Free guide. MurraySwift & Associates
The Owner’s Operating Playbook
Six stages from diagnosis to execution, the questions behind each one, and a 20-question scorecard to tell you where to start.
Most owners know when something is off long before they know what it is. Revenue is up and profit is not keeping pace. Cash feels tight in a good month. The team moves when you are in the room and stalls when you are not. The trouble is not a lack of effort or ideas. It is that the next step is not obvious, every problem looks connected to every other problem, and picking the wrong one to work on first is expensive. So the owner does what feels responsible: works harder on all of it at once. This playbook is the alternative. It lays out the order we work in, the questions we ask at each step, and what each step should leave on paper, so you can see where your own business stands and where to begin.
How to use this playbook
The six stages run in order: Diagnose, Strategize, Commercialize, Model, Decide, Execute. The order is the argument. You do not set strategy before you know what is broken, you do not price a plan that does not exist, and you do not commit money to a decision you have not modeled.
Short on time? Take the scorecard at the end first. It will tell you which stage is weakest, and that is the stage to read.
Each stage has the same parts: the question it answers, the questions we ask, what it produces, the mistake we see most often, where AI helps, and one thing to do this week. Answer the questions in writing. A rough number on paper beats a precise one carried in your head.
Stage 1
Diagnose
Where is the money actually going, and which problems are costing you the most?
The questions we ask
- 01
What were your revenue, gross margin and net margin over the last twelve months, and which way has each moved over the last two years?
- 02
Walk us through what it costs to deliver one unit of your main product or service, including the costs nobody likes to count: rework, returns, waste, rush freight, write-offs.
- 03
Is there a line on your P&L you wince at every month? When was the last time anyone went through the recurring expenses line by line?
- 04
When did you last raise prices, and what happened? Where do you discount, and who decides?
- 05
Where do new customers come from, and if your top channel stopped tomorrow, what would revenue look like in ninety days?
- 06
What share of revenue comes from customers buying again, and when did you last win back a customer who had gone quiet?
What it produces
A written diagnostic. It scores the business in eight areas, from positioning and pricing through direct cost, overhead, the offer, the sales pipeline, repeat business and referral partnerships, against your real financials rather than opinion. Each weak area carries an estimate of what it is worth to you, with the math shown, and the whole thing ends in a ranked order of what to fix first. It also produces a baseline: the numbers as they stand today, so that every later stage has something to be measured against.
The common mistake
Treating the loudest problem as the biggest one. The issue that interrupted your week is rarely the one costing you the most.
The issue that interrupted your week is rarely the one costing you the most.
Where AI helps here
- Drafting a first variance read from your monthly P&L: what moved, by how much, and which lines moved together. You check the reasons; the tool only finds the movement.
- Sorting twelve months of transactions or vendor bills into categories, so recurring spend you forgot about shows up as a list.
- Summarizing your vendor and software contracts into a single table of what each one costs, when it renews, and what it promised.
Further, built with us
- A month-end anomaly monitor: every account in the books compared with its own history, and the outliers explained in a one-page note before the month closes.
- Every vendor contract and recurring bill read into one register with renewal dates, auto-renew clauses and price escalators, so nothing renews without a decision.
- Accurate reports shared from the accounting system, the CRM and the point of sale, so the diagnosis runs on all of your data, not a sample someone exported.
The cautionA diagnosis is only as good as the books underneath it. If your categories are inconsistent or the bookkeeping runs months behind, AI will produce a confident summary of bad data. Fix the inputs or label the gaps before you trust the read.
Do this week
Pull last year's P&L and your current one side by side. Circle the three lines that grew faster than revenue. Next to each, write one sentence on why. If you cannot write the sentence, that line is your first question.
Stage 2
Strategize
Where will this business play over the next one to three years, who is it really for, and how does it win?
The questions we ask
- 01
Ten years out, if this business is exactly what you want it to be, what does it look like? And what are you personally trying to get out of it: wealth, freedom, a handoff, an exit?
- 02
Describe the customer who is easiest to serve, pays without friction and sends you others. What is happening in their world right before they come looking for you?
- 03
In one sentence: who is this business for, what does it do for them, and why does it win over the strongest alternative?
- 04
What do you do better than anyone in your market, and if a well-funded competitor decided to copy it tomorrow, what would it cost them and what would they still not have?
- 05
Of the weak areas the diagnosis found, what is the one move over the next twelve months that closes each? State each as an outcome, not an activity.
- 06
What are the hard limits on the next twelve months: capital, cash runway, team capacity, regulation, your own bandwidth?
