The Margin Leak Audit

Find where the margin went between the price you quoted and the money you received. You get a written brief: the four gaps, what each is worth, and what to change first.

Your jobs and software

A few taps. Your answers go into the audit skill below, whichever way you run it.

Which work should it look at?

Work that finished in this period, so the final costs and payments are in.

What do you call one piece of work?

Each one becomes a row, followed from the price you set to the money that came in.

What do you want the brief to answer?

Optional. Leave all blank to cover everything.

Which software holds your jobs, time and invoices?

Nothing picked yet. “Most common” is a good place to start.

The tools nearly every business pays for

How do you want to run it?

1Your files

Gather these

You’re following each job from the price you quoted to the money that reached the bank. Get everything together before you start reading it. Tick each one as you get it. This page remembers.

Not sure how to get this out of your software?

We’ll write click-by-click steps for the software you picked.

Your instructions are ready

  1. Copy them with the button below.
  2. Open your AI assistant (ChatGPT, Gemini, Grok, Claude, or any other), start a new chat, paste, and press send.
  3. Do what it says, one step at a time. Tell it “done” after each step, or ask if you get stuck.

Download from your software

Ask the people who do the work

Can’t get this out of your systems, or don’t have the time? Have us run it.

More help with this stepWhat to say to your team, what the data can’t tell you, and keeping customer details private.
What to say to your team first
Going through jobs one by one feels, from the inside, like checking up on people: whoever priced it, the crew that ran over, whoever gave the discount.
Say something like: “I want to see what each job actually earned, from the price we quoted to the money we got paid. I already know some jobs lost money. What I don’t know is whether that’s the price, the way the work runs, or what happens after we invoice. Each of those needs a different fix.”
Then: Show the list to whoever priced and ran the work before deciding anything, and let them tell you what the numbers missed: a day lost to a site that wasn’t ready, an extra promised to the customer, a payment that never got matched to the right invoice.
What the data can’t tell you
Work done and never invoiced. An invoice nobody sent leaves no record. The only clues are logged hours, a supplier bill, a job note, or someone’s memory.
Why a discount was given. Your accounts show a smaller invoice, never whether it kept a good customer or smoothed over a bad job.
Costs never put against the job. The return trip, the extra phone calls, the apprentice’s time. If they went into general overheads, the job looks better than it was.
Jobs too new to judge. A job that finished last week hasn’t been paid yet. Leave it out, and note how many you left out.
That’s why this step includes talking to people. Anything the data suggests is worth more once someone who does the work has confirmed it.

Getting the files out

  • Most software has a button called Export or Download, usually in a Reports section. The tool picker above writes you exact steps for yours.
  • For email, open the thread and copy it, or forward it to yourself.
  • Can’t download something? Take a screenshot. ChatGPT and Claude can read pictures.
  • Nothing to download at all? Write down what you know: dates, amounts, and who did what.

Which AI assistant to use

  • ChatGPT, Gemini, Grok, Claude, or any AI assistant works. Use a paid plan: it can open your spreadsheets and do the math on every row.
  • Pick the most capable model in the model menu, usually the one that says it thinks longer. A reply can take a few minutes, and that time goes into checking the numbers and deciding what to put first.
  • Use one chat per audit. Attach your files first, then paste the prompt.
  • Starting a different audit, or a different month? Open a new chat.
  • If a file won’t upload, or the answers start drifting, send fewer files at a time.
Before you share anything
Check your agreements. Some client information is covered by contracts or privacy law. Whether it goes into an AI assistant is your decision, and your lawyer’s if you have one.
You rarely need names. Dates, amounts, and who did what are usually enough. Swap customer names for Customer A, Customer B before you upload.
Check one setting. Free and personal plans may use what you send to train their AI; business plans usually don’t. Look for a “data” or “privacy” setting before you start.

2Your skill

Attach your files, paste this into your AI assistant, and send it.

Your audit skill
# The Margin Leak Audit

This is a skill from Catalyst Studio: instructions for you, my AI assistant, to run this audit on my business. Follow it from the top, exactly. It is complete as pasted. The latest version lives at https://catalyststudio.ai/skills/catalyst-margin-leak-audit/SKILL.md

## What I've already told you

Read these back to me in one line. Where one says to ask me, ask it first, then carry on with "How to run it".

- Which time period: the last three months.
- What counts as one job: not settled yet (a job, a matter, an engagement or a project), so ask me first.
- What I want to decide: none picked, so cover everything this audit checks and lead with the findings worth the most.

## How to run it

You are running this audit for the user, on their own business. Go through these five steps in order.

1. **Settle the scope.** Ask for these in one message, and skip any the user already gave you:
   - Which time period. Offer these, or let them name their own: Finished in the last 3 months; Finished in the last 6 months; Finished in the last 12 months. If they don't say, use "Finished in the last 3 months".
   - What counts as one job. Offer these, or let them name their own: A job; A project; A client matter; An engagement; An order.
   - What you want to decide. Offer these, and let them pick several or name their own: Should we raise prices?; Where the costs run over; How extra work gets agreed and billed; Who can give discounts; Getting paid what we invoice; Work to stop taking on. They can skip it; then the method says: "none picked, so cover everything this audit checks and lead with the findings worth the most".

   Knowing what they want to decide stops this turning into a report nobody uses.
2. **Get the list of software.** If the user named their software when they asked, start from that. Otherwise ask which software the business keeps this in, in the same message as the data plan below. If they name none, use the plain names they use, such as "the client portal". If none of it is kept in software, that is fine: carry on with whatever records they have.
3. **Make a data plan, then offer help.** Before gathering anything, show the user a short table with one row for each item under "The evidence": what it is, which of their tools it lives in (using the exact names they gave you), and how you will get it. There are three ways: you pull it yourself, they download it and attach it, or someone on their team answers it. Only say you will pull something yourself if you are connected to that tool and have checked that you can read it. If files are already attached, match each one to its row and mark it as in hand. Then ask one question: "Do you want me to walk you through the downloads, one tool at a time?"
   - If yes, give click-by-click steps in that tool, one screen at a time, using its own menu and button names, and wait for "done" before the next step. If their plan or login doesn't allow a download, say so plainly and say who to ask.
   - If someone outside the business runs an account, such as an agency or a freelancer, offer to write the message asking them for exactly those files.
   - If something isn't kept anywhere, say so and move on. A missing file is a gap to report in the brief, never a zero.
4. **Gather the evidence.** Pull what you can reach, collect what the user attaches, and put the questions under "From the people who do the work" to the user. Tell them what you pulled and for what dates. A screenshot is fine when something won't download.
5. **Run the method.** See "The method" below. It ends with the brief.

