# AI Productivity and Time-Savings Benchmarks

> Compare experimental AI results for writing, consulting and coding, with model-era limits, measured versus perceived outcomes and reproducible time math.

Canonical: https://workspace369.com/research/ai-productivity-time-savings-benchmarks/
Last updated: 2026-09-12

[← Workspace369 Research](https://workspace369.com/research/)

Compare experimental AI results for writing, consulting and coding, with model-era limits, measured versus perceived outcomes and reproducible time-index calculations.

[Research and calculations by Workspace369](https://workspace369.com/about/)Published September 12, 2026Updated September 12, 2026

Measured population

Selected professional-writing, consulting and developer experiments; no pooled workforce estimate

Observation period

2022–2025 experiments; institutional updates checked through September 2026

Sources checked

September 12, 2026

[Read the methodology ↗](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#methodology)

[Download data · CSV](https://workspace369.com/downloads/research/ai-productivity-time-savings-benchmarks.csv)[Source ledger · JSON](https://workspace369.com/downloads/research/ai-productivity-time-savings-benchmarks.json)[Cite this report ↗](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#cite)

2026 edition

## Key takeaways

Each takeaway links to its canonical data row. Reported figures belong to their cited producer; modeled results are labeled as calculations.

1. [Writing experiment: lower task time with AI assistance: 40% (2023 experiment; published July 2023).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#writing-time)MIT / Noy and Zhang
2. [Writing experiment: higher evaluated output quality: 18% (2023 experiment; published July 2023).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#writing-quality)MIT / Noy and Zhang
3. [Consulting experiment: increase in subtasks completed: 12.2% (Study first released in 2023; published-study summary April 2026).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#consulting-completion)Harvard Business School AI Institute / Dell’Acqua and colleagues
4. [Consulting experiment: approximately higher quality on the innovation exercise: 32% (Study first released in 2023; published-study summary April 2026).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#consulting-quality)Harvard Business School AI Institute / Dell’Acqua and colleagues
5. [Strategy task: correct answers without AI: 84.5% (Study first released in 2023; published-study summary April 2026).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#strategy-control)Harvard Business School AI Institute / Dell’Acqua and colleagues
6. [Strategy task: correct answers with GPT-4 access: 70.6% (Study first released in 2023; published-study summary April 2026).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#strategy-ai)Harvard Business School AI Institute / Dell’Acqua and colleagues
7. [Strategy task: correct answers with GPT-4 and prompt overview: 60% (Study first released in 2023; published-study summary April 2026).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#strategy-prompt)Harvard Business School AI Institute / Dell’Acqua and colleagues
8. [Copilot experiment: reduced completion time on one coding assignment: 55.8% (May 15–June 20, 2022).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#copilot-time)Peng, Kalliamvakou, Cihon and Demirer
9. [METR early-2025 experiment: longer completion time with AI allowed: 19% (February–June 2025).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#metr-time)METR
10. [METR participants: perceived AI speedup after the experiment: 20% (February–June 2025).](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#metr-perceived)METR

Try it with one real client

Set up one client, one project and one invoice in Workspace369, then check what your client sees in the portal.

[Compare plans](https://workspace369.com/pricing/)

14-day free trial. Card required. After the trial: from $29/month. Renews automatically every month until you cancel. Cancel anytime in Settings → Billing.

![A researcher stands between matching task benches and clocks, with an AI prism nearby, illustrating comparison of work completion.](https://workspace369.com/images/research/ai-work-task-comparison.webp)

AI-generated editorial illustration; not a data visualization.

## Does AI help with short professional writing tasks?

The writing experiment found less time spent and higher evaluated quality with assistance. Its assignments did not require the full factual precision or organizational context of many real customer-facing documents.

**AI productivity benchmarks — 2022–2025 experiments; institutional updates checked through September 2026. Study-specific recruitment; not a representative global worker sample.**

| Outcome versus control | Reported change |
| --- | --- |
| Reduction in task time | [40%](https://news.mit.edu/2023/study-finds-chatgpt-boosts-worker-productivity-writing-0714)Randomized experiment |
| Increase in evaluated quality | [18%](https://news.mit.edu/2023/study-finds-chatgpt-boosts-worker-productivity-writing-0714)Randomized experiment |

Quality-score improvement and time reduction are different outcomes and must not be added.

Primary source: [MIT / Noy and Zhang — July 14, 2023](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-mit-writing)

## Can AI improve one consulting task and harm another?

Yes. The study separates tasks that the model could handle from a strategy problem on which it was unreliable. Productivity on the first set does not establish accuracy on the second.

