How to diagnose what is actually wrong in your business: current state, desired state, gap 

A business diagnosis works out what exists today from evidence, what should exist in terms you can test and what is really stopping the move between the two. Then it lists the constraints and the options (doing nothing included) and ends with a recommendation and a plan you can act on.

The point is to find the real cause of a problem before you spend a single rupee fixing it. In this piece I'll walk you through the eight steps so that you or a senior manager you trust can run them without outside help.

Why is the problem you describe first rarely the real problem?

Owners describe what hurts and not what causes the hurt. Cash is tight, the team is overloaded, the reports are late and the customers are unhappy. All of it is real but each one sits at the end of a chain and the cause is usually somewhere earlier in it.

Take cash that's always tight, where the first guess is nearly always that sales are too low. But the cause could be invoices that go out late, a customer who pays in ninety days while nobody chases, a discount given on the phone and never recorded or a buying habit that ties up money in stock. Each of those has a different fix and if you fix the wrong one then you've spent the money and your cash is still tight.

There's a second reason and it's a sneaky one. The problem usually arrives with a solution already attached, like "we need new software", "we need a manager" or "we need a CRM". In my opinion a solution is just a guess about the cause and a diagnosis tests your guess before it turns into an expensive project.

It is also the cheapest piece of work in any change. It takes days or weeks rather than months, it costs far less than a wrong project and quite often it shows that the right project is smaller than the one you had in mind.

What are the eight steps of a business diagnosis?

The sequence is the same for a ten-person firm and a large group and nothing in it is proprietary. The value sits in the judgement you apply at each step. Run the steps in order because skipping ahead to options is the most common mistake and it walks you straight back to the solution you started with.

Step 1: Current state

Write down what exists today from evidence and not from the organisation chart. Who really approves a discount in your business, you or the sales head? Which spreadsheet does the month-end report actually come from? How long does an order take from enquiry to delivery when you measure ten real orders instead of guessing?

Current state means the process as it's actually run, workarounds and all. I'd pay the most attention to the workarounds because they show you exactly where the official process gave up. Use plain words and simple pictures - a one-page flow of how an order moves through your business is plenty.

Step 2: Desired state

Describe what should exist concretely enough that you can test it. "Faster reporting" is not a desired state, but "management figures reach the owner by the seventh day of the month and the owner can see profit by site" is one and you can check it on a calendar.

Keep it in the terms your business already uses, like days, rupees, orders and hours. If you can't say how you'd know you had arrived then your target isn't ready yet.

Step 3: Gap

The gap is the distance between steps one and two in the same units. Reports arrive on day twenty-one and should arrive on day seven. Jobs get re-quoted after the site visit in most cases when they should be quoted once. If you can't put a number or a plain description on the gap then it's still a feeling and I've yet to see anyone fix a feeling.

One warning though - some gaps aren't worth closing at all. Size each one by what it costs you in money, time or risk because that sizing is what you'll use to rank them later.

Step 4: Root causes

Ask why the gap exists and then ask why again about the answer you get. Reports are late because the bank entries are posted late. The entries are posted late because the statements come as paper from two branches and they come as paper because nobody owns that task. So the root cause is the missing owner and not the accounting.

My rule of thumb is that about five rounds of why gets you to something you can act on. Stop when the answer is a decision someone can make, a process someone can change or a fact someone can fix. Be wary of a cause that's only a person to blame, because blame ends the questioning too early and the system that allowed the mistake is nearly always the better answer.

Step 5: Constraints

List what cannot or should not change. I'd put cash first, then people, existing contracts, regulation, the time of year and your own appetite for upheaval. A restaurant can't rebuild its kitchen systems in the festival season and a firm locked into a three-year software contract can't switch next month.

Don't treat constraints as excuses because they're simply the edges of the space you can work in and naming them early stops you from designing something nobody can ever carry out.

Step 6: Options

Now list what could be done including the cheap and boring options, and put doing nothing and doing very little on your list too. For each option you'll want to write what it fixes, what it costs, what it risks and what has to be true for it to work.

I'd aim for three or four real options because one option is a decision you've already made and ten is a brainstorm. The comparison is where you actually learn something, so state the trade-offs honestly even when they favour the choice you didn't want.

