Home Command Centre Our work About
Scattered pink data points drawn into one violet source, from which two lines, budget and actual, run side by side and track each other
← All articles
Management reporting

Build next year's budget on numbers your whole team agrees on

For a company whose financial year follows the calendar year, October is usually when next year's budget starts, and it often begins with three exports. Sales brings this year's revenue from the CRM, finance brings it from the ledger, operations brings a spreadsheet of its own, and the three do not match. The budget is built on whichever version wins the week, saved in a file of its own, and by the second month of the new year the budget against actual review has become an argument about whose numbers are right.

110 AnalyticsOctober 202614 min read

Why does the budget lose touch with the actuals?

This year's numbers come from three places. Sales counts revenue when a deal is signed, finance when it is invoiced, operations when the work is delivered. Each figure is right by its own definition, so the first budget meeting is spent deciding which one to start from, and the losing departments never quite trust the result. The longer version is in why two departments never agree on the number.

The budget lives in a file that cannot see the actuals. Once it is approved, the budget sits in a spreadsheet on someone's drive, while the actuals sit in the accounting system, the CRM and payroll. The two are often cut differently as well, the budget by outlet, product or department and the ledger by account code and cost centre. Comparing them means translating one into the other by hand, and the translation changes with whoever does it.

The comparison is a monthly job that nobody owns. Budget against actual needs fresh exports, that translation and someone with a spare morning. When the month is busy, it slips, and by spring the budget is a document people refer to in the past tense. If the monthly pack already takes days to build, the cost of that is larger than it looks, and a comparison that depends on someone finding the time is the easiest part to skip.

A small example shows how the translation goes wrong. Picture a hotel group that budgets food and beverage revenue by outlet. In this group's ledger, the restaurant and the bar post to separate account codes, and room service posts to a rooms code. Until someone decides, once and in writing, which codes make up the budget line, each month's comparison is rebuilt by whoever has the time, a little differently each time, and the variance on that line says more about the rebuild than about the business.

How does one data ecosystem fix it?

The fix is one data ecosystem the business owns, built so that the budget and the actuals sit in the same place and read the same definitions.

1. Actuals from every system, in one source. The accounting system, the CRM, the point of sale or booking system, payroll and the operational spreadsheets load into one data warehouse in the company's own account, on a schedule, usually overnight. Nobody exports anything, and the history the source systems overwrite is kept.

2. One definition per number. Revenue, margin, cost centre and headcount are agreed once with the finance lead and the department heads, written down and modelled in the warehouse. Every report inherits them, so next year's budget starts from figures that sales, finance and operations already accept.

3. The budget loaded beside the actuals. Once it is approved, the budget loads into the same warehouse and is mapped once to the same lines and the same definitions. From then on, management reporting shows budget against actual by month and year to date, with the variance on every line, refreshed overnight, so the comparison updates itself. Rules raise a flag when a line runs over, before anyone has to spot it.

On top of that sits an AI layer that answers questions in plain language from the same numbers. A managing director can type “where are we against budget this month, and what moved?” and get an answer drawn from the warehouse and the agreed definitions, the same figures the management pack shows.

The AI layer is the last thing switched on, once the definitions are settled, because an answer is only as sound as the numbers underneath it. Deciding what to do about a variance stays with the people who run the business.

What does it look like when it works?

Picture a Monday morning early in the new year, a few days after the first month has closed. The managing director opens the Command Centre and sees the month against budget, line by line, with the variance on each and the year to date beside it. Nobody built it over the weekend. It refreshed overnight, as it does every night, and the month showed as provisional until finance posted the closing entries.

The finance lead opens the same screen and sees the same figures. The sales director's revenue matches finance's revenue, because there is one definition of revenue and both of them read it. Each department head sees their own lines and their own variances, with access set by role. Any line opens onto the transactions behind it, so a question about a number becomes a look at the detail, with no request for another export.

One cost line ran over during the month, and a rule flagged it at the time, so the conversation about it has already happened. Before the monthly meeting, the managing director types “where are we against budget this month, and what moved?” into the AI layer. The answer comes from the warehouse and shows which lines are ahead, which are behind and where the gap is widest. A follow-up about which customers drove the revenue gap is answered from the same figures, in the same way.

What no longer happens matters as much. Nobody exports from three systems at month end. Nobody emails the budget file round for an update. The meeting does not open with a debate about which number is right, and the comparison does not wait for one person to come back from leave. The time goes on the few variances that matter and what to do about them, and the board pack is printed from the same figures the meeting used.

How do you get there, step by step?

The order matters, because each step depends on the one before it.

1. A data audit of the systems and the definitions. About three hours with your team, mapping every system that holds an actual, where each figure comes from and which reports are built by hand today. Nothing is installed. It ends with a quote for the work.

2. One source in the company's own account. The warehouse is built in your environment, under your account and your credentials, on a platform recommended before the build to suit what you already run. The systems behind the first view start loading on a schedule. If the engagement ever ended, the warehouse would keep running and stay yours, with no lock-in.

3. The definitions modelled once. Revenue, gross margin, cost centres and customer groups are agreed with the finance lead and the department heads, written in plain language and modelled in the warehouse. Your team's part is the agreeing; the building is ours.

4. The budget loaded beside the actuals. The approved budget is loaded into the same warehouse, and each budget line is mapped to the actuals it will be compared against. The monthly phasing loads with it, so each month is judged against its own share of the year. Where a line has no matching actual, the gap shows up early, while it can still be fixed.

5. The first live view. Usually three to four weeks from an accepted quote, the first view is live and refreshing on its own, starting with the part of the reporting agreed as the priority in the audit. Before anyone relies on it, it is checked with the finance lead against the version they built by hand.