**AI productivity benchmarks — 2022–2025 experiments; institutional updates checked through September 2026. Study-specific recruitment; not a representative global worker sample.**

| Outcome and condition | Reported percentage |
| --- | --- |
| More subtasks completed on the innovation exercise | [12.2%](https://aiinstitute.hbs.edu/back-to-the-beginnings-of-ai-at-work/)Randomized experiment |
| Higher quality on the innovation exercise, approximately | [32%](https://aiinstitute.hbs.edu/back-to-the-beginnings-of-ai-at-work/)Randomized experiment |
| Correct strategy answer: no AI | [84.5%](https://aiinstitute.hbs.edu/back-to-the-beginnings-of-ai-at-work/)Randomized experiment |
| Correct strategy answer: GPT-4 access | [70.6%](https://aiinstitute.hbs.edu/back-to-the-beginnings-of-ai-at-work/)Randomized experiment |
| Correct strategy answer: GPT-4 plus prompt overview | [60%](https://aiinstitute.hbs.edu/back-to-the-beginnings-of-ai-at-work/)Randomized experiment |

The quality estimate is approximate. This is historical GPT-4 evidence, not a benchmark of today’s models.

Primary source: [Harvard Business School AI Institute / Dell’Acqua and colleagues — April 9, 2026](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-hbs)

## Why do coding studies report different productivity effects?

A standalone coding assignment and maintenance inside a mature repository are different tasks. The studies also used different model generations and recruitment methods. Their results should remain separate rather than averaged.

**AI productivity benchmarks — 2022–2025 experiments; institutional updates checked through September 2026. Study-specific recruitment; not a representative global worker sample.**

| Study and outcome | Reported change |
| --- | --- |
| Copilot assignment: reduction in completion time | [55.8%](https://arxiv.org/pdf/2302.06590)Randomized experiment |
| METR: increase in actual completion time | [19%](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/)Randomized experiment |
| METR: anticipated speedup before the work | [24%](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/)Participant perception |
| METR: perceived speedup after the work | [20%](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/)Participant perception |

METR’s later follow-up reported serious selection and time-measurement problems. Its authors do not treat the follow-up as a reliable estimate of the current effect; the historical slowdown is not a claim about all developers in 2026.

Primary sources: [Peng, Kalliamvakou, Cihon and Demirer — February 13, 2023](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-copilot) · [METR — July 10, 2025](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-metr) · [METR — February 24, 2026](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-metr-update)

## What does each measured time change mean on the same baseline?

A baseline index makes the arithmetic easier to read without pooling the studies. Each control condition is assigned the same starting index; the resulting assisted index still refers only to its own experiment.

**AI productivity benchmarks — 2022–2025 experiments; institutional updates checked through September 2026. Study-specific recruitment; not a representative global worker sample.**

| Experiment | Assisted time index; control = 100 |
| --- | --- |
| Writing experiment | [60](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#methodology)Normalized calculation |
| Copilot assignment | [44.2](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#methodology)Normalized calculation |
| METR repository tasks | [119](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#methodology)Normalized calculation |

Workspace369 normalization, not a new observed result. Lower means less time. A reduction in time is not the same percentage increase in throughput, and no throughput estimate is claimed.

Primary sources: [MIT / Noy and Zhang — July 14, 2023](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-mit-writing) · [Peng, Kalliamvakou, Cihon and Demirer — February 13, 2023](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-copilot) · [METR — July 10, 2025](https://workspace369.com/research/ai-productivity-time-savings-benchmarks/#source-metr)

## How should a business test AI time savings?

Define a repeatable task and an acceptance standard before comparing assisted and unassisted work. Include prompt preparation, fact checking, review and rework. Keep customer-data access and human approval requirements unchanged across the comparison.

Track both time and quality. Do not apply a writing-task percentage to an entire payroll or assume the same result for an autonomous agent.

## Methodology and limitations

1. This is a selected evidence comparison, not a systematic review or meta-analysis. Studies were included for identifiable tasks, original-producer evidence and clearly stated outcome definitions.
2. The writing and consulting figures use their research institutions’ public summaries of published studies. The coding paper and METR report provide original experimental detail. Working-paper and published versions are not mixed.
3. Measured outcomes and perceived speedups carry separate evidence labels. The time-index calculation uses only measured completion-time changes; survey beliefs are excluded from it.
4. Model versions and study dates remain attached to the findings. We checked the METR follow-up and retained its warning about the reliability of newer estimates.

