Why Year End Numbers Arrive Too Late

Your year ends in December. The accounts arrive in May. They are correct, they are properly prepared, and by the time you read them the information is five months stale.

Nobody is at fault. Statutory accounts exist to satisfy a filing obligation, and they do that well. The mistake is expecting them to also tell you how the business is doing.

Three reasons they cannot run a business

They are too late to act on. A problem visible in the May accounts started in the previous February. Fifteen months of it happened before anybody could respond.

They are aggregated past the point of usefulness. One revenue line, one cost of sales line, one figure for wages. The question you actually have is which type of work makes money and which does not, and that is precisely what aggregation removes.

They are built for a different reader. The categories exist because a tax authority requires them, not because they map to how you think about the business. Nothing in a set of statutory accounts corresponds to the three or four things you actually decide.

What management information is meant to be

A separate thing, produced monthly, deliberately less precise and dramatically more useful.

It does not need to reconcile to the penny. It needs to be available on the fifth of the month, split the way your business is actually organised, and answer whether the thing you changed last quarter worked.

Most owners of businesses between five and fifty million have never had this, and assume the reason is cost. Usually the reason is that assembling it required somebody to pull figures out of three systems that disagree, every month, forever.

Why the monthly pack never survived

Almost every business has attempted one. It runs for four months and stops.

The cause is always the same. It took two days to produce, those two days came out of somebody’s month end, and the moment things got busy it was the first thing to go. And because it was manual, each month’s version was slightly different, so comparing March to January required knowing what somebody did differently in February.

A report that depends on a person having a spare two days is not a report. It is an intention.

What AI changed about this

The obstacle was never the arithmetic. It was reconciling systems that describe the same business differently, which is judgement work and therefore needed a person.

Your accounting package, your job system and your CRM each hold a version of the same client, the same project and the same month, with different names, different categories and different cut off dates. Matching them was the two days, and it is exactly the sort of messy, contextual matching that AI now handles.

What that makes possible: the pack assembles itself on the fifth, identically every month, from all three systems. Nobody spends two days on it, so it survives a busy quarter.

Two things beyond speed. It can explain a movement rather than only show it, because it can look across the underlying records and say that the margin fell because three jobs in one category ran over, not because pricing moved. And it can be asked a question in plain language between reports, which is when most questions actually occur.

What belongs in it

• Revenue and margin split the way you actually run the business. By service line, by client type, by branch. Not by statutory category.

• Cash, forward. What is owed to you, what you owe, and what the next ninety days look like. This is the number that ends companies and it is absent from most monthly packs.

• Two or three operational numbers that lead the financial ones. Enquiries, quotes issued, win rate, utilisation. These move first and give you warning.

• Comparison against the same month last year, not against last month, so seasonality does not read as a trend.

Four things, one page. Anything longer does not get read, and a pack nobody reads is the same as no pack.

The precision trap

The objection to monthly management information is always accuracy, and it is the reason most attempts get abandoned.

Somebody notices the March figure moved when April was produced, because an invoice was posted late or an accrual changed. The pack loses credibility, people stop trusting it, and within two quarters everybody is back to waiting for the statutory accounts.

The correct response is to state the tolerance up front. This pack is accurate to within a few percent, it will move slightly as things settle, and it is for direction rather than for filing. A number that is roughly right on the fifth beats a number that is exactly right in May, and saying so in advance prevents the first restatement from killing the whole exercise.

What a buyer will ask for

One more reason to build this, and it is the one that pays largest.

Anybody looking at buying your business, lending to it or investing in it will ask for monthly management accounts going back two or three years. A company that produces them on the fifth every month reads as well run. A company that has to reconstruct them reads as a risk, and that shows up in the price or in the structure of the deal.

It is the cheapest thing you can do now that materially affects a transaction later, and it takes three years of history to be worth anything, which means starting matters more than perfecting.

Your accountant is still doing their job

Worth being clear, because this reads as a criticism and is not.

Statutory accounts are a compliance product with a legal purpose and a required format, prepared to a standard. Asking them to double as a management tool is asking the wrong thing of the right document.

Many accountants will produce management information if asked, and some are very good at it. The point is that it is a separate engagement with a separate purpose, and if nobody has ever raised it, it is because you have not asked rather than because it is unavailable.

Where to start in January

Write down the three questions you wish you could answer on the fifth of every month. Most owners produce something like: is the pipeline strong enough for the next quarter, which work is actually profitable, and will cash hold if a large client pays late.

Those three questions are the specification. Build the smallest possible thing that answers them, produced automatically, and ignore everything else until it has run for six months without anybody touching it.

AI Optimize reconciles the systems that disagree and assembles the same one page pack on the fifth of every month, without anybody losing two days to it. That work sits under Reporting & Data.