To explain engineering velocity to investors, show what the team delivered over the quarter in the words of the product, how much of it was reviewed and tested, and how many defects were confirmed since, each line traceable to a pull request. A CEO who has the Leadership profile asks Claude for those facts; DeployIt draws them from the commits merged into the company's repositories.
This article is for a CEO preparing a board meeting or an investor update, who does not read code and has an hour, not a week. It gives the questions to ask, a way to turn the answers into one slide, and the limits to state before an investor finds them.
I am preparing the board update. Who delivered what this quarter, with what code score and how many confirmed defects?
Here is the quarter, developer by developer; each score reads with the commits on its row. Defects from the last 60 days are not mature yet, so the last column can still rise.
| Developer | Merged commits | Code score | Confirmed defects |
|---|---|---|---|
| Camille R. | 132 | 3.9 | 2 |
| Idriss B. | 118 | 3.2 | 5 |
| Léa M. | 97 | 4.1 | 1 |
What investors mean by engineering velocity
Velocity is not a number of commits. It is a rhythm of delivery that repeats, tied to something a customer can see. When investors ask about it, they are asking three things:
- How often does the team put something in users' hands?
- How reliably: was the work reviewed, was it tested, what had to be fixed afterwards?
- With what effect on activation, retention, expansion or risk?
A fourth question now comes quickly: how much of the code do AI coding agents write, and who reviews it? In DeployIt's statistics, bots are excluded and detected coding agents are counted separately; an undeclared agent counts as human work. You can give the share of merged work that came from detected agents, and say that it is a floor.
Why burn-up charts and ticket counts do not convince
A burn-up chart, a count of closed tickets, a number of story points: these measure activity, in units the team defines for itself. Points inflate when work is sliced thinner. Closed tickets rise with small chores. Neither says what changed for a customer.
A score quoted alone has the same weakness. The code score DeployIt gives is a model's reading of each commit. It means little without the number of commits behind it, and it answers no investor question by itself.
What survives a second question is a line an investor can check. Before and after (example, not customer data):
| Line on the slide | What an investor can do with it |
|---|---|
| "200 tickets closed this quarter" | Nothing: the unit is yours |
| "Velocity up 18%" | Ask "18% of what?" |
| "Prices shown with VAT in the checkout: 23 commits, pull requests reviewed by a colleague, no defect confirmed since" | Open the pull request, check the date, ask what it changed in conversion |
What to ask Claude before the board meeting
Four requests cover the engineering part of an update. The figures come from facts written once per merged commit and updated every night.
| What you ask | What comes back |
|---|---|
"What changed in the product over the last 90 days?" (recent_changes, or the what_changed prompt) | What landed on the tracked branch, grouped into deliveries: date, type, title, modules, number of commits. No person is named. |
"Give me the team's report for the quarter." (the team_report prompt) | The team's figures and one row per tracked developer: merged commits, pull requests reviewed by a colleague, tests, defects, code score with its number of commits. |
| "How many defects were confirmed this quarter, and how many are still to confirm?" | Defects by status. A defect is the commit that a fix came to repair. |
| "Which modules depend on a single person?" | Who wrote each module over the last 12 months. |
The first request gives the story, the second the reliability, the third the quality, the fourth the dependency on one person that a board will want to know about.
Claude composes the table or the paragraph; DeployIt supplies the facts. No report is sent to you: the update is built when you ask for it. The two prompts bring back the same table from one period to the next, so quarters compare.
You read all of it without code or file paths: for the Leadership profile, a filter removes them (filtered and tested; no filter is infallible). The datasets are listed in the statistics section of the docs, and the monthly version of this exercise is the engineering report for the CEO.
How to explain engineering velocity to investors in one slide
From the answers to the slide
Start from deliveries, not from people
Take the quarter's deliveries from the first request and keep five to ten. One line each.
Rewrite each delivery as a customer sentence
"Asynchronous tax calculation" becomes "the checkout no longer waits for the tax service" (example). If you cannot write the sentence, ask Claude what the change does:
ask_codeanswers from the code, in plain words, with a status.Group by theme
Three to five themes: reliability, performance, activation, expansion, compliance. Investors read an allocation of effort faster than a list.
Add one reliability line
The share of pull requests reviewed by a colleague, the share of changes with tests, the confirmed defects, each next to the previous quarter. Say that the last 60 days can still move.
Quantify the effect only where you have the data
Adoption and revenue come from your product analytics and your finance tools, not from the repository. Where nothing is measured yet, write "delivered, adoption being measured". Invented precision reads as exactly that.
Ask the same way every quarter
Same prompts, same periods. The trend is what an investor reads.
What the repository cannot tell investors
State the limits yourself.
- Deployments and incidents are elsewhere. The statistics count what reaches the tracked branch of each repository, when it arrives. A production date, an outage or a recovery time comes from your delivery pipeline and your monitoring.
