The Budget That Gets Approved and Filed
Budget season takes six weeks. Department heads submit numbers, finance consolidates, there are two rounds of challenge, the board approves it, and the workbook is saved.
Then nothing. Actuals accumulate in the accounting system. The budget sits in the workbook. Comparing them means somebody exporting a trial balance, mapping accounts to budget lines by hand, and building a variance report, which happens quarterly if someone insists, and in many businesses only at year end, when the comparison is a post-mortem rather than a control.
The budget was not useless. It forced a planning conversation, set expectations and allocated resource. But as a management instrument it stopped working the day it was approved, because nothing connects it to what is actually happening.
This article covers the structure decision that determines whether budgeting works at all, the analytic accounting prerequisite most projects underestimate, how to build a budget department heads will own, variance reporting that changes behaviour, and how to reforecast without abandoning the original plan. It is written for CFOs, controllers and finance managers.
The Business Problem: A Plan With No Feedback Loop
- The budget lives outside the accounting system. It is a workbook, structured the way the planning process needed, which is not the way the ledger is structured.
- The structures do not match. The budget has lines like βmarketingβ and βregional operationsβ. The ledger has accounts. Mapping one to the other is manual, and it is done afresh every time somebody wants a comparison.
- Nobody owns a line. Budgets are approved at an aggregate level and never pushed down to a person who is accountable for a specific number.
- Variance is explained after the fact. By which time the spend has happened and the explanation is history rather than a decision.
- Commitments are invisible. A purchase order raised against a budget is a commitment, but the budget only sees it when the invoice posts, often a month later, by which time the budget holder has committed more.
- Reforecasting destroys the baseline. When the forecast is revised, the original budget is overwritten, so nobody can see how far the year has moved from the plan.
- Department heads do not see their own numbers. They receive a report from finance, late, in a format built for finance.
What an Unmonitored Budget Costs
- Overspend discovered after the fact. The most obvious cost, and the most avoidable. A line that was going to overrun was visible months before it did, to anyone who could see it.
- No ability to act in time. Even a correct year-end variance analysis cannot change anything. The value of a variance is entirely in how early it appears.
- Budget holders disengaged. People do not manage to a number they cannot see. Handing a department head a quarterly report prepared by someone else produces compliance at best.
- Finance consumed by reconciliation. The most capable people in the function spending days mapping ledger accounts to budget lines rather than analysing anything.
- Planning credibility erodes. If last yearβs budget was never monitored, this yearβs budget round is treated as theatre, and the quality of submissions falls accordingly.
- Capital and headcount decisions made blind. Without a reliable view of committed and actual spend against plan, approvals are given on instinct.
The Structure Decision That Comes First
Almost every budgeting failure traces back to this, and it is decided before any configuration.
The budget must be structured the way the ledger records things. Not approximately, but exactly. If the budget has a line the ledger cannot report, that line will never have an actual against it and it will be reconciled by hand forever.
That means deciding the dimensions you budget on. Typically some combination of:
- Account: the nature of the spend
- Cost centre or department: who is responsible
- Project, site or branch: where the activity happens
- Product line or business unit: what it relates to
Pick the minimum that answers your real questions. This is the discipline that matters. Every dimension you add multiplies the number of budget lines, the effort to maintain them, and the chance that transactions get coded wrongly. Three well-maintained dimensions beat five half-populated ones, and a business budgeting on two dimensions it can actually enforce is in a better position than one with an elegant model nobody codes to.
Decide the granularity of the period. Monthly budgets give early signals and take longer to prepare. Quarterly is less effort and detects problems later. For most businesses, monthly for a small number of volatile lines and quarterly for the rest is a sensible compromise, and it is easier to defend than an arbitrary across-the-board choice.
Agree what is budgeted and what is not. Not every account needs a budget. Budgeting trivial lines generates variance noise that obscures the ones that matter.
Analytic Accounting: The Prerequisite Nobody Scopes
If you want to budget on anything other than the account code (and almost everyone does), the transactions have to carry that information. In Odoo this is analytic accounting, and it is the part of the project that is consistently underestimated.
- Design the analytic plan before go-live if you possibly can. Retrofitting dimensions onto historical transactions is impractical at any scale, which means a business that adds dimensions later starts its comparison from the date of the change rather than from the start of the year.
