Kultrix research
40% of Our Project Hours Go to Development, 32% to Testing and Fixing
Where development hours go: an analysis of 25,199 logged hours across 39 Kultrix projects, March 2025 to October 2026. By Oleksandr Padura, Kultrix.
Published · Updated · Version 1.0 · Download CSV
Key findings
- Development is the largest share: 40.0% of hours (39 projects, 4,345 records, 25,199 hours).
- Testing and bug fixing together take 32.0% of hours: QA 7.9% plus bug fixing 24.1% (39 projects). Bug fixing is an upper estimate, see Limitations.
- Fixed-price projects spent more on QA and bugs: 39.9% of hours (14 projects, 4,894 hours) against 30.0% in hourly-billed projects (25 projects, 20,305 hours). The gap depends on how an entry was labeled, so treat it as a direction, not a precise number.
- Meetings and management took 5.6% of hours in fixed-price projects (14 projects) and 11.5% in hourly-billed ones (25 projects).
- 13.2% of hours (3,327 hours, 468 records) could not be classified. We show them as "other" instead of redistributing them.
Methodology
Source and period
The data comes from the internal time-tracking system Kultrix uses on its own client work. We exported it read-only on 8 October 2026. The records cover 3 March 2025 to 7 October 2026.
Sample
- 39 projects: 14 fixed-price and 25 hourly-billed.
- 4,345 time records, 25,199 hours in total.
- Nothing was excluded for volume. Project names, client names, people, rates, budgets and task descriptions are not published.
- Groups with fewer than 5 projects are never shown.
This is the data of one company and its clients. It is not a market sample.
How each record was classified
Every record gets exactly one stage. The rules are keyword based and explicit. We read the task name first (available for 2,003 of 4,345 records, in 14 projects). If it gives no match, we read the description. The first match in this priority order wins:
- Meetings: meeting, standup, sync, call, planning, management, communication, discussion
- Bug fixing: bug, fix, hotfix, issue, defect, crash, troubleshoot
- QA: qa, test, testing, e2e, regression, playwright, cypress, verification
- Design: design, ui, ux, wireframe, figma, mockup, prototype
- Development: develop, implement, backend, frontend, api, integration, refactor, feature, database, deploy, migration and similar
- Other: nothing matched (empty descriptions, entries such as "Week 14-18")
By label source: 1,447 records were labeled by task name, 2,430 by description, and 468 (10.8% of records, 13.2% of hours) fell into "other".
All shares are weighted by hours across the whole sample, not averaged per project. Larger projects weigh more.
Results: share of hours by stage
All projects
| Stage | Share of hours |
|---|---|
| Development | 40.0% |
| Bug fixing | 24.1% |
| Other (not classified) | 13.2% |
| Meetings and management | 10.4% |
| QA | 7.9% |
| Design | 4.4% |
Sample: 39 projects, 4,345 records, 25,199 hours.
Fixed-price vs hourly-billed
| Stage | Fixed (14 projects, 4,894 h) | Hourly (25 projects, 20,305 h) |
|---|---|---|
| Design | 1.3% | 5.2% |
| Development | 36.4% | 40.9% |
| QA | 7.9% | 7.9% |
| Bug fixing | 32.0% | 22.1% |
| Meetings | 5.6% | 11.5% |
| Other | 16.7% | 12.4% |
Testing and fixing together: 39.9% of hours in fixed-price projects (7.9% + 32.0%) and 30.0% in hourly-billed ones (7.9% + 22.1%).
Design looks small because it is rarely logged as a separate entry. Do not read it as the real effort spent on design.
- Design
- Development
- QA
- Bug fixing
- Meetings and management
- Other (not classified)
How much the label source changes the answer
The result depends on where the label comes from. This is the main caveat of the study.
| Stage | By task name only (14 projects, 6,923 h) | By description only (14,949 h) |
|---|---|---|
| Design | 10.7% | 2.5% |
| Development | 54.2% | 42.4% |
| QA | 15.8% | 6.0% |
| Bug fixing | 17.3% | 32.6% |
| Meetings | 2.1% | 16.5% |
The headline numbers above blend both sources. If you cite one stage, cite it with this range.
Solid bar: label from the task name. Outlined bar: label from the description.
Limitations
- One company. These are Kultrix projects, not a sample of the market. Other teams will differ.
- Bug fixing is an upper estimate. Bug keywords take priority over development keywords, so an entry such as "fixed login screen" during feature work counts as a bug.
- Label source matters. Shares differ up to 2-3 times depending on whether we use the task name or the description (see the table above).
- Design is under-counted. It is rarely logged on its own.
- 13.2% of hours are unclassified and shown as "other".
- Hours are weighted. Large projects dominate the shares.
- Keyword rules, not manual review. A record can be misclassified; the rules are listed in full so you can judge them.
Download the data
CSV with the aggregates shown on this page: dataset_public.csv (3 rows: all, fixed, hourly; columns: group, projects_n, records_n, hours_total and the share of each stage). Version 1.0.
How to cite
Padura, O. (2026). Where Development Hours Go: 25,199 Hours Across 39 Projects. Kultrix. Version 1.0, published 12 October 2026. https://kultrix.com/research/where-development-hours-go
Short form: "Kultrix time-tracking data, 39 projects, 25,199 hours (March 2025 to October 2026), kultrix.com/research/where-development-hours-go".
You may quote the figures with a link to this page. If you reuse the charts, keep the caption and source line.
About the data owner
Kultrix is a software development agency in Lviv, Ukraine. We build MVPs, mobile apps and web products.