What it produces
A strategy brief. It holds a positioning statement you can defend, a specific profile of your best customer, the shape of your offer from first purchase to most valuable, what makes you hard to copy, and a three-level horizon: a short direction for ten years, the big moves for the next three, and no more than three priorities for the next twelve months, one marked as the most important. Each priority ties back to a problem the diagnosis found and carries a rough value.
The common mistake
A strategy built on adjectives. "Great service" and "growing companies" are not a position or a customer; every competitor says the same thing. The second is choosing seven priorities. Three is already a lot for a company that is also running its day.
If their answer does not match your sentence, theirs is closer to the truth.
Where AI helps here
- Reading your last few hundred customer reviews, support emails or sales call notes and pulling out the words customers actually use to describe their problem. Use their words in your positioning, not yours.
- Building a first pass at a competitor table from public websites and pricing pages: what each one claims, what they charge where it is published, and who they say they serve.
Further, built with us
- Win and loss mining: sales call recordings and lost-deal notes analyzed for the reasons customers actually choose you or leave, by segment.
- A monthly competitor watch over public pricing pages, job postings and reviews that flags what changed and what it might mean for you.
- Each three-year move stress-tested against several market conditions, written up as a short memo per scenario for the leadership team.
The cautionAI is good at gathering and organizing what is public and what you give it. It is not good at knowing what your customers will pay for or what your team can actually deliver. Every claim in the competitor table needs a source and a date before it shapes a decision.
Do this week
Write your positioning in one sentence: who it is for, what it does for them, why it wins. Then call two of your best customers and ask why they chose you. If their answer does not match your sentence, theirs is closer to the truth.
Stage 3
Commercialize
Does the way you price, package and sell actually carry the strategy, and does each sale make the money it should?
The questions we ask
- 01
What is the easiest, lowest-risk thing a new customer can buy first, and what is the path from there to your most valuable offer?
- 02
What is your highest-margin offer, and is it the one you lead with?
- 03
What does it cost you to win a customer, and how long until that customer has paid it back?
- 04
Where does recurring revenue live in your business, if anywhere?
- 05
Do you have a premium option for the customers who would pay more for more, or one price for everyone?
- 06
What makes saying yes feel safe to a buyer: a trial, a pilot, a guarantee, proof?
What it produces
The commercial model. It sets out your unit economics (average order value, margin per unit, cost to acquire a customer, what a customer is worth over the relationship, and how long payback takes), the pricing structure with the gaps named, the offer ladder with an entry point and a premium tier, and the one number that tells you the business is winning, with the few inputs that drive it.
The common mistake
Leading with the product you like best instead of the one that makes the most money. A close second is discounting for volume without checking whether the discounted sale makes anything once delivery cost is counted.
Leading with the product you like best instead of the one that makes the most money.
Where AI helps here
- Pulling order history into a list of customers who bought once and never came back, and customers whose order frequency has dropped. That list is the start of a win-back plan.
- Grouping your order history by product to show margin per product and per customer segment, so you can see which offers carry the business and which ride on it.
Further, built with us
- Price testing from your own history: how volume moved when prices or discounts changed, by product and customer, before you touch a list price.
- Reorder and churn signals: every account scored weekly on whether its buying pattern is slipping, with a call list for the sales team.
- Quoting assistants that work from your price book and margin floor, so every quote is consistent and nothing goes out below the floor.
The cautionOrder history lives in several places in most companies: the accounting system, the CRM, the point of sale, a spreadsheet someone keeps. Before you trust any AI analysis of customers, confirm it is reading all of them and that the customer names match across systems. Duplicates and mismatched names are where these lists go wrong.
Do this week
List every product or service you sell. Next to each, write its price and your honest best estimate of its margin. Mark the one you lead with in sales conversations. If it is not near the top on margin, ask why.
Stage 4
Model
What do the numbers say the plan will do to profit and cash, and what has to be true for it to work?
The questions we ask
- 01
What does the plan assume about volume, price, cost to deliver and headcount, and which of those assumptions are you least sure of?
- 02
If the plan works, what happens to cash month by month over the next four quarters? When is the low point, and is there enough to get through it?
- 03
What does the picture look like if the weakest assumption comes in at half of what you hoped?
- 04
How long does a dollar sit in inventory or receivables before it comes back to you?
- 05
What is the one number in the model that, if it moves, moves everything else?
What it produces
A financial model built from the bottom up: from units, prices, costs and timing to revenue, margin and cash, with every assumption written down next to where it came from. It carries a base case and at least one case where the riskiest assumption goes wrong, and it shows which inputs the result is most sensitive to. Where a figure is unknown, it gets a reasoned estimate, flagged as an estimate, with the logic shown, rather than a blank or a guess presented as fact.