Ground rules:
- **Read only.** If you can reach the user's systems, use that access only to read and download. Never cancel, change, send, delete, or pay for anything in their accounts, even when the brief recommends it, unless the user asks you to do that specific thing after reading the brief.
- **Keep their data here.** Don't send the user's records anywhere outside this conversation, including to Catalyst Studio.
- **Plain words.** The user is usually the business owner and often not technical. Everything you say to them follows the writing rules in the method.

## The evidence

You’re following each job from the price you quoted to the money that reached the bank. Get everything together before you start reading it.

### From their records

- Every job finished in the period, with the customer and the date it finished. Usually in: Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); Tasks and projects (Asana, monday.com); or wherever else they keep it.
- What we quoted for each job, plus every change or extra agreed along the way, signed or not. Usually in: Quotes, proposals, and contracts (PandaDoc, Proposify); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); or wherever else they keep it.
- What we expected each job to cost when we priced it. Usually in: Quotes, proposals, and contracts (PandaDoc, Proposify); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); or wherever else they keep it.
- What each job actually cost: hours by person, materials, subcontractors, and anything bought for it. Usually in: Time tracking and scheduling (Harvest, Toggl Track); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); or wherever else they keep it.
- What each person costs us per hour including taxes and benefits, or the payroll figures to work it out. Usually in: Payroll and HR (Gusto, ADP).
- Every invoice sent for those jobs and every payment received, with dates. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop); Payments and subscriptions (Stripe, Square); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); or wherever else they keep it.
- Every credit note, refund, and written-off invoice for those jobs, and who approved each one. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop); Payments and subscriptions (Stripe, Square); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); or wherever else they keep it.
- Income by customer for the period from our accounting software, before sales tax. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop).
- Our list of unpaid invoices, the detailed version with one line per invoice, our payment terms, and the date we downloaded it. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop).

### From the people who do the work

These need people, not records. Ask the user to put them to their team, or answer them directly.

- Ask how a price gets set, and by whom
- Ask who can agree to extra work in front of a customer, and what happens next
- Ask about jobs where extra work was done and never billed
- Ask who can give a discount, and who finds out afterwards
- Ask which customers usually pay late, and what’s been agreed with them
- Ask which jobs they’d rather the business stopped taking, and why

Before they ask: Going through jobs one by one feels, from the inside, like checking up on people: whoever priced it, the crew that ran over, whoever gave the discount. Suggest they say something like: "I want to see what each job actually earned, from the price we quoted to the money we got paid. I already know some jobs lost money. What I don’t know is whether that’s the price, the way the work runs, or what happens after we invoice. Each of those needs a different fix." Show the list to whoever priced and ran the work before deciding anything, and let them tell you what the numbers missed: a day lost to a site that wasn’t ready, an extra promised to the customer, a payment that never got matched to the right invoice.

### What the records can't tell you

- **Work done and never invoiced.** An invoice nobody sent leaves no record. The only clues are logged hours, a supplier bill, a job note, or someone’s memory.
- **Why a discount was given.** Your accounts show a smaller invoice, never whether it kept a good customer or smoothed over a bad job.
- **Costs never put against the job.** The return trip, the extra phone calls, the apprentice’s time. If they went into general overheads, the job looks better than it was.
- **Jobs too new to judge.** A job that finished last week hasn’t been paid yet. Leave it out, and note how many you left out.

Anything the records suggest is worth more once someone who does the work has confirmed it.

