Where does margin leak away, what does a file really cost and where does capacity get stuck? Answers in euros, based on your own measured data.
Every analysis explains what's happening, what it costs and what it yields if you tackle it.
Distribution of all entries
Written less than measured, per day of delay
Screenshot of the live report, with example data
Live, in development and planned. Scroll sideways for more.
Minutes per entry, rolling 12 months
Entries cluster on round blocks of 30 and 60 minutes, far above what is natural.
Per employee, rolling 12 months
Anyone who writes exactly 8 hours every day is registering a norm instead of the work.
Measured hours outside office hours
In March work outside office hours rises to 444 hours: an early signal of workload.
Share of measured time
How much time goes to production, meetings and client contact, per employee side by side.
Early warning based on measured patterns
Two employees are above the alert threshold, while it does not show in their hours yet.
Share of all input
Over a third of all hours are entered on Friday: reconstruction after the fact.
Share with full description and tag
The later it is written, the worse the description: from 91% to 34% complete.
Thinking time and conversations per month
In nine months AI thinking time grew from 22 minutes to over 4.5 hours a month.
How heavily the AI assistant is used
Advanced models: 8% of conversations, but 46% of thinking time.
Share of employees using the AI assistant
Within nine months the whole team uses the AI assistant.
What the AI assistant is used for
Annual-account notes and client communication take the most thinking.
With leakage percentage
Every month 8 to 13% of measured time disappears before it reaches the invoice.
Hours per week against available capacity
Around weeks 11 to 15 the workload is structurally above available capacity.
Obligations, with percentage on time
In February the share on time drops below 81%, right in the tax-return peak.
Hours per annual account, with index
Annual accounts in construction take almost 2.5x as many hours as in services.
Average number of loops per file
Compilation is above the norm all year; in the peak it rises to 2.8 loops.
Internal meetings and client appointments, with share of time worked
In the peak months over 11% of time worked goes to meetings.
Advisory hours and the potential fee in thousands of euros
Clients who demonstrably take advice without paying for it: there is demand, but no offer yet.
Example data from the product designs
Your measured figures next to an anonymous peer group of similar firms. An own norm says little; a peer median shows whether 28.5 hours for an annual account is a lot or a little.
Number of firms per band, small business (hours)
Your firm sits at P71: 28.5 hours per annual account against a peer median of 20.4 hours.
Your firm, peer median and top quartile
Compared anonymously with firms of the same size: where does work cost more than at peers?
Percentile within the peer group per quarter
The trend matters more than the position: hours and margin improve, but leakage pulls away from the peer group.
Your firm, peer median and top quartile
| Indicator | Yours | Peer | Top 25% |
|---|---|---|---|
| Leakage | 10.8% | 7.4% | 4.1% |
| Latency work → writing | 1.8 d | 1.1 d | 0.4 d |
| Review loops per file | 2.3 | 1.8 | 1.3 |
| Annual accounts lead time | 46 d | 34 d | 22 d |
| Completeness score | 81% | 87% | 94% |
| Margin per client | 12.3% | 18.1% | 26.4% |
This shows whether a problem is industry-wide or specific to your firm. Only the second can be solved with your own measures.
Shared anonymously and aggregated only · example data from the product designs
Insights is built to improve a firm, not to hold people to account. The employee stays in control of their own activity.
Which data sources count is arranged per employee. No consent, no measurement, and that is visible in the report.
Firm questions are answered at firm level. Where an analysis looks at people, it's about workload and bottlenecks, not a ranking.
The outcomes land in one report for management, with the business case attached. You can export it, so you have the conversation with numbers.
