Appearance
How your team's effort is estimated
Nobody on your team fills in a timesheet. Once a day, Flowstate estimates what each person worked on, as a share of their day on each project, from the tools they already use. This page explains where those numbers come from, so you can check them with confidence before you submit.
The short version
Each person's day is built up in layers. When two layers disagree about the same day, the later one wins.
- Ticket activity from Jira, Linear or Azure DevOps.
- GitHub activity, if GitHub is connected.
- AI analysis, if your organisation has switched it on.
- Your changes as team lead, and your submission.
- Leave.
1. Ticket activity
Everything starts from the tickets in your project tool.
- A ticket counts towards someone's day from the moment it moves into an in-progress status until it moves on.
- A ticket left in progress stops counting once nobody has touched it for a while, and no ticket keeps counting indefinitely. A forgotten ticket can't keep collecting effort. To see when a ticket stops counting, open its row's ticket list (Show tickets).
- Each day is shared across the projects the person had work in progress on. It never adds up to more than 100%.
- A day with no ticket activity stays empty. Flowstate never fills it with a guess.
- Work on a ticket whose project isn't linked shows as Unattributed, for you to sort out.
Rows built this way are tagged Ticket activity. Which statuses and ticket types count is set when your project tool is connected — see Effort reporting settings.
2. GitHub activity
When GitHub is connected, code sharpens the estimate. On any day someone commits code or opens a pull request, that code activity shapes how their day is shared across projects. Ticket activity fills the days without code.
- Only code linked to a ticket counts.
- Only people whose GitHub account is matched to them in Flowstate are affected.
- Rows built this way are tagged GitHub activity, or Tickets + GitHub when both played a part.
Your admin can turn this off, or use each commit's own time instead. See Choose how GitHub shapes effort.
Commit-timestamp attribution changes your CapEx and R&D figures
It changes effort, and so the cost that lands on each project. Turn it on deliberately, with your Finance lead.
3. AI suggestions (optional)
AI Effort Analysis is off unless an admin turns it on in Effort reporting settings. When it's on, after the daily estimate Flowstate reviews each person's tickets and can:
- move part of a day to a different project, where the work belongs there;
- suggest a link for an unlinked project in your tool that belongs to a Flowstate project. The suggestion appears in the Linked Project column on your tool's tab under Delivery → Roadmap, where you can Accept or Dismiss it.
The week doesn't mark which numbers AI changed, and nothing about AI is shown per person. Check the week as usual before you submit.
4. Your changes and submission
The estimate is a draft until you submit the week. Before then you can:
- change any cell — the row is tagged Adjusted, and Flowstate won't overwrite what you entered;
- attribute an unattributed ticket to a project for the week, or remove a ticket from the week;
- put time that belongs to no project in Non-project work — Flowstate never fills this row itself;
- copy a person's submitted week from last week.
Once you submit, the daily estimate no longer changes the week. You can withdraw a submission for a short time afterwards, and very old weeks are locked. See Approve or send back a week.
To see where a row's numbers came from, look at its tag — Ticket activity, GitHub activity, Tickets + GitHub or Adjusted — and hover over it for the rule. Select the arrow at the start of the row (Show tickets) to see the tickets, pull requests and commits behind it. See Check where the numbers came from.
5. Leave
A day of leave shows as zero on every project, in its own Annual leave row. Leave isn't CapEx or OpEx.
Your organisation can choose to carry leave forward, so finance reports put a leave day against the projects the person was working on just before. The week still shows the leave. See Handle leave in effort reports.
Which numbers finance uses
The menu at the top right of the week switches between two versions of the same weeks.
| Version | What it includes | Used for |
|---|---|---|
| Team lead's version | Only weeks that have been submitted. | R&D tax claims, the Spend timeline, and Capitalisation when it's set to use submitted effort. |
| Flowstate estimate | Submitted weeks, plus Flowstate's estimate for weeks nobody has submitted yet. | Capitalisation when it's set to use the estimate, and initiative actuals. |
Until you submit, your team's work doesn't appear in R&D claims or the Spend timeline. Activity that arrives after a week was submitted doesn't count until the week is sent back or withdrawn, and submitted again.
The week itself only shows CapEx, OpEx, Annual leave and Unattributed. Finance may also ask about two gaps that appear in Capitalisation, not in effort reporting:
- Uncosted effort — effort by people with no salary on record. It counts as zero cost until a salary is added.
- Untracked — Capitalisation uses this word for two different things. On its Overview, it's the cost of effort by people who aren't in any team. In its table of projects, Untracked Cost is the part of people's days that no effort covers.
How effort becomes cost is in How costs are calculated.
If your numbers look wrong
Someone's day is empty. None of their tickets was in progress that day, or their account in your project tool isn't matched to them. Ask them to move tickets as they work, or check Settings → Integrations → User attribution.
A ticket kept counting after the work stopped. It was left in an in-progress status. Move it on in your project tool, or choose Remove from this report on the ticket for the week.
A ticket is on the wrong project. See Correct a ticket on the wrong project.
Someone's day moved to a different project after GitHub was connected. On days with code, code activity shapes the split. Select the row's arrow (Show tickets) to see the code behind it, and correct the week before you submit if it's wrong.