- So are usage and revenue. Whether a delivery moved activation or churn is a question for product analytics and finance.
- Recent defects are incomplete. A defect shows up when its fix arrives. Quote confirmed defects only: "to confirm" is the machine's count until an administrator or an investigation has confirmed it.
- A closed quarter can still move. A commit that landed late, a defect dismissed: each summary carries the time it was last computed. Put that date under the figure.
- The code score is an indication. On a slide, quote it with its number of commits or leave it out.
A quiet quarter is not an empty one. A refactor, a migration or a security questionnaire leaves few visible features. Grouped by theme, it reads "reliability and compliance took the largest share of the quarter" (example): an allocation choice, not an absence of news.
Who carried the quarter: what you read, what you show
The statistics are individual: DeployIt tracks each developer, commit by commit. That is why the table at the top exists, and how you can answer when a board member asks who holds billing. The framework is written in the terms of service: the company informs the tracked people before it connects its repositories; a score set by a model is an indication, never on its own the reason for a decision; the human decision prevails over a defect counted by the machine.
An investor update rarely needs names. Show the team's totals and their trend, and keep the table by developer for yourself and your CTO. One individual fact does belong in front of a board: a module that depends on a single person. It is a risk for the company, and a hiring or pairing decision follows from it.
To go further, read developer performance metrics read from merged commits and the bus factor of a codebase: which modules depend on one developer; the product itself is described in DeployIt: ask your code and follow each developer's work from Claude.
What DeployIt does here
DeployIt is an MCP server. Your company gets one link, https://mcp.deployit.ai/<company>/<code>; each person adds it to Claude (on the web, on the desktop or in Claude Code) and signs up with a name and an email, which is not verified; their questions are logged under that name. Everyone starts with the Everyone profile: questions about the product, answered from the code; a filter removes the code, the file paths and the names of team members it recognises from the answers, without proving their absence, and a leak test tells you what still gets through. An administrator gives the Developers profile (excerpts) or the Leadership profile (developer statistics) to whoever needs it. Repositories are connected from GitHub or GitLab; DeployIt makes no commit, branch or pull request, and only adds one webhook per repository.
Next step: open the trial, have your CTO connect the repositories, add your company link to Claude by the three steps of Connecting, and ask for the team's report on the last quarter (statistics need the Leadership profile, which an administrator gives; the demo has none, so try it on your own repositories).
Price. Free plan: 15 questions a week for the whole company. A 30-day trial, no card. Plans from €290 a month excluding tax for up to 25 tracked developers: see pricing.
Ask your code from Claude
30-day trial, no card. Or try the demo first: add https://mcp.deployit.ai/demo to your Claude connectors.
Frequently asked questions
How do you explain engineering velocity to investors?
Show what the team delivered over the quarter in the words of the product, how much of it was reviewed and tested, and how many defects were confirmed since. Each line should lead back to a pull request an investor could open. Put the previous quarter next to each figure: the same request, quarter after quarter, is what makes a rhythm visible.
Is the number of commits a good measure of engineering velocity?
Not on its own. A count of commits says how much landed, not what it changed for a customer or whether it held. Read it with the share of pull requests reviewed by a colleague, the share of changes that came with tests and the defects confirmed afterwards; DeployIt gives these for the team and for each developer, from the commits merged into the tracked branch.
What engineering figures belong in a board update?
A short list: the deliveries of the quarter in plain words, merged commits against the previous quarter, the share reviewed by a colleague, the share tested, and confirmed defects. Add the modules that depend on a single person if the board follows key-person risk. Leave out any figure you cannot trace to its commits.
How do you show engineering progress to a board without technical detail?
Start from deliveries, not from code. With DeployIt, asking Claude what changed over the last 90 days returns what landed on the tracked branch, grouped into deliveries with a date, a type, a title and the modules touched, without naming anyone. Rewrite each delivery as one sentence about what a customer can now do.
Should individual developer statistics be shown to investors?
That is your decision, and an investor update rarely needs names. DeployIt tracks each developer, and the company informs the tracked people before connecting its repositories; a score set by a model is an indication, never on its own the reason for a decision. Show the team totals and their trend, and keep the table by developer for the people who manage the team.
Continue reading
Find the commit that introduced a bug: follow the fix back to it
For a CTO and their developers: how a fix leads back to the commit it repairs, why the defect it reveals stays “to confirm” until an administrator or an investigation confirms it, and why the last 60 days are never final.
Developer performance metrics that hold up: the ones read from merged commits
For a CTO or a CEO choosing what to measure: the per-developer metrics that hold up come from merged commits (delivered, reviewed, tested, defects, code score), and each one has limits to state.
Developer metrics in your data warehouse: one key, two MCP calls
For a data team: how to load per-developer statistics into a warehouse or a BI tool with a Data tool key. The request and its header, stats_schema then stats_query, the row limit, the cursor, and what to reread on each pass, shown as the trace of a call.