Step 7: Recommendation

Pick one and say why you picked it over the others. Write your reasoning so that someone who disagrees with you can find the exact point where they disagree. A recommendation that cannot be challenged can't be trusted either.

Keep it short, since one page is enough for most small and mid-sized businesses and clarity is what you're delivering, not the page count.

Step 8: Execution plan

A recommendation isn't finished until it says who does what, in what order, by when and at what cost. Most plans skip this step and if you ask me it is exactly where most good ideas die.

Write the plan with the people who will carry it out. A plan that reads well but needs skills you don't have is a document and not a plan. Name an owner for each action and then name the first action and make it small enough that someone can start it this week.

1. Describe what exists today from evidence, which is your current state.
2. Describe what should exist in terms you can test, which is your desired state.
3. State the gap between the two with a number or a clear description.
4. Ask why until you reach a root cause that someone can act on.
5. List the constraints, meaning what can't or should not change.
6. List three or four real options and include doing nothing among them.
7. Pick one as your recommendation and give reasons someone can challenge.
8. Write the execution plan with owners, order, dates and cost.

What evidence should you collect, and where do you find it?

Opinions are a start but not a finding. For each step try to find at least one piece of evidence that didn't come from a conversation, because conversations tell you what people believe while documents and records tell you what actually happened.

Talk to the people who do the work and not only to those who manage it, because the person who keys in the invoices knows things your finance head doesn't. Ask open questions like "walk me through the last time this happened" and then check what you hear against a record.

Don't wait for perfect data either. If the numbers are poor then that is a finding in itself and plenty of businesses find that the figure they argue about is defined three ways in three places. Agreeing on one definition can be the first and most useful fix you make.

When should you stop and do nothing at all?

Sometimes the gap is small, or it'll close by itself, or the fix costs more than the problem. Then the right answer is to change nothing or very little and a diagnosis that ends here isn't a failure because it just saved you the cost of a project.

Four signs point this way, and the obvious one is a gap that costs less than the fix. Another is a business about to change anyway through a move, a merger or a new product, so any fix would only be redone. A third is data that isn't ready, where a new system would just automate the confusion. The last is a root cause that is a decision you haven't made yet, and no tool under the sun can make it for you.

You'll also need to tell caution apart from avoidance and here's how I'd do it. Caution has a reason you can write down and a date to look again, while avoidance has neither. If you choose to wait then write down when you'll review it and what would change your mind.

Which traps tend to spoil a business diagnosis?

Four traps catch most owners, and the first is starting with the answer. If you already believe you need a new system then you'll only collect the facts that support it. So write your favourite solution on a sheet of paper at the start, set it aside and see at the end whether the evidence still points to it.

The second is listening only to the loudest voice, because the person who complains most isn't always closest to the cause. Speak to those who feel the problem and those who create it and then compare what each of them says.

The third is mixing up a symptom with a cause. Late reports, high staff turnover and slow quotes are all symptoms, so if your root cause reads like one of them then keep asking why.

The fourth is endless study, which is tricky because it feels like work. A diagnosis should end so set a date to stop collecting and start deciding. Most of the value comes from the first few findings while the last few add very little.

Share your findings with the people who gave you the evidence. They'll correct errors you missed and they're far more likely to back the plan if they've seen how you reached it, whereas a diagnosis that surprises the team on announcement day tends to meet resistance.

What does a good diagnosis produce in the end?

You should end up with a short written record that an outsider could read and understand. Here's what it needs to hold -

• A view of the current state that's backed by evidence rather than assertion.
• A gap stated in the terms your business already uses.
• Options with honest trade-offs, including the option of doing nothing.
• A recommendation with reasoning that you can challenge.
• An execution plan with owners, sequence and cost.

I judge the output by one simple test and it's a tough one. Could your own team carry out the plan without the person who wrote it? If yes then the diagnosis worked, but if the plan only makes sense with the author in the room then it isn't finished yet.

If the recommendation involves new technology or AI then the diagnosis also tells you whether it fits. Automating a process nobody understands only makes the confusion faster, so run the diagnosis first and the tool question gets a lot easier.

You can run these eight steps yourself. If you'd rather have an outsider do it so that history and office politics don't colour the findings then that's exactly what strategic advisory is for. Hope this helps, and happy digging!