Microsoft Copilot Cowork

Cowork Team Dashboard

Contoso — Customer Success Impact & how Cowork is used

Generated 2026-09-18 ·

Use the information in this report to gauge the impact of Cowork on your team, not as individual or team performance scores. Treat these as directional estimates of tool-assisted time savings, and read them with team context in mind (e.g., project phase, seasonality). Anonymized & team-level only: nothing is shown per person.

New to this report? The tab explains every number, tab and control — and every section title has a clickable ? for a quick explanation.

Use the Period selector, Hourly rate box, and Recapture rate box below to pick the reporting window, the hourly rate (default $75/hr), and the productivity recapture rate (default 70%). Value / cost-reduction figures = effective recaptured hours × hourly rate and recompute instantly; time-saved hours and counts stay the same.

$/hr
%

What the data saysA plain-language reading of the team's posts, generated automatically. It re-words itself when you change the hourly rate below.

The four highlights below summarize the team's Cowork impact at a glance: the total time reclaimed and its dollar value, the task category driving the most savings, the business process where Cowork is applied most, and the type of business value it advances most — so you can quickly see where the impact is concentrated.

Team impact at a glanceThe headline totals for the selected period. Value = manual hours × the hourly rate in the control bar, so it recomputes whenever you change the rate.

About the ranges: under Time saved and Value / cost reduction the headline is the typical (mid-point) estimate; the low–high range beside it is the conservative-to-optimistic span from the research time bands (each task category carries a low / typical / high band — see ).

Where Cowork is applied — top business processesThe business processes the team uses Cowork for most, ranked by the selected measure (time saved or value). This is a preview — the full How Cowork is used tab expands every process into its deliverables and the skills behind them.

The business processes where Cowork does the most work for the team — the clearest signal of how it's actually being used. Open the full breakdown to drill into each one.

Where the time went — by task categoryThe method used, task by task. Each task is sorted by its output file type and goal keywords (e.g., spreadsheets → Analysis; code / HTML → Write or debug code). Each category carries a research time band (minutes saved per run); time saved = run tasks × band. The reach line shows how many contributors used it. Full mapping is on the How to read tab.

Understand how much time is saved by the team across each task category. Refer to the section to understand the research-based time ranges for each category.

Each task category row shows how many contributors used it — e.g., “used by 4 of 5 contributors” — so you can see where usage is concentrated vs. spread, not just the volume of hours. This is an aggregate count and never names anyone. To protect a small team, when fewer than 3 people used a category the exact number is withheld and shown as “used by <3 contributors”.

Roles Cowork stood in for

Roles a services firm would have billed — expand for the specific skills behind them.

Outputs produced — by format

Every output the team produced with Cowork, by file type. Note that outputs count each file and version, so this may be higher than the “Deliverables” count on the tab, which counts distinct deliverables. Per-item detail sits under “” on the tab.

Work by business processWhat the team actually does with Cowork, grouped into the shared canonical process set. Click any row to expand its deliverables and the skills behind them. Deliverables with a de-identified name list individually; ones a post carried only by file type collapse into a single row (e.g., “HTML · 5 deliverables”).

The business processes the team used Cowork for, ranked by time saved. Click any process to expand it and see the deliverables it produced and the skills behind them.

Named deliverables (e.g., “Team ROI dashboard”) list on their own row. A row that shows only a format (e.g., “HTML”) is a deliverable whose name wasn't included in that teammate's post. All type-only deliverables of one format collapse into a single row — e.g., “HTML · 5 deliverables” — with their hours and value summed.

Cowork fit — how well the work suited CoworkEvery task is graded High / Medium / Low on how well it fit Cowork's agentic, cross-app strengths. High = work only Cowork can do (builds & packaged skills, executed automations/connectors, many-source synthesis); Low = a single-app Copilot could have done it. Click a fit level to see the tasks in it — de-identified to business process & method, never a person or file.

A composition of graded hours by how well the work fit Cowork, shown as a waterfall — the bands add up to all graded time. Click High, Medium or Low below to expand the tasks in that band.

Category mixHow each contributor group splits its time across categories. A role only breaks out when 3+ people share it; otherwise everyone is combined into one bar — no individual is ever shown.

How saved time splits across task categories — grouped by Role where privacy allows.

Glossary of terms

Active days — Person-days with at least one Cowork task in the window.

Anonymity — A role or attribute is shown separately only when at least 3 contributors share it; otherwise contributors are combined. Nothing is ever shown per person.