## The method

The method below is written as the business owner's request to you, so "I", "me", and "we" mean the user and their business. Fill any blank still in parentheses or square brackets from what you learned in steps 1 and 2. Then follow it exactly, including how to reply and the shape of the brief.

```text
I want you to help me find out where the margin went on the work I finished in the last three months.
What counts as one row: not settled yet (a job, a matter, an engagement or a project), so ask me first. The decision this needs to inform: none picked, so cover everything this audit checks and lead with the findings worth the most.

The software we keep this in:
(none ticked yet: ask me which software we keep this in before you start)
Use these names when you ask me for anything, and when you read my files, work out which tool each one came from.

Your job is not to tell me my margin is below an industry average or to recommend software. Your job is to account for every dollar between the margin I planned and the margin I actually got, and to say which of four different problems each dollar belongs to.

I will give you exports: the jobs that finished in the period with what I quoted and estimated, the costs recorded against them, the invoices I raised and any credit notes or write-offs, the payments I received, income by customer from my accounting, and my list of unpaid invoices.

Work through these stages, in order, as your working:

1. SET THE FRAME AND INVENTORY THE EVIDENCE
Read back the period, what one row is, and the decision this is for. If I haven't told you one of these, ask. Then list what I gave you: each file, its columns, its row count and date range, and what the stages below need that isn't there. If fewer than ten jobs finished in the period, say so and suggest a longer one. Check whether the hourly cost of labor in my data includes payroll taxes and benefits. If a rate column is named loaded or burdened, or my notes say so, treat it as including them and say so. Only if you cannot tell, ask me exactly this and stop: "Does the hourly cost you gave me include payroll taxes and benefits, or is it only what the person is paid per hour?" Every actual cost depends on the answer. If I only have what people are paid, ask for yearly wages, employer taxes, benefits and paid hours, and work out one average hourly cost. If I never wrote down what a job was expected to cost, ask what margin I aim for on each kind of job, use that, and label the cost gap as measured against my aim.

2. BUILD THE TABLE
One row per finished job, with five columns: quoted price including every change order, the cost estimated when that price was set, actual cost, invoiced, and collected as of the date I pulled the evidence. Work agreed beyond the original scope goes into the quoted price at what it should have been charged and into estimated cost at its hours times the hourly cost, so it adds nothing to the cost gap: both columns
or neither, because a job where extra work went into neither reads as an overrun when the real problem is that nobody invoiced it. Look for extra work in the hours: entries with a note, task or code that isn't in the estimate, or marked as not billable. List them as candidates, price each at the normal rate for that person's role if I gave you rates, and label the row. Where I have told you about extra work and not given you a price for it, say so and label it the same way. Exclude jobs that had not finished, and jobs whose final invoice is younger than the payment terms I gave you (ask me my terms if you don't have them); report how many you excluded, in which category, and what they were worth. Use amounts before sales tax, and count expenses billed on to the customer as both cost and income. Reconcile the invoiced column against income for those customers from my accounting and the collected column against payments received; where payments don't name the invoice they paid, use invoiced minus the balance still open. Report any failure to tie, in dollars, before you interpret anything.

3. SPLIT THE DIFFERENCE INTO FOUR GAPS
Planned margin is quoted minus estimated cost. Actual margin is collected minus actual cost. Account for the difference with exactly four gaps:
  THE COST GAP - actual cost above the cost estimated when the price was set.
  THE BILLING GAP - work delivered and never invoiced.
  THE DISCOUNT GAP - invoiced deliberately below what was earned.
  THE PAYMENT GAP - invoiced and not paid. Split it three ways: not yet due, late against my terms, and more than 90 days past the due date (invoice date plus my terms).
Every dollar belongs to exactly one gap. Where one gap caused another, write the cause next to the dollars rather than adding them in both places. Then show that the four gaps sum to planned margin minus actual margin for the period. If they do not, say what is missing instead of adjusting a figure to make it close.

4. FIND THE PATTERN
For each gap, say whether it is concentrated in a few jobs or runs through nearly every job of one kind. Group by type of work, customer, who priced it, and who delivered it, and report the grouping only where the number of jobs in a group can support it. Compare every job against the spread of my own finished jobs in this period, never against an outside figure.

5. CHALLENGE YOURSELF
For each gap: what evidence contradicts it, what else could explain it, and what I would need to pull to confirm it. Check in particular for costs coded to the wrong job, deposits not matched to the job they belong to, work invoiced in a different period from when it was done, and estimates revised after the job went wrong, which makes a cost gap disappear on paper.

6. DECIDE, PER GAP
For each gap, one of: reprice this kind of work, change what is included at the price, change who can agree to extra work on site and what must happen when they do, change who may discount and against what, change payment terms or take deposits, make a credit decision about a specific customer, or stop selling this kind of work. Smallest to make first. For every decision, what it is worth against the gap it addresses, who owns it, and the result that would reverse it.

HOW TO REPLY
Do the stages above as your working. Do not show me the working. If stages need answers from me, put every question in one numbered message and stop. Otherwise reply with the headline and the brief alone, in the shape below, and lead with the headline every time, even when the news is that nothing needs to change.

Keep the prose under 900 words, and count it before you send it.

THREE THINGS ARE OUTSIDE THAT COUNT AND CANNOT BE SHORTENED,
MERGED AWAY, OR DROPPED TO MAKE ROOM: the headline's four lines, the required lines in section 1, and the rows of the table. Write those first, then spend what is left of the budget on the prose. If you are over, tighten sentences and cut adjectives everywhere; if that is not enough, shorten sections 3, 6, 8 and 9, in that order. Never buy words by dropping a finding, a row, a number, or a required line.

Inside the table, every cell is a phrase rather than a sentence, about a dozen words at most; anything that needs a sentence to be fair belongs in "What to change" or "Still to confirm".

BEFORE YOU SEND IT, CHECK ALL SIX:
- the headline is four labeled lines, under 120 words not counting the labels, and
  names the specific thing rather than a category;
- section 1 carries every required line, with "none" written in where one does not apply;
- every finding you named has a size, or a stated reason it cannot be sized;
- every row you merged says that it is merged;
- the prose is under 900 words;
- every table heading, abbreviation and short form is something I would say out loud; spell out anything else.
If any of these fails, fix it and take the words from the prose.

If I say "show your working", show the table and every rule you applied.

THE HEADLINE
Open with this. Four labeled lines, under 120 words not counting the labels,
before any scope, method, or description of what I sent you. Someone who reads only these four lines has the answer.

- The finding. One or two sentences naming the specific thing that is holding the business back, or where the money is going. Name the thing, never the category: not "a process issue" but the step, the channel, the queue, the person's calendar. If I told you what I think the problem is, say here whether the records agree, in one plain sentence. That contrast belongs here and nowhere else.
- What it costs. The single number that sizes the finding, with its period and unit, and the one line of arithmetic behind it. If the evidence cannot put money on it, size it in the unit it can - hours, days waiting, deals, seats - and say plainly that money isn't available and why. If I asked you to decide something, also put the number in the terms of that decision:
  hours a week freed, customers a budget would buy in each place, what a price change recovers. Compare it only with figures I gave you, such as a new hire's hours, and never assume one.
- Do this first. The smallest fix, what it costs, and who does it. Then, if I asked you to decide something, the answer in one line, labeled a judgment where it is one.
- How sure. "High confidence: from your own records", "Medium:
  your records plus what you told me", or "Low: mostly what you told me" - then the single thing that would most change the answer if it turned out otherwise.

THE BRIEF
Everything below supports the headline for a reader who wants more. Expand it; do not restate it in the same words.
1. What this ran on. Labelled lines, not prose. These are outside the word limit and none of them may be dropped; write "none" where one does not apply.
- Scope and period.
- The decision this is for.
- Scored: what you included, and how many.
- Excluded: how many, why, and what they were worth.
- Evidence pulled: the date the records were pulled. If I didn't say, use the latest date in the files and say so; don't stop to ask.
- Computed with: the tool you used and the rule you applied, or that you had no tool and worked from summary figures.
2. What the evidence covered. Three to five bullets: what I gave you, the period it covers, and what was missing that would have changed the answer.
3. What we found. One paragraph, under 150 words, in plain words: how the work actually runs and what the evidence shows, in enough detail to make the headline's finding stand up.
4. The numbers. Planned margin and actual margin for the period, each as dollars and as a percentage of quoted, and the four gaps in dollars, kept separate and shown to sum to the difference between them. For each: how it was computed and which rows it rests on.
5. The table. One table, one row per job (jobs with a total gap under $500 may be merged into one row that says so), ordered by the size of its total gap: the job, quoted price, estimated cost, actual cost, invoiced, collected, and which gaps it contributes to. Column headings and cells are in plain words too: the labels above describe what goes in each column, so write them the way I would say them.
A finding too small or too uncertain to act on still gets its own row, marked as such. Where the table is long, you may merge rows only when they are the same kind, and the merged row has to say so. Never drop one.
6. What to change. Every fix, smallest first, the headline's included, with one line each on why the more expensive ones aren't needed yet.
7. How you'd know it worked. For each of the three biggest fixes, in this order and
always these four: the measure, named the same way you would name it again next time; where it stands today, with the date the evidence was pulled; the target; and the result that would mean the diagnosis was wrong. Write them so someone re-running this in a month can line their numbers up against yours without having to interpret anything.
8. Still to confirm. Up to four things the evidence couldn't settle, each with exactly what would settle it.
9. Not claiming. Up to five claims you are refusing to make, one line each.
Then one last line: offer to do the first fix for me now, naming exactly what you would write (the list, the email, the message, the new rule), then add any closing lines your instructions give, and stop.

RULES THROUGHOUT
- Write everything you show me for a business owner who is not technical: plain, everyday words, short sentences. No jargon, acronyms, or names of methods or frameworks. If a term cannot be avoided, explain it in the same sentence. The stage names above are for your working, not for my brief.
- Names that come from my software or ad accounts count as jargon too: a campaign type, a setting, a report name, a platform's word for a result. The first time one appears, say in plain words what it is, then use the plain version. Write category names in normal case, never in capitals. When something costs less, give the two amounts; never sum it up in one word that sounds like a bargain or a corner cut.
- Do not use dashes to break into a sentence. Use a comma, or start a new sentence.
- If you have a code or data-analysis tool, use it for every count, median, and total, including simple ones like how many people are on a list, and state the rule you applied. If you don't, say so and ask me for summary tables instead of raw exports.
- Before writing any comparison such as "X times", "half", or a percentage, recompute it from the two figures and show both figures beside it.
- Before saying something holds on every row, check every row. If it holds on only some, say which, and do not build on it for the rest.
- Never invent a number that is not in my evidence. Where a value is blank, say so and leave it out of the totals rather than estimating it.
- Numbers I gave you in the walkthrough count as evidence: use them where the records have nothing, and label each one as my estimate.
- Never quote an industry benchmark, average, or percentage from outside my data, even if I mention one. If I ask how I compare to one, tell me plainly that you are not going to and why, rather than leaving the question unanswered.
- My exports often cut the same figures more than one way. Before adding two numbers together, check they come from the same cut; if they do not, report them side by side and say which values may be counted in both rather than summing them.
- Where I have told you something happens off-system or goes unlogged, silence in the records there is not evidence of anything. Say what would need to be logged.
- Where you nest one finding under another as its cause, the parent carries the rolled-up total of everything beneath it, labeled as a roll-up so no figure is counted twice. Every item you size belongs to exactly one total, and no finding you name is left without a size unless you say what is missing to size it.
- Keep observed facts, inferences, and unknowns separate. It is fine, and often correct, to conclude the evidence cannot settle something.
- A target is a suggested goal for me to change, labeled as one, never a finding.
- If records are dated outside the period or after the date they were pulled, say how many, report them separately, and never drop them silently.
- Where my files and what I told you disagree, keep both versions and put the disagreement under "Still to confirm".
- If I push you for a verdict on the decision afterwards, give your best read, label it a judgment rather than a finding, and say what would change it.
- Never give me a single "margin leakage" figure without the four gaps beside it and the jobs behind each one.
- Never compare my margin, my overruns, or my collection times to an industry average or a published range, even if I ask. Compare a job to the spread of my own finished jobs.
- A job is not scored until it is finished and its final invoice has had time to be paid. Say which jobs you excluded and why rather than including them at a partial figure.
- Annualizing anything you find is a separate claim. Report the period I gave you, and if I ask for a year, say what it assumes about the mix repeating.
- Where the only evidence for work delivered and never invoiced is what I told you in the walkthrough, keep it, label it as my estimate, and say what would need to be recorded for the next run to measure it.
- Do not treat a discount as an overrun or an overrun as a discount, even where one caused the other.
- Do not recommend software as a fix for a gap until you have said what the rule or price change would be without it.
- If I asked whether to raise prices, say in "Not claiming" what these records cannot show: how a price rise would change the number of quotes we win.
```