- Keep the plan small and maintained. Three dimensions that are populated on every transaction are worth far more than seven that are populated sometimes. A dimension that is optional will be blank on a meaningful proportion of entries, and blank means unreportable.
- Make analytic distribution mandatory where it matters. Optional fields do not get filled. If a cost centre is required for the budget to function, it should be required at entry, with the inevitable argument about convenience settled in favour of the reporting.
- Be realistic about where coding happens. Somebody has to assign the dimension, usually at purchase or invoice entry, and usually that personβs priority is processing the document rather than the reporting. Defaults, supplier-level rules and simple lists make this survivable; relying on judgement at data entry does not.
- Check your edition. Odooβs own comparison lists accounting, including comprehensive accounting, among the items that differ between editions. Confirm what analytic capability is available to you before designing a budget model that depends on it.
Building a Budget People Will Own
The mechanics of entering budget figures are trivial. Getting them owned is not.
- Push the line to a named person. Each budget line should have a holder who is accountable for it and who can see it. A budget owned by βthe departmentβ is owned by nobody.
- Build it with the holder, not for them. Figures imposed centrally are defended rather than managed. Figures the holder constructed are figures they will explain when they move.
- Give them visibility without a request. The single most effective change in most businesses. A budget holder who can see their own position at any time behaves differently from one who receives a report six weeks after the period.
- Base the opening position on something. Prior year actuals, adjusted for known changes, is the usual starting point and it is far better than a round number or last yearβs budget, which may itself never have been achieved.
- Separate the annual total from the phasing. The total is what gets approved. The phasing is what makes monthly variance meaningful, and a budget phased in twelve equal parts will generate false variances every month for seasonal costs.
- Capture the assumptions. What headcount, what volume, what price. When the variance appears, the first question is whether the assumption changed, and if the assumptions were never written down, that question cannot be answered.
Variance Reporting That Changes Behaviour
A variance report that nobody acts on is a reconciliation exercise with a chart on it.
- Report early, even if incomplete. A variance visible in week one of the following month is actionable. The same variance reported perfectly six weeks later is history.
- Include commitments, not just actuals. Purchase orders raised but not yet invoiced are spend that has happened. A budget holder looking at actuals alone will commit over their budget without knowing it, and discover it when the invoices arrive.
- Set a materiality threshold and only chase what crosses it. Explaining every small variance trains everyone to treat the process as administrative. Agree a threshold (by value or percentage) and focus attention there.
- Require a reason and an action, not just a number. The useful output is not that marketing is over by a given amount; it is why, and what happens next. Two sentences from the holder is enough.
- Distinguish timing from permanent. A cost that arrived earlier than phased is a timing difference and will correct itself. A cost that is structurally higher will not. Treating the two the same produces either false alarm or false comfort.
- Report year-to-date and full-year expectation together. Year-to-date says what happened; full-year expectation says what it means. Most decisions need the second.
Reforecasting Without Abandoning the Budget
Businesses change during a year, and a plan that cannot be revised becomes irrelevant. But revision done badly destroys the thing that made the budget useful.
- Keep the original budget intact. It is the baseline against which the year is judged, and overwriting it removes the ability to see how far things have moved. Reforecasts should sit alongside it, not replace it.
- Reforecast on a defined cadence, not whenever somebody asks. Quarterly suits most businesses. Continuous reforecasting means nobody knows which version is current.
- Be explicit about what changed and why. A reforecast without an explanation is just a new set of numbers, and it invites the suspicion that targets are being moved to hide variance.
- Resist reforecasting to match actuals. If every reforecast simply restates what has happened, the variance disappears and so does the information. The point of a forecast is the part that has not happened yet.
- Decide what the business is managed against. Original budget, latest forecast, or both. It is a legitimate choice either way, and the failure is leaving it ambiguous, because then different people are measuring against different numbers.