The common mistake
Building the model from the top down. "We will grow revenue by a fixed amount next year" is a target, not a model. The second mistake is modeling profit and ignoring cash. Plenty of profitable plans run out of money in the middle.
Plenty of profitable plans run out of money in the middle.
Where AI helps here
- Turning your plan's assumptions into a first-draft spreadsheet structure, with inputs separated from calculations so you can change one number and see the result.
- Checking a model for the errors people miss: a formula that stops a row early, a sum that double counts, an input typed as a fixed number where it should be a reference.
Further, built with us
- A driver-based model connected to the books and refreshed from actuals each month, instead of a spreadsheet rebuilt every quarter.
- A rolling cash forecast built from bank, receivables and payables data, with the low point flagged weeks before it arrives.
- Hundreds of scenarios run across the riskiest assumptions at once, so you see the range of outcomes rather than one number.
The cautionDo not let AI fill in the numbers you do not have. A tool asked for a figure will give you one, and it will look exactly as confident as a real one. Every input in the model should trace to your books, a quote, a contract, or an estimate you can explain out loud.
Do this week
Take the one priority you marked as most important in Stage 2 and write down the three numbers it depends on. Next to each, write where the number came from. Any number with no source is your first modeling task.
Stage 5
Decide
When the business faces a fork that costs real money, how do you choose, and how do you know later whether you chose well?
The questions we ask
- 01
What does success look like, in concrete terms, at a specific date? Write it before you look at the options, not after.
- 02
What are at least three realistic options, including doing nothing or waiting, and a combination of the others?
- 03
For each option, what happens next, and what happens after that? Which options can you reverse cheaply, and which lock you in?
- 04
How does each option score against the outcome you wrote down, against your team's capacity, and against the time you have?
- 05
Imagine it is a year from now and the decision failed. What killed it?
- 06
What is decided, who owns the next step, what happens in the next seven days, and when do we look at this again?
What it produces
A decision memo. One page per decision that holds the outcome you wanted, the options you considered, the consequences of each, how they scored, the risks of the chosen path with a mitigation and an owner for each, and the date the decision gets reviewed. In our work these memos are kept in one place alongside the plan and the numbers, so every major decision can be checked against what actually happened.
The common mistake
Framing the decision as A or B. Real choices almost always have more paths than two, and the option you did not write down is often the better one. The other mistake is writing the success criteria after you have already picked a favorite, so that only your favorite can pass.
The option you did not write down is often the better one.
Where AI helps here
- Generating options you had not considered. Give it the outcome and your constraints and ask for five different ways to get there; most will be weak, but one may be worth a look.
- Laying out side-by-side comparisons of vendor proposals or contract terms, with differences in price, term, exit clauses and service commitments pulled into one table.
Further, built with us
- Decision memos drafted from the model and the data, with a second AI pass assigned to argue against the recommendation before anyone signs.
- A searchable decision log: every past memo, its assumptions and what actually happened, ready when a similar fork comes up again.
- Pre-mortems built from your own history of what went wrong, so the risk list is specific to your company, not generic.
The cautionThe decision is yours, and so is the reasoning. AI can widen the option list and organize the comparison. It cannot weigh what matters to you, your family or your team, and it will cheerfully argue for whichever option you seem to prefer. Write your outcome first and ask it to attack your favorite, not support it.
Do this week
Pick the decision you have been circling for longest. On one page, write the outcome you want and the date you will judge it by. Then write three options, one of which is "do nothing for ninety days."
Stage 6
Execute
Who does what, by when, with what resources, and how do you know the plan is working before it is too late to change it?
The questions we ask
- 01
For each of this year's priorities, who owns it by name, and does that person have the time, the authority and the budget to deliver it?
- 02
What are the results you expect, and by what dates? Are they written down where the team can see them?
- 03
What will you stop doing to make room for this? Which projects, meetings or reports go away?
- 04
What systems, vendors and people does the plan depend on, and who is accountable for each of them?
- 05
At thirty, sixty and ninety days: were we right, were we wrong, and why?
- 06
If you are building from scratch: what is the smallest version that proves customers will pay, and what can wait until they do?
What it produces
An operating plan on paper. It sets out the priorities for the quarter with an owner and a due date for each, a short set of weekly numbers that show whether the plan is on track, a resource plan covering people, cash, systems and vendors, and a review rhythm: the same day each month, the numbers, the plan, the exceptions and the next decision. Every fix is measured against the number it was supposed to move, and what did not move gets reworked, not explained.
For a company going through a larger change, such as a system replacement, a leadership turnover or preparation for a sale, the plan adds the sequence of the change: what moves first, what depends on what, and who carries it when the owner is not in the room. For a startup, the same plan is lighter and more focused: own the customer before you build everything behind it, and add capacity as demand proves itself.