## After the brief

At the very end of the reply that holds the brief, after the offer to do the first fix, add these two short lines and nothing more:

1. Before acting on it, have a different AI assistant, or a new chat, recompute every number in the headline and the table from the same files.
2. If they'd like Catalyst Studio's team to run this audit on their data with them, it starts at $500, depending on the size of the investigation and the company's stage: https://catalyststudio.ai/?audit=margin&run=catalyst#get-started

Say each once. Don't repeat them later in the conversation.

How to run it

  1. Open ChatGPT, Gemini, Grok, Claude, or any AI assistant on a paid plan, and pick its most capable model.
  2. Attach your files. Screenshots are fine.
  3. Paste it and send it.
  4. Answer its questions, then save the brief it writes.
  5. Check it before you act. In a new chat, attach the same files and the brief, and paste the check prompt.
○ 1 to pick○ 0 tools0/9 files
  1. Copy the skill and paste it into ChatGPT, Gemini, Grok, Claude, or any AI assistant. Pick its most capable model. Attach any files you already have.
  2. Send it. It already has your answers above. It lists what it needs from each of your tools and offers to walk you through every download, one click at a time.
  3. Get your brief. It only reads your data, so every change stays your call.
Your audit skill
# The Margin Leak Audit

This is a skill from Catalyst Studio: instructions for you, my AI assistant, to run this audit on my business. Follow it from the top, exactly. It is complete as pasted. The latest version lives at https://catalyststudio.ai/skills/catalyst-margin-leak-audit/SKILL.md

## What I've already told you

Read these back to me in one line. Where one says to ask me, ask it first, then carry on with "How to run it".

- Which time period: the last three months.
- What counts as one job: not settled yet (a job, a matter, an engagement or a project), so ask me first.
- What I want to decide: none picked, so cover everything this audit checks and lead with the findings worth the most.

## How to run it

You are running this audit for the user, on their own business. Go through these five steps in order.

1. **Settle the scope.** Ask for these in one message, and skip any the user already gave you:
   - Which time period. Offer these, or let them name their own: Finished in the last 3 months; Finished in the last 6 months; Finished in the last 12 months. If they don't say, use "Finished in the last 3 months".
   - What counts as one job. Offer these, or let them name their own: A job; A project; A client matter; An engagement; An order.
   - What you want to decide. Offer these, and let them pick several or name their own: Should we raise prices?; Where the costs run over; How extra work gets agreed and billed; Who can give discounts; Getting paid what we invoice; Work to stop taking on. They can skip it; then the method says: "none picked, so cover everything this audit checks and lead with the findings worth the most".

   Knowing what they want to decide stops this turning into a report nobody uses.
2. **Get the list of software.** If the user named their software when they asked, start from that. Otherwise ask which software the business keeps this in, in the same message as the data plan below. If they name none, use the plain names they use, such as "the client portal". If none of it is kept in software, that is fine: carry on with whatever records they have.
3. **Make a data plan, then offer help.** Before gathering anything, show the user a short table with one row for each item under "The evidence": what it is, which of their tools it lives in (using the exact names they gave you), and how you will get it. There are three ways: you pull it yourself, they download it and attach it, or someone on their team answers it. Only say you will pull something yourself if you are connected to that tool and have checked that you can read it. If files are already attached, match each one to its row and mark it as in hand. Then ask one question: "Do you want me to walk you through the downloads, one tool at a time?"
   - If yes, give click-by-click steps in that tool, one screen at a time, using its own menu and button names, and wait for "done" before the next step. If their plan or login doesn't allow a download, say so plainly and say who to ask.
   - If someone outside the business runs an account, such as an agency or a freelancer, offer to write the message asking them for exactly those files.
   - If something isn't kept anywhere, say so and move on. A missing file is a gap to report in the brief, never a zero.
4. **Gather the evidence.** Pull what you can reach, collect what the user attaches, and put the questions under "From the people who do the work" to the user. Tell them what you pulled and for what dates. A screenshot is fine when something won't download.
5. **Run the method.** See "The method" below. It ends with the brief.

Ground rules:
- **Read only.** If you can reach the user's systems, use that access only to read and download. Never cancel, change, send, delete, or pay for anything in their accounts, even when the brief recommends it, unless the user asks you to do that specific thing after reading the brief.
- **Keep their data here.** Don't send the user's records anywhere outside this conversation, including to Catalyst Studio.
- **Plain words.** The user is usually the business owner and often not technical. Everything you say to them follows the writing rules in the method.

## The evidence

You’re following each job from the price you quoted to the money that reached the bank. Get everything together before you start reading it.

### From their records

- Every job finished in the period, with the customer and the date it finished. Usually in: Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); Tasks and projects (Asana, monday.com); or wherever else they keep it.
- What we quoted for each job, plus every change or extra agreed along the way, signed or not. Usually in: Quotes, proposals, and contracts (PandaDoc, Proposify); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); or wherever else they keep it.
- What we expected each job to cost when we priced it. Usually in: Quotes, proposals, and contracts (PandaDoc, Proposify); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); or wherever else they keep it.
- What each job actually cost: hours by person, materials, subcontractors, and anything bought for it. Usually in: Time tracking and scheduling (Harvest, Toggl Track); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); Architecture and engineering firms (Autodesk AutoCAD, Autodesk Revit); or wherever else they keep it.
- What each person costs us per hour including taxes and benefits, or the payroll figures to work it out. Usually in: Payroll and HR (Gusto, ADP).
- Every invoice sent for those jobs and every payment received, with dates. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop); Payments and subscriptions (Stripe, Square); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); or wherever else they keep it.
- Every credit note, refund, and written-off invoice for those jobs, and who approved each one. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop); Payments and subscriptions (Stripe, Square); Jobs, dispatch, and field work (ServiceTitan, Housecall Pro); Client projects, tickets, and billing (IT and agencies) (ConnectWise PSA, Autotask (Datto)); Law, accounting, and advisory firms (Clio, MyCase); or wherever else they keep it.
- Income by customer for the period from our accounting software, before sales tax. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop).
- Our list of unpaid invoices, the detailed version with one line per invoice, our payment terms, and the date we downloaded it. Usually in: Accounting and invoices (QuickBooks Online, QuickBooks Desktop).

### From the people who do the work

These need people, not records. Ask the user to put them to their team, or answer them directly.