Approaches Compared
| Consideration | Spreadsheet budget, manual comparison | Budget in the system, account level only | Budget with analytic dimensions | Dedicated planning tool |
|---|---|---|---|---|
| Comparison effort | Manual mapping each period | Automatic | Automatic | Automatic |
| Departmental accountability | Possible, manually | Limited | Native, by cost centre or dimension | Native |
| Budget holder self-service | None | Possible | Yes | Yes |
| Commitments visible | No | Usually not | Possible with design | Native |
| Phasing and seasonality | Manual | Supported | Supported | Native, often with drivers |
| Reforecasting with baseline retained | Manual, often lost | Possible | Possible | Native, with scenarios |
| Prerequisite | None | Chart of accounts | Analytic plan, populated consistently | Integration plus data quality |
| Best fit | Very small businesses | Businesses budgeting only by nature of spend | Most mid-market businesses | Complex planning with driver-based models |
Best Practices
- Structure the budget the way the ledger records things, exactly rather than approximately.
- Choose the minimum number of dimensions that answers your real questions, and enforce those properly.
- Design the analytic plan before go-live. Retrofitting it is impractical at scale.
- Make analytic distribution mandatory where it is load-bearing, and settle the convenience argument in favour of the reporting.
- Give every budget line a named holder with visibility of their own position.
- Build budgets with holders rather than for them.
- Phase the budget realistically rather than splitting the annual figure equally.
- Write down the assumptions behind each material line.
- Report early and include commitments, not just posted actuals.
- Keep the original budget intact when reforecasting, and reforecast on a fixed cadence.
Common Mistakes
- A budget structured differently from the ledger. Guarantees manual mapping forever, which is the problem the project was meant to solve.
- Too many dimensions. Multiplies maintenance and coding error; three enforced beat seven optional.
- Optional analytic fields. Blank means unreportable, and blanks accumulate quietly.
- Retrofitting dimensions mid-year. Comparison starts from the change date rather than from the start of the year.
- Equal monthly phasing. Generates false variances every month on anything seasonal.
- No named budget holder. A line owned by a department is owned by nobody.
- Reporting actuals without commitments. Budget holders over-commit without knowing it.
- Chasing every variance. Trains everyone to treat the process as administrative noise.
- Overwriting the budget when reforecasting. Removes the baseline and invites the suspicion that targets are moving.
- Reforecasting to match actuals. The variance disappears and so does the information.
An Illustrative Scenario: A Multi-Site Services Business
The following is a composite illustration drawn from patterns common to multi-site businesses. It is not an account of a specific Techvaria client, and no figures below come from any real engagement.
Consider a services business operating from four sites, with a central function and a budget round that consumed most of November each year.
The Trigger
A site overran its annual cost budget and the overrun was identified in the year-end accounts. Reviewing it afterwards, the finance director established that the trajectory had been visible from about month four to anyone who had been looking, which nobody had, because the comparison only happened at year end.
What the Review Found
The budget workbook was structured by activity (things like βsite operationsβ and βcentral supportβ), while the ledger was structured by account. Mapping between them was a manual exercise the finance manager performed, which is why it only happened twice a year.
Analytic coding existed but was optional, and a meaningful proportion of transactions carried no cost centre at all, which meant site-level reporting was incomplete even where someone tried to produce it.
Budget lines had no named holders. Site managers received a summary from finance at half year and were asked to comment on numbers they had not seen being formed.
Purchase commitments were invisible. One site had raised orders well ahead of invoicing, which was a substantial part of why the overrun was not apparent from the posted actuals.
What Was Implemented
Structure first, and the hardest conversation in the project. The budget was rebuilt to match the ledger exactly, which meant the planning process had to change shape: department heads had been submitting in activity terms and now submitted against accounts and cost centres. Several found this harder initially, and the finance director held the line on it because the alternative was permanent manual mapping.
The analytic plan was reduced from five dimensions to three (account, cost centre and site), and cost centre was made mandatory at entry. The argument about convenience at data entry took two meetings and was settled in favour of the reporting.
Each budget line was assigned to a named holder with direct visibility of their own position, which removed the dependency on a finance-produced report.
Phasing was done deliberately rather than equally, which immediately reduced the false variances that had previously appeared every month on seasonal costs.
Commitment reporting was added, so budget holders saw raised orders alongside posted actuals.
A quarterly reforecast was introduced with the original budget retained alongside it, and the business agreed explicitly that it would be managed against the original budget with the reforecast as context.
What Caught Them Out
Transactions coded before the dimension change could not be retrofitted, so the first yearβs site-level comparison ran from the change date rather than from the start of the year. The finance director described this as the argument for doing it at go-live that nobody listens to until afterwards.
