Business health check
Most companies past their first few million are not badly run. They are running on an arrangement that fit the business two years ago and has never been redesigned since. This finds where that arrangement has started to cost you, and in what order to fix it.
In short
A business health check is a structured read on whether the way a company is organized will carry the next two years. This one asks eighteen questions across six operating areas, then returns the one weakness holding the others in place and the order to fix things in. About five minutes, and the result is yours whether or not you ever speak to us.
What we take on
A read, not a score
A number on its own is a vanity metric. This returns which of six areas is the binding constraint, why the others are probably symptoms of it, and the sequence that costs least.
Written for a real operating company
The questions assume revenue, staff, systems that disagree with each other and a leadership team that already knows something is wrong. Nothing here asks whether you have a business plan.
Five minutes, then it is yours
Eighteen questions, three options each. The full result copies out as plain text, ready to paste into a leadership doc without editing.
An outside read if you want one
Send it in and a person reads it and writes back within a business day. If the answer is that you do not need an outside firm this quarter, the note says that.
Why most health checks are worth nothing
The usual version asks twenty questions, adds them up, and tells you that you scored 64 percent. That is a number with no decision attached to it. Nobody has ever changed how a company runs because of a percentage, and the reason is simple: a total hides the only thing that matters, which is which single weakness is holding the others in place.
Operating weaknesses are not independent. A company that cannot produce its numbers usually cannot produce them because one person owns all of them, which is a dependency problem wearing a reporting costume. A company whose work sits in queues usually has queues because four people have to approve things, which is the same dependency problem again. Fix the reporting and the queues come back. Fix the dependency and both improve without being touched.
That is why this returns a binding constraint rather than a ranked list of everything wrong. The constraint is the one to spend money on. The rest is a watch list, and treating a watch list as a work list is how companies end up with three concurrent improvement projects and no improvement.
The second thing most checks get wrong is the horizon. A finding with no sequence attached becomes a slide. So the output here ends with three moves in order: what to measure in the first two weeks, what to change by week six, and what to touch only after that. The order is not decoration. Measuring before changing is what stops you automating a process that should have been deleted.
None of the eighteen questions ask what you plan to do. They ask what happens on an ordinary Tuesday. The gap between those two is where the honest version of a business lives, and it is the gap that an acquirer, an auditor or a bad month will find whether or not you looked first.
This is deliberately not a sales instrument. Roughly half the results that come back through it describe a company that should fix one specific thing internally and not hire anybody. Saying so is cheaper than the alternative, which is taking on work that was never going to succeed and being remembered for it.
Eighteen questions, about five minutes
Six areas that decide whether a company scales on its systems or on its people. Answer honestly rather than generously; the output is only worth what the inputs are. Nothing is stored, and the result is yours whether or not you ever speak to us.
How to use it, about five minutes
- Answer eighteen questionsSix areas, three questions each. Answer for how the company works on an ordinary Tuesday rather than how it is supposed to work.
- Read the constraint, not the totalThe result names the single area holding the others in place. A strong total with one weak area is still a weak company, and the read says so rather than averaging it away.
- Work the ninety day sequenceMeasure in the first two weeks, change the constraint by week six, and only then touch the second weakest area.
1. Flow and handoffs
Three questions. Answer for how it works on an ordinary Tuesday, not for how it is meant to work.
Take your most common piece of recurring work. Do you know how long it takes from request to finished?
When that work is in progress, is it mostly being worked on or mostly waiting?
How many approvals does a routine item pass before it is done?
2. Information and reporting
Three questions. Answer for how it works on an ordinary Tuesday, not for how it is meant to work.
If your board or your bank asked for last month's operating numbers this afternoon, what happens?
When two systems disagree about the same figure, is there an agreed answer for which one wins?
How many people can produce the operating numbers without asking anyone?
3. Systems and data
Three questions. Answer for how it works on an ordinary Tuesday, not for how it is meant to work.
How many systems does one customer record pass through before the job is finished?
Is anything important still running through a spreadsheet that nobody owns?
When you add a system, does a named person own the connection to the others?
4. Dependency and key person risk
Three questions. Answer for how it works on an ordinary Tuesday, not for how it is meant to work.
If your most loaded operator took three weeks off with no notice, what happens?
How much of the week's work needs a decision from you or one other person?
Are your core processes written down well enough for a new hire to run them?
5. Commercial operations
Three questions. Answer for how it works on an ordinary Tuesday, not for how it is meant to work.