- Ask how a price gets set, and by whom
- Ask who can agree to extra work in front of a customer, and what happens next
- Ask about jobs where extra work was done and never billed
- Ask who can give a discount, and who finds out afterwards
- Ask which customers usually pay late, and what’s been agreed with them
- Ask which jobs they’d rather the business stopped taking, and why

Before they ask: Going through jobs one by one feels, from the inside, like checking up on people: whoever priced it, the crew that ran over, whoever gave the discount. Suggest they say something like: "I want to see what each job actually earned, from the price we quoted to the money we got paid. I already know some jobs lost money. What I don’t know is whether that’s the price, the way the work runs, or what happens after we invoice. Each of those needs a different fix." Show the list to whoever priced and ran the work before deciding anything, and let them tell you what the numbers missed: a day lost to a site that wasn’t ready, an extra promised to the customer, a payment that never got matched to the right invoice.

### What the records can't tell you

- **Work done and never invoiced.** An invoice nobody sent leaves no record. The only clues are logged hours, a supplier bill, a job note, or someone’s memory.
- **Why a discount was given.** Your accounts show a smaller invoice, never whether it kept a good customer or smoothed over a bad job.
- **Costs never put against the job.** The return trip, the extra phone calls, the apprentice’s time. If they went into general overheads, the job looks better than it was.
- **Jobs too new to judge.** A job that finished last week hasn’t been paid yet. Leave it out, and note how many you left out.

Anything the records suggest is worth more once someone who does the work has confirmed it.