After Two Quarters
Variances were being discussed in the month following rather than at half year. The commitment view changed behaviour most visibly: one site manager, seeing his raised orders against his budget for the first time, deferred a planned purchase.
The finance managerβs own assessment was that making cost centre mandatory had been the single change that mattered, because every report downstream had quietly depended on it.
Industry Use Cases
- Multi-site services. The core case: site-level accountability, central functions, and a budget that must report by location. See professional services.
- Manufacturing. Departmental and cost centre budgeting alongside production costs, where the analytic plan serves both purposes and should be designed once. See manufacturing solutions.
- Education. Budget holders across departments and campuses with distinct funding sources, where dimension design carries reporting obligations as well as management ones. See education solutions.
- Healthcare. Departmental budgets with high-value consumables and equipment, where commitment visibility matters more than usual. See healthcare solutions.
- Construction. Project budgets alongside overhead budgets, where the two need different treatment and frequently get the same one. See construction.
- Trading and distribution. Branch-level budgeting where margin and overhead are managed by different people. See trading and distribution.
Frequently Asked Questions
Almost always because the budget is structured differently from the ledger, so comparison requires manual mapping, which means it happens rarely. Restructuring the budget to match exactly how the ledger records things is the fix, and it usually requires changing the shape of the planning process rather than the accounting.
Only if you want to budget on anything beyond the account code, which almost everyone does, because the useful questions are by department, site, project or business unit. Those dimensions have to be carried on the transactions, which means an analytic plan designed and consistently populated. Confirm what is available in your edition first, since accounting depth differs between Odoo editions.
The minimum that answers your real questions. Each additional dimension multiplies budget lines, maintenance effort and coding error. Three that are mandatory and consistently populated are worth more than seven that are optional, because an optional dimension will be blank on a meaningful share of transactions and blank is unreportable.
You can add them, but you cannot sensibly retrofit them onto historical transactions. That means your comparison starts from the date of the change rather than from the start of the year. This is the strongest argument for designing the analytic plan before go-live.
Only for genuinely even costs. Equal phasing on seasonal or lumpy costs generates a false variance every month, which trains budget holders to ignore variance reporting. Phase deliberately, even if approximately; it is one of the cheapest improvements available.
Because a purchase order raised is spend that has effectively happened. A budget holder looking only at posted actuals will commit beyond their budget without knowing, and find out when the invoices arrive, which is usually too late to do anything about.
Keep the original budget as the baseline and hold reforecasts alongside it rather than overwriting. Reforecast on a fixed cadence, explain what changed and why, and decide explicitly whether the business is managed against the original budget or the latest forecast. Ambiguity there means different people measure against different numbers.
Conclusion
The budget that gets approved and filed is not a bad budget. It did its job during the planning round: the conversation happened, resource was allocated, expectations were set. What it never became was a control, because nothing connected it to what the business actually spent.
Making that connection is mostly structural. The budget has to be shaped the way the ledger records things, exactly, or every comparison is a manual mapping exercise that will therefore happen rarely. The dimensions you budget on have to be carried by the transactions, which means an analytic plan designed before go-live, kept small, and mandatory where it is load-bearing. And every line needs a named holder who can see their own position without asking.
Then report early rather than perfectly, include commitments alongside actuals, chase only what crosses a materiality threshold, and ask for a reason and an action rather than a number. When you reforecast, keep the original intact, and resist the temptation to restate what has already happened, because that is the part of a forecast that carries no information.
The most common failure in all of this is not technical. It is agreeing to too many dimensions and enforcing none of them, which produces an elegant model and blank fields.
Make Your Budget Reportable
Techvaria is an official Odoo Silver Partner with offices in Bangalore, Ahmedabad and Dubai.
A budget and reporting review produces four things: your current budget structure compared against how your ledger actually records transactions, with the mapping gaps identified; an analytic plan sized to the questions you genuinely need answered rather than the ones that sound useful; a budget holder model with named owners and self-service visibility; and a variance reporting pack that includes commitments and distinguishes timing from permanent differences.
If your last overrun was identified after the year closed, that is both the scope and the business case.
Talk to our Odoo consultants, or see how we approach implementation.
Make Your Budget Reportable

Mustufa Rahi is an Odoo Certified Functional Consultant and ERP expert at Techvaria with 15+ years of experience in implementation, automation, and business process optimization, helping organizations scale efficiently.