The common mistake
A plan with no owner. The plan is right, and nobody is carrying it, so it stalls the first week the owner is pulled elsewhere. The second mistake is adding priorities without removing anything, so the team is asked to do more with the same hours and quietly does less.
The plan is right, and nobody is carrying it.
Where AI helps here
- Turning meeting notes into a dated action list with an owner on each line, ready for the owner to confirm before it goes out.
- Drafting the monthly review pack from the weekly numbers: what is on track, what slipped, and what needs a decision. Someone still has to read the numbers and add the why.
- Mapping where your team re-keys the same information between systems by hand. Those handoffs are usually the first and safest place to automate, and the place hours and errors live.
Further, built with us
- Tools built for one job, like the daily production plan across two lines and packaging we built with a fruit processor in a four-week sprint.
- A company knowledge assistant over your procedures, contracts and past decisions, with permissions, so people stop waiting on the one person who knows.
- Workflow agents that do the re-keying between systems and stop for a named person's approval before anything is sent, paid or changed.
The cautionPut one named person in charge of every AI tool you deploy, and measure it against the hours or errors it was supposed to remove. A tool nobody owns drifts, and a tool nobody measures never gets switched off.
Do this week
Write down your three priorities for this quarter. Put one name next to each. Then write one thing the company will stop doing to make room. Book a thirty-minute review in your calendar thirty days from today and keep it.
Twenty questions
The 20-question self-scorecard
Rate each statement from 1 to 5.
1 = not true at all. 3 = partly true. 5 = completely true, and you could show us the proof.
Be hard on yourself. A generous score only hides the problem.
Diagnose
I know our gross and net margin for the last twelve months without looking them up.
I can name the three things costing this business the most money, in order.
Someone has reviewed every recurring expense and vendor contract in the last twelve months.
Our books close within a few weeks of month end and I trust what they say.
Strategize
I can say in one sentence who we serve, what we do for them, and why we win, and my team would say the same sentence.
We have no more than three priorities for the next twelve months, and each is written as an outcome.
I know what makes us hard to copy, and our prices reflect it.
Commercialize
I know which of our products or services makes the most margin, and it is the one we lead with.
I know what it costs us to win a customer and how long it takes to earn it back.
We have a clear path for customers to buy more from us over time, and someone works it on purpose.
Model
We have a forward view of cash, month by month, for at least the next twelve months.
Every major assumption in our plan is written down with where it came from.
I know what happens to cash if our riskiest assumption comes in well below plan.
I know the one number that, when it moves, moves the rest of the business.
Decide
When we face a major decision, we write down what success looks like before we compare options.
Our big decisions are written down with an owner, a next step and a date to review them.
In the last year we have gone back and checked at least one big decision against what actually happened.
Execute
Every current priority has one named owner with the time and authority to deliver it.
We review the plan on a fixed schedule, and it gets corrected in writing when it is wrong.
The business keeps moving on its priorities when I am out for two weeks.
How the scoring works
Add up your score for each stage and divide by the number of statements in that stage to get an average between 1 and 5.
- If Diagnose averages below 3, start there, whatever the other stages say. Every later stage stands on the diagnosis. If it is wrong, everything built on it compounds the error.
- Otherwise, start with your lowest-scoring stage. If two are tied, start with the one that comes first in the sequence.
- Any stage averaging below 2 is likely holding back the stages after it. Fix it before investing heavily further down the list.
- If every stage averages 4 or above, your work is in Execute: keeping the review rhythm running and making sure the business does not depend on you to hold it together.
Date your averages and score yourself again in ninety days. The movement between the two is the most honest measure of progress you will get.
Where to go from here
Everything in this playbook can be done by an owner with a pen, a P&L and an honest afternoon. If you would rather have help, here is how working with us starts.
We begin with a diagnostic, not a pitch. It takes about a month and costs $5,000. You get a written read of how the business actually operates, where margin is leaking, what each problem is worth in dollars, and the order to fix them in. You leave knowing what to do first, what it will take, and whether you need us for any of it. If you go further with us, the diagnostic becomes the plan on paper and nothing is re-diagnosed.
Before any of that, the first call is free: thirty minutes to see whether we're a fit and what we'd look at first. Bring the three things that keep you up at night. Nothing to sign.
MurraySwift & Associates is two operators who have sat in your chair, working with owner-operated companies and well-capitalized startups from Traverse City, Michigan.
Start with a free 30-minute call.
Bring the three things that keep you up. In thirty minutes you’ll know whether we’re the right fit and what we’d look at first. Nothing to sign.