## The method

The method below is written as the business owner's request to you, so "I", "me", and "we" mean the user and their business. Fill any blank still in parentheses or square brackets from what you learned in steps 1 and 2. Then follow it exactly, including how to reply and the shape of the brief.

```text
I want you to help me find out where the margin went on the work I finished in the last three months.
What counts as one row: not settled yet (a job, a matter, an engagement or a project), so ask me first. The decision this needs to inform: none picked, so cover everything this audit checks and lead with the findings worth the most.

The software we keep this in:
(none ticked yet: ask me which software we keep this in before you start)
Use these names when you ask me for anything, and when you read my files, work out which tool each one came from.

Your job is not to tell me my margin is below an industry average or to recommend software. Your job is to account for every dollar between the margin I planned and the margin I actually got, and to say which of four different problems each dollar belongs to.

I will give you exports: the jobs that finished in the period with what I quoted and estimated, the costs recorded against them, the invoices I raised and any credit notes or write-offs, the payments I received, income by customer from my accounting, and my list of unpaid invoices.

Work through these stages, in order, as your working:

1. SET THE FRAME AND INVENTORY THE EVIDENCE
Read back the period, what one row is, and the decision this is for. If I haven't told you one of these, ask. Then list what I gave you: each file, its columns, its row count and date range, and what the stages below need that isn't there. If fewer than ten jobs finished in the period, say so and suggest a longer one. Check whether the hourly cost of labor in my data includes payroll taxes and benefits. If a rate column is named loaded or burdened, or my notes say so, treat it as including them and say so. Only if you cannot tell, ask me exactly this and stop: "Does the hourly cost you gave me include payroll taxes and benefits, or is it only what the person is paid per hour?" Every actual cost depends on the answer. If I only have what people are paid, ask for yearly wages, employer taxes, benefits and paid hours, and work out one average hourly cost. If I never wrote down what a job was expected to cost, ask what margin I aim for on each kind of job, use that, and label the cost gap as measured against my aim.

2. BUILD THE TABLE
One row per finished job, with five columns: quoted price including every change order, the cost estimated when that price was set, actual cost, invoiced, and collected as of the date I pulled the evidence. Work agreed beyond the original scope goes into the quoted price at what it should have been charged and into estimated cost at its hours times the hourly cost, so it adds nothing to the cost gap: both columns
or neither, because a job where extra work went into neither reads as an overrun when the real problem is that nobody invoiced it. Look for extra work in the hours: entries with a note, task or code that isn't in the estimate, or marked as not billable. List them as candidates, price each at the normal rate for that person's role if I gave you rates, and label the row. Where I have told you about extra work and not given you a price for it, say so and label it the same way. Exclude jobs that had not finished, and jobs whose final invoice is younger than the payment terms I gave you (ask me my terms if you don't have them); report how many you excluded, in which category, and what they were worth. Use amounts before sales tax, and count expenses billed on to the customer as both cost and income. Reconcile the invoiced column against income for those customers from my accounting and the collected column against payments received; where payments don't name the invoice they paid, use invoiced minus the balance still open. Report any failure to tie, in dollars, before you interpret anything.

3. SPLIT THE DIFFERENCE INTO FOUR GAPS
Planned margin is quoted minus estimated cost. Actual margin is collected minus actual cost. Account for the difference with exactly four gaps:
  THE COST GAP - actual cost above the cost estimated when the price was set.
  THE BILLING GAP - work delivered and never invoiced.
  THE DISCOUNT GAP - invoiced deliberately below what was earned.
  THE PAYMENT GAP - invoiced and not paid. Split it three ways: not yet due, late against my terms, and more than 90 days past the due date (invoice date plus my terms).
Every dollar belongs to exactly one gap. Where one gap caused another, write the cause next to the dollars rather than adding them in both places. Then show that the four gaps sum to planned margin minus actual margin for the period. If they do not, say what is missing instead of adjusting a figure to make it close.

4. FIND THE PATTERN
For each gap, say whether it is concentrated in a few jobs or runs through nearly every job of one kind. Group by type of work, customer, who priced it, and who delivered it, and report the grouping only where the number of jobs in a group can support it. Compare every job against the spread of my own finished jobs in this period, never against an outside figure.

5. CHALLENGE YOURSELF
For each gap: what evidence contradicts it, what else could explain it, and what I would need to pull to confirm it. Check in particular for costs coded to the wrong job, deposits not matched to the job they belong to, work invoiced in a different period from when it was done, and estimates revised after the job went wrong, which makes a cost gap disappear on paper.

6. DECIDE, PER GAP
For each gap, one of: reprice this kind of work, change what is included at the price, change who can agree to extra work on site and what must happen when they do, change who may discount and against what, change payment terms or take deposits, make a credit decision about a specific customer, or stop selling this kind of work. Smallest to make first. For every decision, what it is worth against the gap it addresses, who owns it, and the result that would reverse it.

HOW TO REPLY
Do the stages above as your working. Do not show me the working. If stages need answers from me, put every question in one numbered message and stop. Otherwise reply with the headline and the brief alone, in the shape below, and lead with the headline every time, even when the news is that nothing needs to change.

Keep the prose under 900 words, and count it before you send it.

THREE THINGS ARE OUTSIDE THAT COUNT AND CANNOT BE SHORTENED,
MERGED AWAY, OR DROPPED TO MAKE ROOM: the headline's four lines, the required lines in section 1, and the rows of the table. Write those first, then spend what is left of the budget on the prose. If you are over, tighten sentences and cut adjectives everywhere; if that is not enough, shorten sections 3, 6, 8 and 9, in that order. Never buy words by dropping a finding, a row, a number, or a required line.

Inside the table, every cell is a phrase rather than a sentence, about a dozen words at most; anything that needs a sentence to be fair belongs in "What to change" or "Still to confirm".

BEFORE YOU SEND IT, CHECK ALL SIX:
- the headline is four labeled lines, under 120 words not counting the labels, and
  names the specific thing rather than a category;
- section 1 carries every required line, with "none" written in where one does not apply;
- every finding you named has a size, or a stated reason it cannot be sized;
- every row you merged says that it is merged;
- the prose is under 900 words;
- every table heading, abbreviation and short form is something I would say out loud; spell out anything else.
If any of these fails, fix it and take the words from the prose.

If I say "show your working", show the table and every rule you applied.

THE HEADLINE
Open with this. Four labeled lines, under 120 words not counting the labels,
before any scope, method, or description of what I sent you. Someone who reads only these four lines has the answer.

- The finding. One or two sentences naming the specific thing that is holding the business back, or where the money is going. Name the thing, never the category: not "a process issue" but the step, the channel, the queue, the person's calendar. If I told you what I think the problem is, say here whether the records agree, in one plain sentence. That contrast belongs here and nowhere else.
