It's quarter-end. Your CFO emails: "Need the R&D allocation breakdown for tax credits by Friday. How much engineering time went to R&D vs maintenance last quarter?" You groan. This again.
The Problem
You go through the ritual. Pull up the team list. Start guessing. "Alex probably spent 80% on R&D... no wait, he was doing that database migration most of the quarter, that's maintenance. Maybe 60%? Or 50%?" You have no real data. You're reconstructing three months of work from memory. For 50 engineers. You Slack team leads: "Quick, what percentage of your team's time was R&D vs maintenance last quarter?" They guess too. "Mostly new features, so... 70% R&D?" Are they right? Who knows. You compile the guesses into a spreadsheet. Finance uses it to claim $400K in R&D tax credits. Maybe you're leaving money on the table. Maybe you're over-claiming and creating audit risk. You honestly don't know. The process is painful, time-consuming, and produces data you don't trust. And you'll have to do this again next quarter. Forever.
How It Cascades
You leave tax credits on the table. Maybe your engineers actually spent 75% on R&D but you guessed 60%. That's tens of thousands in unclaimed credits. Multiply that by four quarters.
Or worse: you over-claim. Finance submits 80% R&D when it was actually 65%. In an audit, you can't defend the numbers. The company pays penalties plus interest. Your CFO is not happy with you.
Every quarter-end becomes a scramble. Finance needs the data. You drop everything to compile estimates. Your engineers get interrupted for their guesses. Everyone hates it.
Capitalization reporting is similarly weak. Some engineering work should be capitalized on the balance sheet, other work should be expensed. You're guessing there too, which affects financial statements that investors and boards review.
You can't optimize. Maybe certain teams consistently do more R&D work. You could structure comp or teams around that. But without data, you'll never know.
The Insight
The problem isn't that the finance team is being difficult. They genuinely need accurate R&D allocation for tax purposes and financial reporting. The problem is that engineering effort categorization is fundamentally a data problem that you're solving with human memory and guesswork. The actual work is happening in commits, PRs, and tickets. The categorization should happen automatically from that data.
"I go through and guess on allocations of people when I do these exercises, which is not fun or very realistic. If we had the ability to look at code impact by category over six months to a year, that would really streamline what we could do for R&D tax credits."
The Solution
Maestro automatically categorizes every piece of engineering work: Features (new capabilities—counts as R&D), Bugs (fixing broken functionality—typically maintenance), KTLO (keeping lights on—definitely maintenance), infrastructure and tooling (depends on whether it's net-new innovation or upkeep). The categorization happens continuously, in real-time, using AI analysis of commit messages, PR descriptions, and ticket types. When quarter-end arrives, you don't scramble. You open Maestro and export the report. Q3 2024: 73.2% R&D, 18.4% maintenance, 8.4% support. Broken down by month, by team, by engineer if needed. The data is precise, defensible, and backed by an audit trail of actual work. Finance is thrilled. Instead of trusting your guesses, they have comprehensive data they can stand behind in an audit. You file for $480K in R&D tax credits instead of the $400K you would have conservatively guessed. The extra $80K more than pays for Maestro. One VP reported: "I used to guess on allocations, which was not fun or realistic. Being able to look at code impact by category over six months really streamlines R&D tax credit reporting."
The Outcome
Finance teams get accurate R&D allocation data without engineering scrambles, engineering leaders reclaim hours every quarter, companies maximize legitimate tax credit claims with defensible data, capitalization reporting becomes precise and audit-ready, and quarter-end becomes routine instead of chaotic.