Business process — The business need served by the work — e.g., Business Value & ROI Analytics.

Contributors — The number of teammates who posted their de-identified stats this period. Never named.

Cowork fit — How well a task suited Cowork's agentic, cross-app strengths, graded High / Medium / Low. High = work only Cowork can do (builds & packaged skills, executed automations, many-source synthesis); Low = a single-app Copilot could have done it. Shown as a quantified waterfall on the How Cowork is used tab.

Deliverables — The count of distinct pieces of work produced.

Hands-on time — The actual time the team spent working with Cowork.

Outputs — Every output file and version created. May be higher than deliverables because it counts each file and version.

Reach — How many contributors used a given task category. Withheld as “<3” when fewer than 3 share it.

Research time band — The low / typical / high minutes of manual time saved per run for a task category, drawn from published studies (see “” below).

Run tasks — A single unit of work run with Cowork — one discrete request or action (e.g., analyzing a file, drafting a document). Run tasks are grouped into sessions.

Sessions — Distinct Cowork chats run across the team. A session may contain one or multiple run tasks.

Skills — The specific capabilities behind roles (e.g., Data Visualization, Python).

Task category — How the work was done (the method) — e.g., Analysis & Research, Write or debug code. Each carries a research time band (see “” below).

Team speed multiplier — How much faster the work went: estimated hours without Cowork ÷ actual hands-on with Cowork hours.

Time saved — Manual hours Cowork saved this period: for each task, run tasks × the research time band for its category. The headline is the typical estimate, with a low–high range alongside.

Recapture rate — The share of time saved the team can realistically harvest into productive output. Set in the control bar (default 70%). It scales every value figure but leaves time-saved hours and counts unchanged.

Effective time recaptured — Time saved × the recapture rate — the productive hours the team actually reclaims. This is what the value figure prices.

Value / cost reduction — Effective time recaptured priced out: time saved × recapture rate × the hourly rate in the control bar. Recomputes whenever you change the rate or the recapture rate.

How task categories are derived

Each task is sorted into up to two categories from three signals:

1 · Output file type

  • spreadsheets (.xlsx / .csv / .json) → Analysis & Research
  • code & web (.py / .html / .sql / .js) → Write or debug code
  • documents (.docx / .pdf / .pptx / images) → Document & content creation
  • packaged skills (.zip / .skill) → Specialized workflows

2 · Goal keywords

  • “analyze / research / ROI / benchmark” → Analysis
  • “debug / refactor / script / API / automation” → Write or debug code
  • “email / inbox / reply” → Email
  • “meeting / transcript / standup / agenda” → Meeting

3 · Input signal

  • if the sources are data files, or there are 3 or more of them → Analysis

A task with no saved file and no signal falls to General assistance / Other. When more than one matches, they rank (code › analysis › specialized › document › communication › meeting › email › general) and the top two are kept.

Privacy & anonymity

Nothing is shown at an individual level. Members appear only as counts. The one per-attribute view (category mix) breaks a Role out only when 3+ contributors share it; otherwise they collapse into a single combined bar. The only attribute used is the directory Role (job title) that a teammate's post carries — never names, never country, never file names or prompts.

Value model
  • Time saved = Count of run tasks × the research band (minutes saved/run).
  • Recapture rate = the share of time saved the team realistically converts into productive output (default 70%; adjustable in the control bar).
  • Effective time recaptured = time saved × recapture rate.
  • Value / cost reduction = effective recaptured hours × hourly rate (default $75/hr) = time saved × recapture rate × rate.
  • Speed multiplier = manual hours ÷ modeled hands-on hours.

All figures come from the posts; the dashboard only re-totals and re-prices them.

Research bands & sources

Each task category carries a research-anchored time band — the minutes of manual time saved per run, at a low / typical / high estimate.† Time saved = run tasks × the band for that category.

Task categoryLowTypicalHigh
Analysis & Research306792
Write or debug code305696
Document & content creation122442
Meeting workflows123143
Email workflows3712
Communication workflows246
Specialized workflows102540
General assistance / Other258

† Minutes of manual time saved per run (low / typical / high). Sources: Stanford-WB (SSRN 5136877), Microsoft Research 2026 (DiD n=72,186), Noy & Zhang (Science 2023), Cambon et al. (MSR 2024), Cui et al. (CACM 2024), Brynjolfsson, Li & Raymond (QJE 2025), Forrester TEI 2024.