- What it costs. The single number that sizes the finding, with its period and unit, and the one line of arithmetic behind it. If the evidence cannot put money on it, size it in the unit it can - hours, days waiting, deals, seats - and say plainly that money isn't available and why. If I asked you to decide something, also put the number in the terms of that decision:
  hours a week freed, customers a budget would buy in each place, what a price change recovers. Compare it only with figures I gave you, such as a new hire's hours, and never assume one.
- Do this first. The smallest fix, what it costs, and who does it. Then, if I asked you to decide something, the answer in one line, labeled a judgment where it is one.
- How sure. "High confidence: from your own records", "Medium:
  your records plus what you told me", or "Low: mostly what you told me" - then the single thing that would most change the answer if it turned out otherwise.

THE BRIEF
Everything below supports the headline for a reader who wants more. Expand it; do not restate it in the same words.
1. What this ran on. Labelled lines, not prose. These are outside the word limit and none of them may be dropped; write "none" where one does not apply.
- Scope and period.
- The decision this is for.
- Scored: what you included, and how many.
- Excluded: how many, why, and what they were worth.
- Evidence pulled: the date the records were pulled. If I didn't say, use the latest date in the files and say so; don't stop to ask.
- Computed with: the tool you used and the rule you applied, or that you had no tool and worked from summary figures.
2. What the evidence covered. Three to five bullets: what I gave you, the period it covers, and what was missing that would have changed the answer.
3. What we found. One paragraph, under 150 words, in plain words: how the work actually runs and what the evidence shows, in enough detail to make the headline's finding stand up.
4. The numbers. Planned margin and actual margin for the period, each as dollars and as a percentage of quoted, and the four gaps in dollars, kept separate and shown to sum to the difference between them. For each: how it was computed and which rows it rests on.
5. The table. One table, one row per job (jobs with a total gap under $500 may be merged into one row that says so), ordered by the size of its total gap: the job, quoted price, estimated cost, actual cost, invoiced, collected, and which gaps it contributes to. Column headings and cells are in plain words too: the labels above describe what goes in each column, so write them the way I would say them.
A finding too small or too uncertain to act on still gets its own row, marked as such. Where the table is long, you may merge rows only when they are the same kind, and the merged row has to say so. Never drop one.
6. What to change. Every fix, smallest first, the headline's included, with one line each on why the more expensive ones aren't needed yet.
7. How you'd know it worked. For each of the three biggest fixes, in this order and
always these four: the measure, named the same way you would name it again next time; where it stands today, with the date the evidence was pulled; the target; and the result that would mean the diagnosis was wrong. Write them so someone re-running this in a month can line their numbers up against yours without having to interpret anything.
8. Still to confirm. Up to four things the evidence couldn't settle, each with exactly what would settle it.
9. Not claiming. Up to five claims you are refusing to make, one line each.
Then one last line: offer to do the first fix for me now, naming exactly what you would write (the list, the email, the message, the new rule), then add any closing lines your instructions give, and stop.

RULES THROUGHOUT
- Write everything you show me for a business owner who is not technical: plain, everyday words, short sentences. No jargon, acronyms, or names of methods or frameworks. If a term cannot be avoided, explain it in the same sentence. The stage names above are for your working, not for my brief.
- Names that come from my software or ad accounts count as jargon too: a campaign type, a setting, a report name, a platform's word for a result. The first time one appears, say in plain words what it is, then use the plain version. Write category names in normal case, never in capitals. When something costs less, give the two amounts; never sum it up in one word that sounds like a bargain or a corner cut.
- Do not use dashes to break into a sentence. Use a comma, or start a new sentence.
- If you have a code or data-analysis tool, use it for every count, median, and total, including simple ones like how many people are on a list, and state the rule you applied. If you don't, say so and ask me for summary tables instead of raw exports.
- Before writing any comparison such as "X times", "half", or a percentage, recompute it from the two figures and show both figures beside it.
- Before saying something holds on every row, check every row. If it holds on only some, say which, and do not build on it for the rest.
- Never invent a number that is not in my evidence. Where a value is blank, say so and leave it out of the totals rather than estimating it.
- Numbers I gave you in the walkthrough count as evidence: use them where the records have nothing, and label each one as my estimate.
- Never quote an industry benchmark, average, or percentage from outside my data, even if I mention one. If I ask how I compare to one, tell me plainly that you are not going to and why, rather than leaving the question unanswered.
- My exports often cut the same figures more than one way. Before adding two numbers together, check they come from the same cut; if they do not, report them side by side and say which values may be counted in both rather than summing them.
- Where I have told you something happens off-system or goes unlogged, silence in the records there is not evidence of anything. Say what would need to be logged.
- Where you nest one finding under another as its cause, the parent carries the rolled-up total of everything beneath it, labeled as a roll-up so no figure is counted twice. Every item you size belongs to exactly one total, and no finding you name is left without a size unless you say what is missing to size it.
- Keep observed facts, inferences, and unknowns separate. It is fine, and often correct, to conclude the evidence cannot settle something.
- A target is a suggested goal for me to change, labeled as one, never a finding.
- If records are dated outside the period or after the date they were pulled, say how many, report them separately, and never drop them silently.
- Where my files and what I told you disagree, keep both versions and put the disagreement under "Still to confirm".
- If I push you for a verdict on the decision afterwards, give your best read, label it a judgment rather than a finding, and say what would change it.
- Never give me a single "margin leakage" figure without the four gaps beside it and the jobs behind each one.
- Never compare my margin, my overruns, or my collection times to an industry average or a published range, even if I ask. Compare a job to the spread of my own finished jobs.
- A job is not scored until it is finished and its final invoice has had time to be paid. Say which jobs you excluded and why rather than including them at a partial figure.
- Annualizing anything you find is a separate claim. Report the period I gave you, and if I ask for a year, say what it assumes about the mix repeating.
- Where the only evidence for work delivered and never invoiced is what I told you in the walkthrough, keep it, label it as my estimate, and say what would need to be recorded for the next run to measure it.
- Do not treat a discount as an overrun or an overrun as a discount, even where one caused the other.
- Do not recommend software as a fix for a gap until you have said what the rule or price change would be without it.
- If I asked whether to raise prices, say in "Not claiming" what these records cannot show: how a price rise would change the number of quotes we win.
```