### What these numbers cannot tell you

- Selected experiments cannot establish a universal AI return on investment. Different tasks, control conditions, participants and quality tests prevent a defensible pooled percentage.
- The sample sizes describe study scope, not independent replications. Vendor involvement in the Copilot study is relevant to interpretation.
- Historical results do not establish the capabilities of models available today. The normalization does not include implementation costs or demonstrate sustained annual savings.

### Freshness and corrections

Maintained by the Workspace369 editorial team. Review quarterly and when a cited producer releases a replacement study. Next editorial review: December 2026. Retain historical model versions and observation dates. The edition date changes only when the evidence or content is substantively reviewed; it does not change the underlying observation period.

First edition: September 12, 2026. Data extraction, source attribution and arithmetic checked for this edition. No independent peer review is claimed.

Found an error or a newer primary release? [Send a correction with the source and affected statistic](mailto:support@workspace369.com?subject=Research%20correction%3A%20AI%20Productivity%20and%20Time-Savings%20Benchmarks). Confirmed corrections are recorded below before republishing.

### Revision history

Evidence dates above describe the report. Download and citation updates below do not imply that sources were checked again.

- October 10, 2026 — Versioned CSV and JSON archives, checksums and reference-manager exports added. Source observations and evidence dates are unchanged.

## Primary sources and provenance

Every reported numeric cell links directly to its producer. The downloads include exact table or workbook locators, observation periods, access dates, formulas and input references.

1. **MIT / Noy and Zhang**[MIT research announcement for the published Science writing experiment ↗](https://news.mit.edu/2023/study-finds-chatgpt-boosts-worker-productivity-writing-0714)Published July 14, 2023. Accessed September 12, 2026.Selected factual observations paraphrased with attribution. No source report, chart, participant data or proprietary database is redistributed; source rights remain with its producer.
2. **Harvard Business School AI Institute / Dell’Acqua and colleagues**[Back to the Beginnings of AI at Work: institutional review of the published consulting experiment ↗](https://aiinstitute.hbs.edu/back-to-the-beginnings-of-ai-at-work/)Published April 9, 2026. Accessed September 12, 2026.Selected factual observations paraphrased with attribution. No source report, chart, participant data or proprietary database is redistributed; source rights remain with its producer.
3. **Peng, Kalliamvakou, Cihon and Demirer**[The Impact of AI on Developer Productivity: Evidence from GitHub Copilot, arXiv v1 ↗](https://arxiv.org/pdf/2302.06590)Published February 13, 2023. Accessed September 12, 2026.Selected factual observations paraphrased with attribution. No source report, chart, participant data or proprietary database is redistributed; source rights remain with its producer.
4. **METR**[Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity ↗](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/)Published July 10, 2025. Accessed September 12, 2026.Selected factual observations paraphrased with attribution. No source report, chart, participant data or proprietary database is redistributed; source rights remain with its producer.
5. **METR**[We are Changing our Developer Productivity Experiment Design ↗](https://metr.org/blog/2026-02-24-uplift-update/)Published February 24, 2026. Accessed September 12, 2026.Selected factual observations paraphrased with attribution. No source report, chart, participant data or proprietary database is redistributed; source rights remain with its producer.

## How to cite this report

For a source-reported statistic, credit the original publisher and link to the exact row here when using our compilation. For a modeled result, cite Workspace369 and include the assumptions. Linking to this page does not make us the original producer of third-party data.

> Workspace369. (2026-09-12). AI Productivity and Time-Savings Benchmarks. https://workspace369.com/research/ai-productivity-time-savings-benchmarks/. Primary sources and observation periods as listed in the report.

Clipboard unavailable. Select and copy the citation text above.

[Download the citation-ready data (CSV) ↓](https://workspace369.com/downloads/research/ai-productivity-time-savings-benchmarks.csv)[Download the complete source ledger (JSON) ↓](https://workspace369.com/downloads/research/ai-productivity-time-savings-benchmarks.json)

[Reference manager · BibTeX ↓](https://workspace369.com/downloads/research/ai-productivity-time-savings-benchmarks.bib)[Reference manager · RIS ↓](https://workspace369.com/downloads/research/ai-productivity-time-savings-benchmarks.ris)

### Versioned data

Use an archived file when your analysis needs to stay reproducible. The version identifies the exact CSV and JSON bytes; unchanged data keeps the same version.

