: count the pages, two or fewer.

When I will stop

  • Time: 9 billable days, then stop even if the paper is unfinished.
  • Size: if the job needs more than the four pieces above, I will list the extra work and ask first.
  • Ask: before going further, I will ask the operations director who gives me access to the records.

Please confirm

Please confirm this scope, or tell me what to change, and who will give me access to the records.

  • Check:Each piece is marked stated or assumed: Nine pieces listed: 3 stated and 6 assumed.
  • Check:Each decision is in, out or later: Four in, three out and two later, each with a reason.
  • Check:Every piece has a way to check it: Each of the four pieces that is in has a meaning of done and a way to check it.
  • Check:The note is 300 words or fewer: About 290 words.
  • Check:Every done check can be made by looking: Each check is done by looking, counting or opening something; none uses a word like good or ready.
  • Check:Every in-scope piece reaches the note: All four pieces marked in are in the note, and the five left out (three out and two later) are under Not included.
  • For your read:You can live with the boundary: Would the operations director agree with leaving recommendations for later?

Runs and scores

Sample

Claude, Sonnet model

October 4, 2026, typical sample

  • Check:Each piece is marked stated or assumed

    Nine pieces listed: 3 stated and 6 assumed.

  • Check:Each decision is in, out or later

    Four in, three out and two later, each with a reason.

  • Check:Every piece has a way to check it

    Each of the four pieces that is in has a meaning of done and a way to check it.

  • Check:The note is 300 words or fewer

    About 290 words.

  • Check:Every done check can be made by looking

    Each check is done by looking, counting or opening something; none uses a word like good or ready.

  • Check:Every in-scope piece reaches the note

    All four pieces marked in are in the note, and the five left out (three out and two later) are under Not included.

  • For your read:You can live with the boundary

    Would the operations director agree with leaving recommendations for later?

Fixed: Nothing needed fixing: every check passed on the first pass.

Met 7 of 7

1 of these is for your read: the tool hands it to you.

ChatGPT, GPT model

October 5, 2026, tricky sample

  • Check:Each piece is marked stated or assumed

    Fourteen pieces listed: 5 stated and 9 assumed.

  • Check:Each decision is in, out or later

    Five in, five out and four later, each with a reason.

  • First pass failed:Every piece has a way to check it

    Two rows had nothing in How to check by looking: the third and the fifth.

  • Check:The note is 300 words or fewer

    The note is about 285 words.

  • Check:Every done check can be made by looking

    Each check is done by looking, counting or opening something; none uses a word like good or ready.

  • Failed:Every in-scope piece reaches the note

    The two leadership workshops are marked in, but the note does not name them.

  • For your read:You can live with the boundary

    Would the CEO accept that the three-year budget is left for later?

Fixed: The done step ran again once with the rule that every row needs a way to check by looking, and both blanks were filled, which cleared the first failure. The note step ran again once and still left out the leadership workshops, so the check failed again and the run stopped there and flagged it for you.

Met 6 of 7

  • Not met:Every in-scope piece reaches the note: The note still leaves out the two leadership workshops, which are marked in.

1 of these is for your read: the tool hands it to you.

n8n, import format 2.20

October 6, 2026, easy sample

  • Check:Each piece is marked stated or assumed

    Six pieces listed: 3 stated and 3 assumed.

  • Check:Each decision is in, out or later

    Three in, two out and one later, each with a reason.

  • Check:Every piece has a way to check it

    Each of the three pieces that is in has a meaning of done and a way to check it.

  • Check:The note is 300 words or fewer

    The note is about 150 words.

  • Check:Every done check can be made by looking

    Each check is done by looking, counting or opening something; none uses a word like good or ready.

  • Check:Every in-scope piece reaches the note

    All three pieces marked in are in the note, and the three left out (two out and one later) are under Not included.

  • For your read:You can live with the boundary

    Is the office manager content that a check of the layout is left for later?

Fixed: Nothing needed fixing: every check passed on the first pass.

Met 7 of 7

1 of these is for your read: the tool hands it to you.

The steps, and the check on each

  1. Step 1: List everything the job could involve

    Write out every piece of work the job might include, marking which ones the request actually names.

    Hands back: A numbered table of every piece of work, each marked stated or assumed.

    Check:Each piece is marked stated or assumed

    A plain rule. If it fails, run the step again (the n8n file does this for you).

  2. Step 2: Decide what is in, out or later

    Put every piece in, out or later, with a reason, using the time and help you have.

    Hands back: A table that puts every piece in, out or later, with a reason.

    Check:Each decision is in, out or later

    A plain rule. If it fails, run the step again (the n8n file does this for you).

  3. Step 3: Define done for each piece that is in

    For each piece marked in, write what done means and how to see that it is done.

    Hands back: A table that says what done means for each piece in scope and how to check it by looking.

    Check:Every piece has a way to check it

    A plain rule. If it fails, run the step again (the n8n file does this for you).

  4. Step 4: Decide when to stop

    Set the limits that tell you when to stop and ask, so the job does not grow.

    Hands back: Three short stopping rules (a time limit, a size limit and when to ask).

    No separate check on this step.

