Budget vs Actuals Dashboards: A Founder's Checklist

Most founders see budget versus actuals as a PDF that lands three weeks after the month ends, already out of date and full of numbers nobody argues with because nobody trusts them. The damage is quiet: a hiring plan approved on a spreadsheet, a marketing line drifting over for two quarters, GST and TDS outflows that surprise the bank balance. If finance is one person plus a part-time CA, this is the default state — and it is fixable in about a month of part-time work.

Start with the decisions you make monthly

A budget versus actuals dashboard is worth building only for the four or five decisions you actually take each month, so list those first and design backwards. Typical ones for an Indian SMB: whether to add the next hire, whether a client's gross margin justifies the servicing cost, whether marketing spend is buying pipeline or just impressions, and whether cash covers the coming GST and TDS cycle. Write each decision as a question with a threshold — "if the cloud bill is more than 20% over budget two months running, we renegotiate the commitment" — and you have your minimum viable contents. Anything that does not feed one of those questions is decoration, and decoration is what breaks first.

Clean the chart of accounts before you buy any tool

A dashboard can only be as clean as the ledger it reads, so the first real work is mapping your chart of accounts to budget lines one-to-one. In most SMB books the same spend lands in three places: SaaS subscriptions split across "office expenses" and "IT", contractor payments inside "professional fees", and payment gateway charges buried in "bank charges". Fix the mapping in Tally, Zoho Books or QuickBooks, freeze it, and refuse to let a new vendor create a ledger without a deliberate decision. If invoicing and GST entries are still half-manual, GST billing automation is worth doing first — a dashboard reading a ledger with mismatched tax entries reports variance you cannot explain.

Set budget granularity you can actually maintain

Budget at the level you can forecast honestly — usually 15 to 25 lines, not 90 — and keep the extra detail in the actuals only. A services company with ₹4 crore of revenue rarely knows its electricity cost twelve months out, but it does know headcount, salaries, rent, and its two largest vendor contracts. Rules that hold up in practice:

• Budget monthly rather than dividing the year by twelve. Salaries, quarterly SaaS renewals, insurance premiums and advance tax instalments are lumpy.
• Separate revenue from collections. GST collected is not your money, and TDS deducted at source means the bank credit is not the invoice value.
• Keep a single "unbudgeted" line and review it monthly instead of stretching other lines to hide surprises.
• Freeze the budget at sign-off and log changes in a dated note, so variance is a conversation about the business and not about who edited cell F14.

Wire the pipeline and prove one number by hand

Numbers should arrive on a fixed schedule without anyone exporting CSVs, so pick the simplest pipeline that survives a busy week. For most SMBs that is a nightly export from the ledger or billing system into a small database — Postgres, or even a well-structured sheet — with one dashboard layer on top in Metabase, Looker Studio or Power BI. Before trusting it, reconcile one month manually: pull May's bank statement, tick every software subscription back to an invoice, and confirm the total matches the dashboard's software actual to the rupee. If it is off by ₹18,000, you have found either a missed invoice or a mis-mapped ledger, and both are worth knowing. Agree upfront who gets pinged when the refresh fails, because silent stale data is worse than no dashboard.

Review variances in a fixed 45-minute slot

A dashboard nobody opens on a schedule is an expensive screensaver, so book a recurring 45 minutes on day five of the following month and keep the agenda identical every time. Walk the same three things: lines more than 10% and more than ₹25,000 off budget, the unbudgeted line item by item, and a one-line reason for each large miss recorded next to the number. Keep those reasons, because after three months patterns surface — the same client always pays late, the same vendor always overruns. Where transaction volume makes manual scanning unrealistic, anomaly flagging via AI-assisted review can pre-sort the noise, but the decision and the note stay with a human. Move the meeting when you must; never skip two in a row.

Where to start

Do one month by hand before automating anything: take your top 15 spend lines, pull last month's actuals from the ledger, and build the comparison in a sheet. If that takes more than half a day, the problem is the account mapping, not the tool — fix it and the dashboard becomes a two-week job instead of a six-month one. When you want the pipeline, the variance logic and the review habit set up properly in one go, that is the kind of work we do at Maysan Labs.