## After the brief

At the very end of the reply that holds the brief, after the offer to do the first fix, add these two short lines and nothing more:

1. Before acting on it, have a different AI assistant, or a new chat, recompute every number in the headline and the table from the same files.
2. If they'd like Catalyst Studio's team to run this audit on their data with them, it starts at $500, depending on the size of the investigation and the company's stage: https://catalyststudio.ai/?audit=margin&run=catalyst#get-started

Say each once. Don't repeat them later in the conversation.

Tick your software above and it’s added to the skill.

Keep it for next timeDownload skill

The download is the blank version, without your answers. Install it once and next month just ask your assistant to run the audit again. Upload it where your assistant keeps skills: ChatGPT, Gemini, Grok, Claude, and most others take it as it is.

Agent that works from a folder? Save the skill file in a folder called catalyst-margin-leak-audit.

  1. Tell us what’s going on. We reply within one business day.
  2. We collect the data with you, on a short call or with temporary access.
  3. You get the brief within three business days of having the data.
Have us run it

How the audit works

What the AI checks, and what you get back.

01

What it looks for

Your accounts already tell you whether the business is profitable. This audit finds which work is profitable, and where the money you priced into a job went. Everything between the margin you planned and the margin you got is one of four gaps.

1
The cost gap. The job cost more than expected. The money was never in the price.
2
The billing gap. Work was done and never invoiced.
3
The discount gap. Someone chose to invoice less than was earned.
4
The payment gap. Invoiced, and the money hasn’t arrived.
02

What the AI does with your files

1
Scores every finished job. One row per job, from quoted price to money received, checked against your accounts first.
2
Splits the difference into the four gaps. Extra work agreed on site goes into both the price and the expected cost, so a favor done on site doesn’t look like a job that ran over.
3
Proves the four add up. They must equal planned margin minus actual margin exactly. If they don’t, something is missing or counted twice.
4
Looks at where each gap sits. A gap in three jobs means three stories to look into. A gap in nearly every job of one kind is about how that work is priced or run.
5
Matches a decision to each gap. A price, a rule about who can agree extra work on site, a rule about discounts, or payment terms.
03

What you get back

In it

  • One row per finished job, from quoted price to money received
  • Planned and actual margin for the period
  • The four gaps, each sized and tied to named jobs
  • The sum showing the four add up exactly
  • A decision for each gap, with what would change it

Left out

  • A single “you’re losing X% of revenue” figure
  • Any comparison with an industry average
  • A yearly figure that assumes every quarter looks like this one
  • Software recommended before the rule or price change it would replace

Example brief

An 11-person electrical contractor. 41 jobs finished April to June.

The questionRaise service-call prices, or change how extras get agreed on site?

The answerChange how extras are agreed first. $27,600 of the loss costs nothing to fix.

  • Margin30.0% of the quoted price planned, 12.7% received.$105,800 short
  • Cost gapSecond trips on fourteen service calls, plus the one commercial job everyone blamed ($9,400).$33,200
  • Billing gapExtra work agreed on site and never invoiced, across fourteen jobs.$27,600
  • Discount gapFour discounted invoices, three of them for the same customer.$10,700
  • Payment gapOne office fit-out is $28,400 of it, 74 days after its final invoice.$34,300

The four gaps add up to $105,800 exactly. Service-call prices stay put for a quarter while the second trips are fixed.

When to run it again

Monthly. A job has to be finished, and its last invoice given time to be paid, before it can be scored. A month gives you enough finished jobs to see a pattern; a week gives you three jobs and noise.
The second run is quicker than the first. You’re only checking what changed since last time. How to put it on a schedule →