  5. Step 5: Write the one-page scope note

    Put the scope, the done checks and the stopping rules on one short page you can share.

    Hands back: A one-page scope note the person who asked can confirm.

    Check:The note is 300 words or fewer

    A plain rule. If it fails, run the step again (the n8n file does this for you).

After the last step

  • Check:Every done check can be made by looking

    A second look by the AI. If it fails, run the step again (the n8n file does this for you).

  • Check:Every in-scope piece reaches the note

    A second look by the AI. If it fails, run the step again (the n8n file does this for you).

  • Check:You can live with the boundary

    You check this one. If it fails, it is flagged for you.

What you need first

  • The job as it was given to you(needed)

    Paste the request or describe the job in your own words. Rough is fine.

  • Time and help you have(needed)

    How much time you can give it, any deadline, and who else can help.

Ways to adapt it

  • If the person who asked is available, send them the note before you start and change it with their answer.
  • To plan several jobs at once, run the template once per job and compare the notes side by side.
  • If your team has a standard form for requests, paste its headings into the last step in place of the ones shown.
  • To change the size of the note, change 300 in the last step's rules and in its check, both.

Change any step or check in the Editor, then download it again.

Words that fit consultants

  • A scope note is for you: what is in, what is out and when to stop. When the client must sign it, it is a statement of work: use Write a scope of work.
  • Anything a client asks for that sits under Not included is a change request, and it starts a new conversation about time and fee.
  • Billable days are the time limit here. Put them in the time you give, and the stopping lines will use them.
  • A deliverable is something you hand over, such as a paper or a file. A meeting you attend is an activity.
Try it with a sampleSample

A loose request from a client director with a few stated wishes and several gaps (who gives access, whether to talk to drivers, whether to recommend changes); the limits are tight.

The job as it was given to you
The operations director of a regional trucking company emailed me: "Can you look at how we onboard new drivers and tell us what is wrong with it? Our board meets on November 20." On the call she said onboarding takes about 3 weeks, that the paperwork is the part everyone complains about, and that the safety training is run by an outside provider. She did not say whether I should talk to the drivers, whether I should recommend changes or only describe the problems, or who will give me access to the records. The company hired about 60 drivers this year.
Time and help you have
9 billable days before November 20, board paper due November 13, working alone, the client's HR coordinator can pull records for up to 4 hours

The scoring guide

Sample

One criterion for each check, marked met or not met. (The guides in Quality rate a whole piece of work on three levels.) Under each one is how the three sample runs scored.

Download the scoring guide (Markdown)

  • Each piece is marked stated or assumed

    A plain rule

    Meets:Every piece of the job is marked stated or assumed.
    Does not meet:A piece has no mark, or a mark such as probably.
    Claude:Met
    ChatGPT:Met
    n8n:Met
  • Each decision is in, out or later

    A plain rule

    Meets:Every piece has a decision of in, out or later.
    Does not meet:A piece has no decision, or another word such as maybe.
    Claude:Met
    ChatGPT:Met
    n8n:Met
  • Every piece has a way to check it

    A plain rule

    Meets:Every piece has a done-means line and a way to check it by looking.
    Does not meet:A piece has a blank done-means or a blank way to check.
    Claude:Met
    ChatGPT:Met
    n8n:Met
  • The note is 300 words or fewer

    A plain rule

    Meets:The scope note is 300 words or fewer when you count them.
    Does not meet:The scope note runs past 300 words.
    Claude:Met
    ChatGPT:Met
    n8n:Met
  • Every done check can be made by looking

    A second look by the AI

    Meets:Every done check can be made by looking at, counting or opening something.
    Does not meet:A check rests on a feeling word such as good, clear, tidy or ready.
    Claude:Met
    ChatGPT:Met
    n8n:Met
  • Every in-scope piece reaches the note

    A second look by the AI

    Meets:Every piece marked in is named in the note, and every piece marked out or later is shown as not included.
    Does not meet:An in piece is missing from the note, or an out or later piece reads as included.
    Claude:Met
    ChatGPT:Not met
    n8n:Met
  • You can live with the boundary

    You check this one

    Meets:You read the note as the person who asked and they would agree with what is left out.
    Does not meet:The person who asked would object to something left out, and the Decision column needs a fix.
    Claude:For your read
    ChatGPT:For your read
    n8n:For your read

Claude: met 7 of 7 (including 1 for your read)

ChatGPT: met 6 of 7 (including 1 for your read)

n8n: met 7 of 7 (including 1 for your read)

Which one fits?

  • Use Write a scope of work when the scope has to be agreed with another person, as a document that says what is delivered, how each part is accepted and what is left out.
  • Use Break a project into steps when the job is clear and you need its steps in order, with a time and a person for each.

Related templates

All for Consultants

Plan

Review

Review

Last run
October 6, 2026: n8n, import format 2.20 (sample run)
Tools and versions
Claude, Sonnet model; ChatGPT, GPT model; n8n, import format 2.20
What was checked
The files, against each tool’s import format, and 3 sample runs, each check’s finding recorded.
Version
1.0
What changed
First release.

Report a problem with this templateHow we test