6. The AI layer, switched on last. Once the definitions are settled and access is set by role, the AI layer is switched on. It reads from the same warehouse and the same definitions as every report, so its answers and the management pack draw on the same figures.

From there, a fractional team keeps the loads running and the definitions current as the business changes, and adds the next report when the business needs it. The warehouse reads from the accounting system and never posts into it, so the ledger your accountant works in stays exactly as it was. Started in October, with budget against actual agreed as the first view, that sequence leaves room for it to be live before the new year begins, so the new budget is tracked from its first month.

Which mistakes make a budget hard to track?

Much of the trouble with budget against actual is built in during October, before a single actual arrives.

Rolling last year's budget forward. Copying last year's file and adjusting it carries last year's assumptions into next year, including the ones the actuals have already proved wrong. Start from this year's actuals, taken from one source, and change only what the business expects to change.

Budgeting in more detail than anyone will review. A line for every cost and every product feels thorough in October. By spring, most of those lines go unread, and the ones that matter are lost among them. Budget at the level leadership steers by, and keep the finer detail for the lines that move the result.

Spreading the year evenly across the months. A business with a strong summer and a quiet winter that divides its annual budget evenly will look behind in the quiet months and ahead in the busy ones, and the variance stops meaning anything. Phase each line by the pattern this year's actuals show.

Leaving the assumptions in someone's head. Prices, headcount, opening dates and new contracts sit behind every line. Write them beside the numbers, so that when a variance appears you can tell whether the business moved or the assumption was wrong.

Reading an open month as if it were closed. Some figures are only final once the month-end entries are posted, such as accruals, prepayments and depreciation. A comparison taken before the close can look wrong for reasons that have nothing to do with performance. Treat an open month as provisional and judge the variance after the close.

What can you do this week?

These steps help before the budget is built, whatever tools you use.

1. Run this year's comparison once, by hand. For a handful of lines, set this year's actuals beside this year's budget. Note how long it takes, which lines have no matching actual and where people disagree about the figure. That list is the brief for building next year's.

2. Choose one source for each starting number. Before anyone opens the budget template, agree which system this year's revenue, costs and headcount come from, and write it at the top of the file. Everyone then builds from the same starting point, even if it is imperfect.

3. Write down the definitions the budget depends on. One line each for revenue, margin and headcount, agreed by the heads of sales, finance and operations. State whether revenue means signed, invoiced or collected, and which costs sit inside the margin. Keep the page with the budget. For other common numbers, the KPI examples by department note what each definition has to settle.

4. Build the budget on the lines the actuals already use. Use the account codes, cost centres and any outlet or product groupings the accounting system already holds, so every budget line has an actual line to sit beside. Where the budget needs a split the ledger does not have, decide now where that actual will come from. A simple list is enough: each budget line on the left, the account codes or cost centres it covers on the right, and a note against any line that has no matching actual yet.

5. Name an owner and a date for the first comparison. Give each budget line one owner, and put the first budget against actual review in the diary for the first month of the new year, while the budget still has everyone's attention.

None of these steps needs new tools. They make next year's budget easier to track by hand, and they are the same groundwork a data warehouse would start from, so none of the effort is wasted if you build one later.

Frequently asked questions

What is the difference between management reporting and the financial statements?
The financial statements are the formal annual accounts, prepared by your accountant under the reporting framework that applies to your company. Management reporting is internal and far more frequent, laid out by whatever cut leadership steers by, and budget against actual usually sits at its centre. The first is the record of the year; the second is how the business decides what to do next month.

Can we build next year's budget from the financial statements?
Only in part. This year's statements will not exist until after the year closes, and last year's are annual and summarised, so they show neither the recent months nor the outlet, product or customer a working budget is built by. They are a sound check on the totals. The detail has to come from the systems underneath, which is why budgets end up built from exports.

What is a budget variance?
The difference between the actual figure and the budget on a line, for the month or the year to date. A favourable variance helps the result, such as revenue ahead of budget or a cost below it; an adverse one works against it. A variance is only worth discussing when both figures use the same definition and the month has closed.

How often should budget against actual be reviewed?
At least monthly, once the month has closed, with the lines that move fastest, such as sales or staff hours, watched more often. When nobody has to build the comparison, it can be open every day, with the current month marked provisional until the close.

Do we need to stop using our budgeting spreadsheet?
Usually not. Finance can keep building the budget in the spreadsheet it knows. The approved version is then loaded into the warehouse beside the actuals, so the comparison no longer depends on anyone opening the file.

Does this change anything for our accountant?
No. Your accountant keeps preparing the financial statements from the same accounting system. The warehouse reads from that system and never posts into it, so the management figures and the year-end figures start from the same ledger. We do not do bookkeeping, and we do not prepare the statutory accounts.

What if the budget is already approved by the time this is built?
It can be loaded at any point in the year. The warehouse loads the actuals with the history the systems hold, so the comparison can cover every month of the budget, including the months before it was loaded.

At 110 Analytics we build a Command Centre: the whole business on one screen, one set of numbers, built in an account the business owns, with an AI layer on top that answers questions from the company's own data. If you want next year's budget built on numbers your team already agrees on, and tracked against the actuals without anyone rebuilding it, book a strategy call.

This October: our data audit, standard price €8,000, is free for up to three companies. Ask for a place when you book your strategy call.

Glen SultanaFounder · 110 Analytics

Want to talk this through on your own numbers?

Thirty minutes with Glen, and a straight answer on whether 110 Analytics is the right fit.

Thirty minutesFreeNo pitch deck
01Where your data is fragmented, and what it is costing you in decisions and reporting hours.
02The one report or view that would change how the business runs, and whether the data for it already exists.
03A straight answer on whether 110 Analytics is the right fit.