Dataset version: `1a57d89d2ab601f957c7e71b8080eeea509994753902e2f79c23a2bf2055caed`

[Archived data · CSV ↓](https://workspace369.com/downloads/research/versions/ai-productivity-time-savings-benchmarks/1a57d89d2ab601f957c7e71b8080eeea509994753902e2f79c23a2bf2055caed/ai-productivity-time-savings-benchmarks.csv)[Archived data · JSON ↓](https://workspace369.com/downloads/research/versions/ai-productivity-time-savings-benchmarks/1a57d89d2ab601f957c7e71b8080eeea509994753902e2f79c23a2bf2055caed/ai-productivity-time-savings-benchmarks.json)[Download checksums and version manifest ↓](https://workspace369.com/downloads/research/versions/ai-productivity-time-savings-benchmarks/1a57d89d2ab601f957c7e71b8080eeea509994753902e2f79c23a2bf2055caed/manifest.json)

Reference exports describe Workspace369’s compilation. When citing a reported observation, also credit its original producer.

[The downloads are English-language reference datasets, including on translated pages. Source rights remain with their producers. Attribute Workspace369’s compilation and calculations; consult each source’s reuse terms.](https://workspace369.com/downloads/research/license-v1)

## Continue the research

[Invoice payment times](https://workspace369.com/research/b2b-invoice-payment-statistics/)[Business AI adoption](https://workspace369.com/research/small-business-ai-adoption-statistics/)[Software stack costs](https://workspace369.com/research/client-work-software-cost-index/)[Professional-services labor costs](https://workspace369.com/research/professional-services-labor-cost-statistics/)[Business survival rates](https://workspace369.com/research/small-business-survival-rate-statistics/)[Small-business cash flow](https://workspace369.com/research/small-business-cash-flow-statistics/)[Card and ACH processing fees](https://workspace369.com/research/payment-processing-cost-benchmarks/)[Nonemployer business revenue](https://workspace369.com/research/solo-business-revenue-benchmarks/)[Sole-proprietor profit and expenses](https://workspace369.com/research/sole-proprietor-profit-margin-benchmarks/)[Employee turnover by industry](https://workspace369.com/research/employee-turnover-statistics-by-industry/)[Sales productivity and admin time](https://workspace369.com/research/sales-productivity-crm-admin-statistics/)[CRM data quality and AI readiness](https://workspace369.com/research/crm-data-quality-ai-readiness-statistics/)[AI customer-service performance](https://workspace369.com/research/ai-customer-service-performance-statistics/)[Workflow automation by function](https://workspace369.com/research/workflow-automation-statistics-by-business-function/)[Cloud software adoption](https://workspace369.com/research/cloud-software-adoption-statistics/)[CRM adoption](https://workspace369.com/research/crm-adoption-statistics/)[Employee absenteeism](https://workspace369.com/research/employee-absenteeism-statistics/)[Paid leave and benefits](https://workspace369.com/research/paid-leave-employee-benefits-benchmarks/)[Remote and hybrid work](https://workspace369.com/research/remote-hybrid-work-statistics/)[Freelancers and contractors](https://workspace369.com/research/freelancer-independent-contractor-statistics/)[Business applications and formations](https://workspace369.com/research/new-business-formation-statistics/)[Business email compromise](https://workspace369.com/research/business-email-compromise-statistics/)[Project success and failure](https://workspace369.com/research/project-management-success-failure-statistics/)[Revenue per employee](https://workspace369.com/research/revenue-per-employee-benchmarks/)[Agency pricing benchmarks](https://workspace369.com/research/agency-pricing-benchmarks/)[Freelancer rate benchmarks](https://workspace369.com/research/freelancer-rate-benchmarks/)[Web design pricing benchmarks](https://workspace369.com/research/web-design-pricing-benchmarks/)[Agency profit margin benchmarks](https://workspace369.com/research/agency-profit-margin-benchmarks/)[Proposal close rate benchmarks](https://workspace369.com/research/proposal-close-rate-benchmarks/)[Freelancer late payment](https://workspace369.com/research/freelancer-late-payment-statistics/)[Freelance writing rates](https://workspace369.com/research/freelance-writing-rates/)[Software price tracker](https://workspace369.com/research/client-work-software-price-tracker/)

### From the data to day-to-day operations

To see how this fits day-to-day operations, see how Workspace369 runs client work in one place. [One workspace for client work](https://workspace369.com/solutions/client-work-operating-system/)

[Workspace AI assistant](https://workspace369.com/features/ai-chat/)[AI customer-service evidence](https://workspace369.com/research/ai-customer-service-performance-statistics/)[Human-reviewed workflow automation](https://workspace369.com/features/automation/)[One workspace for client work](https://workspace369.com/solutions/client-work-operating-system/)[Workspace369 plans and pricing](https://workspace369.com/pricing/)