Budget Variance Commentary

Budget Variance Commentary is a routine your Townie AI assistant runs after month-end actuals land: it compares actuals to budget and forecast, surfaces the three to five lines that genuinely moved, and drafts the explanation for each — pulling the 'why' from your email, Slack threads, and meeting notes, with nudges drafted for anything still unexplained.

To: your Townie

Subject: Create me a Budget Variance Commentary routine

Build me a routine that drafts the 'why' behind my budget variances, because the variance report itself is easy but writing the commentary every month is the real grind.

Run it once a month after actuals are ready (I'll tell you where they land: a sheet, an export, or our accounting system), and let me kick it off on demand.

When it runs:
- Compare actuals to budget and to the latest forecast, and surface only the 3-5 lines that actually moved (the biggest and most unexpected swings), not every account.
- For each one, draft the explanation: what moved, by how much, and the likely 'why', and go find that why where it already lives: my emails, Slack threads, and meeting notes (a vendor true-up, a delayed hire, a one-time charge, a timing shift).
- Where you can't find a reason, don't invent one. Draft a short nudge to the budget owner asking them to explain their line, and hold it for me.
- Flag anything that looks like a data error (a sign flip, a missing month, a line that doubled) so I catch it before it goes into a deck.

Output: month-end variance commentary, the 3-5 material lines with a drafted 'why' for each, plus ready-to-send nudges to owners for anything still unexplained.

My default: do the analysis, the commentary, and the draft nudges on your own, but hold any message to a budget owner for my okay before it sends.

Before you build this, tailor it to the integrations I actually use and what you already know about me. Don't take the above at face value; ask me the critical questions that make it stronger: which tools and integrations to use, what the output should look like, and how much autonomy you should have (what to do on your own vs. check with me first).

Sent when you sign up — works as-is, or edit anything.

Get started with this routine

Nudges to budget owners wait for your okay.

See it work: the commentary written before you open the file

Draft the ‘why’ behind my budget variances

Only the lines that moved — with the reason already found.

How this routine works

Works with

It runs when actuals are ready
Once a month, after your close — or on demand when finance asks early. You tell it where the numbers land: a sheet, an export, or your accounting system.
It hunts the why, not just the what
For the three to five lines that genuinely moved, it searches your email, Slack threads, and meeting notes for the real driver — a vendor true-up, a delayed hire, a timing shift.
You get commentary, not homework
Each material line arrives with its explanation drafted. Anything unexplained comes with a ready-to-send nudge to the budget owner, held for your okay.

Budget Variance Commentary FAQ

It's a routine your Townie AI assistant runs each month after actuals land: it compares actuals to budget and forecast, surfaces only the three to five lines that materially moved, and drafts the explanation for each — pulling the reason from your email, Slack, and meeting notes. Where no reason exists, it drafts a nudge to the budget owner instead of inventing one.

You do. The commentary is written for you, and every nudge to a budget owner is held for your approval — nothing sends itself. Read more in our security overview.

No — that's the point. If the why isn't in your email, threads, or notes, the line is marked unexplained and a short nudge to its owner is drafted instead. Guessed commentary is worse than none.

Wherever actuals land — a spreadsheet or an export works fine — plus your email, and optionally your team chat and meeting notes, which is where the explanations usually hide.

Yes. Sign flips, a missing month, a line that suddenly doubled — anything that looks like a data problem rather than a business story gets flagged separately so you catch it first.