From an agreed deal to the first invoice going out, how long?
Do you know your close rate and what one inquiry is worth?
When a customer asks for status mid delivery, how fast can anyone answer?
6. Continuity and readiness
Three questions. Answer for how it works on an ordinary Tuesday, not for how it is meant to work.
If a buyer, an investor or an auditor asked for your operating documentation tomorrow, could you produce it?
Has anything important failed quietly in the last year and been found late?
If volume doubled in six months, do you know what breaks first?
Operating health
0 / 36
Nothing answered yet.
Flow and handoffs
Information and reporting
Systems and data
Dependency and key person risk
Commercial operations
Continuity and readiness
The read
Answer the eighteen
The written read back
Want this read by somebody who has fixed it before?
Send the result and you get back a written note: which of the six is actually the constraint rather than the loudest symptom, what a company shaped like yours usually finds underneath it, the sequence to fix it in, and what the first two weeks would need from your team. One business day, from a person, not a template.
If the honest answer is that you do not need an outside firm this quarter, that is what the note says. It happens often enough that it is worth stating up front.
Goes to one inbox, read by Shaheer Shaikh.
This is the broadest of the set. If it points at one area and you want to go deeper on that area specifically, the tools page covers the other eleven, grouped by the question each one answers.
What this usually leads to
The health check names the area. Pricing it is the next move, and the task cost calculator turns a named problem into a number for the year. If you would rather have the same method run properly, with the interviews and the observation included, that is the operations assessment.
Questions we get
How long does it actually take?
About five minutes if you answer from memory, which is the right way to do it. If you find yourself wanting to go and check something before answering, that hesitation is itself the finding, and the honest answer is the one you were about to check.
It does not save partway. It is short enough that it does not need to.
Is this useful if we already know what our problem is?
Usually, yes, and not for the reason people expect. Most leadership teams can name a problem confidently and are naming a symptom. The value here is the ordering: whether the thing you have named is the constraint or downstream of something cheaper to fix.
If it confirms what you already thought, you have lost five minutes and gained a written case for the budget conversation.
What size of company is this written for?
Companies roughly between one and fifty million in revenue, with staff, more than a couple of systems, and no dedicated operations function. Below that the answers tend to be obvious. Above it the questions need to be asked per business unit rather than per company.
Most of the work behind it has been US based, and the questions assume a company operating on US business hours and reporting conventions.
What happens to the answers?
Nothing, unless you press send. The whole thing runs in your browser and there is no account, no tracking of individual answers, and nothing written to a server while you use it.
If you do send it, it goes to one inbox and is read by Shaheer Shaikh. It is not added to a marketing list and there is no sequence of follow up emails behind it.
What does the written read back actually contain?
Which of the six is the real constraint rather than the loudest symptom, what a company shaped like yours typically finds underneath it once somebody looks, the order to fix things in, and what the first two weeks would need from your team in hours and from whom.
It is a page or so of specifics about your answers. It is not a capability deck.
How is this different from the value leak diagnostic on this site?
The value leak diagnostic asks nine questions about revenue, capacity and control, and points at where money is escaping. This is broader and structural: it covers flow, information, systems, dependency, commercial operations and continuity, and it returns a sequence rather than a single leak.
If you want a fast read on where money is going, use that one. If you want to know whether the way the company is organized will carry the next two years, use this one.
Who reads it, and what is the relevant experience?
Shaheer Shaikh, who runs operations for LARVOL, a San Francisco AI company working with global pharmaceutical companies on clinical trial data and AI benchmarking, and was named one of ten top performers company wide.
He cofounded a Delaware public benefit corporation and ran it end to end, holds a business degree, a Six Sigma Green Belt and Anthropic certification on the Model Context Protocol. The method used here is published on this site in full rather than described.
We scored well. Is there anything to do?
Take the lowest of the six and keep it from becoming the constraint. Companies in that range rarely fail; they drift. Approval layers accumulate, documentation ages quietly against a process that moved, and one person gradually becomes the only one who can produce a number.
The useful discipline is to retake this every couple of quarters and watch which bar moves down, because the direction matters more than the total.
More in the guides and every answer in one place.
Shaheer Shaikh, operations lead at LARVOL, a San Francisco AI company working with global pharma on clinical trial data and model benchmarking. Six Sigma on the process side, Anthropic certified on the Model Context Protocol, ten years across eight industries. More about the firm