Outcomes are the realised, often measurable results of delivered AI work. Where an ICE score describes what your team expected and what they judged once the work was done, an Outcome captures what your business actually realised. A Deliverable that saved your support team 200 hours per week is one Outcome. A change request that won a £500K expansion deal because the customer’s CTO referenced your new AI policy is another. Outcomes are the receipts.Documentation Index
Fetch the complete documentation index at: https://docs.kowalah.com/llms.txt
Use this file to discover all available pages before exploring further.
Outcomes attach to a Deliverable or an Expert Request. They’re how the platform turns “we shipped it” into “and here’s what changed in the business.”
Why Outcomes are separate from scoring
ICE scores and Outcomes do different jobs:| ICE Score | Outcome | |
|---|---|---|
| Captured by | The team delivering the work | Anyone with write access to the parent — including you |
| When | At planning (estimated) and at closeout (final) | At closeout, and any time afterwards as new results materialise |
| What it describes | The team’s account of delivery | The realised business result |
| Locks? | Final score locks shortly after closeout | No lock — outcomes can be added or edited at any time |
What an Outcome contains
Every Outcome captures:| Field | Required? | What it’s for |
|---|---|---|
| Title | Yes | A short, plain-English headline |
| Outcome type | Yes | One of: revenue won, cost reduced, productivity gained, adoption milestone, other |
| Narrative | Yes | The story — what changed, who noticed, why it’s attributable to the work |
| Metric value | Optional | The number — e.g. 2000000 |
| Metric unit | Required if value is set | The unit — e.g. GBP, hours/week, users, % |
| Metric period | Optional | one_off, weekly, monthly, quarterly, or annual |
| Evidence URL | Optional | A link to a supporting source — a board paper, a dashboard, an email |
| File attachments | Optional | Documents that back up the result |
The five outcome types
| Type | Example |
|---|---|
| Revenue won | ”Won a £500K expansion contract — their CTO referenced our AI policy work” |
| Cost reduced | ”Reduced AWS supplier spend by £2M annually” |
| Productivity gained | ”Saved 200 hours/week across the support team” |
| Adoption milestone | ”Hit 65% weekly active adoption against a 60% target” |
| Other | Anything that doesn’t fit the four above — and is still worth recording |
Capturing Outcomes
Outcomes can be captured in two places.When a Deliverable or Expert Request closes, the closeout form includes an Outcomes section alongside the final ICE score. Add as many Outcomes as you have at the time of closing — there’s no minimum and no maximum.
Cost reductions often realise months after delivery. Revenue wins can be later still. The Outcome panel on a Deliverable or Expert Request stays open: you can add a new Outcome whenever fresh evidence lands, edit an existing one as the picture sharpens, or soft-delete one if it turns out the result wasn’t really attributable to the work.
There’s no lock window on Outcomes. Unlike the final score — which locks shortly after closeout to preserve the prediction-vs-result narrative — Outcomes are intentionally living. The dashboard becomes more honest over time.
Who can capture Outcomes
Anyone with write access to the parent Deliverable or Expert Request can capture, edit, or soft-delete Outcomes. That includes:- Kowalah team members working on the parent
- Your organization’s admins and Core Team
- Customer-side stakeholders linked to the parent — the people closest to the realised result, who often see the evidence first
Attribution: one Outcome, one parent
An Outcome belongs to exactly one Deliverable or Expert Request. When the same realised business result is genuinely attributable to multiple pieces of work — say, a cost reduction driven by three concurrent Deliverables — capture it as separate Outcome rows on each parent, rather than fanning a single row across many. Duplicated narrative is preferable to ambiguous attribution. Rollup queries can dedupe headline numbers at the dashboard layer if double-counting bites in practice.Evidence: links and files
Every Outcome can carry two kinds of evidence:- An evidence URL — a link to a board paper, a dashboard, a customer email, an internal report
- File attachments — PDFs, spreadsheets, screenshots that back up the claim
Where Outcomes show up
Once captured, an Outcome appears on:- The parent Deliverable or Expert Request page, in an Outcomes section
- The project rollup, summarised across all Deliverables in the project
- The organization-level ROI view, where Outcomes with metrics are aggregated by type
ICE Scoring
The scoring model that Outcomes back up
Deliverables
Where most Outcomes attach — committed work delivered inside a project
Expert Requests
The other parent for Outcomes — standalone expert work
Reports
Where Outcomes roll up at the organization level