Support Center

How IRIS evaluates every asset in your workspace, how to tell it what a good score means, and which changes take effect the moment you save versus which ones need a re‑extraction.

How scoring works

IRIS reads every asset and scores it against a set of dimensions you control.

Each dimension is one question you want answered about an asset - how clear it is, how feasible, how well it fits your strategy. IRIS scores each dimension from 0 to 100, and a higher score is always better. There is no dimension where a low number is the good outcome; if you find yourself wanting one, invert the wording of the dimension instead.

Each dimension carries a weight. The weights combine the individual dimension scores into the single overall score you see on an asset. Weights are pure arithmetic - they decide how much each answer counts, not what the answer is.

Everything else on the setup page exists to tell IRIS how to decide a dimension's score: what strong evidence looks like, what weak evidence looks like, what to do when the information simply isn't there, and what it must explain when it reports back.

Where to find it

Scoring is configured per use case. Ideation and Project Management have separate dimensions, separate weights, and separate publish states — configuring one never changes another.

Settings → Use cases → [your use case] → Scoring Model

An amber dot beside Scoring Model in the sidebar means that use case has changes saved but not yet published.

The setup sequence

Work through these in order the first time. Afterwards you can jump straight to whichever step you need.

1. Choose the use case. Open it from the Settings sidebar. Every change you make applies only to it.
2. Decide which dimensions are active. Use the On toggle. A disabled dimension is not extracted, not scored, and not counted in the total. You need at least four active dimensions.
3. Set the weights to total 100%. Edit weights directly in the Weight column - no need to open the form. Only active dimensions count toward the total, and it must be exactly 100% before IRIS will extract.
4. Write the instructions for each dimension. Open a dimension with the pencil icon and complete the four sections of the form. This is the part that actually determines score quality.
5. Order the list. Drag rows by the grip handle, or use the row menu, to put the dimensions most important to your reviewers at the top. This is presentation only; it does not affect any score.
6. Test the model. Use Test Scores to run your configuration against real assets and read the reasoning IRIS produces. If a Test required badge is showing, testing is mandatory before you can publish.
7. Publish. Publishing puts your instruction changes into effect. Saved-but-unpublished changes do not influence scoring at all.

The dimensions list

Most day-to-day tuning happens directly in this table without opening a form.

Why Strategic Fit is locked

Strategic Fit has no edit action, and that is intentional.

Every other dimension is scored against instructions you write. Strategic Fit is different: it is scored against the strategic goals configured for your organisation. Its meaning is defined by those goals rather than by a description in this form, so letting it be re-worded here would let the score drift away from the strategy it is supposed to measure. That is why the pencil icon does not appear on its row.

You still have real control over it:

• Turn it off with the On toggle if you are not managing strategy in IRIS. It then drops out of the total entirely.
• Change its weight to decide how much strategic alignment counts against everything else.
• Reorder it like any other row.

Every field in the form

The form has four sections. Required fields are marked with an asterisk in the product.

1. Basic setup

2. Score meaning

This is where you define the scale itself. IRIS uses these three descriptions to place an asset on the 0–100 range, so vague wording here produces vague scores.

3. Missing information behavior

Assets are often submitted incomplete. This setting decides what IRIS does when the evidence for this dimension simply is not in the asset. Pick the option that matches how your organisation actually treats a gap.

4. Evidence and explanation guidance

All four are optional, and all four are the cheapest quality improvement available to you. Filling them in is what moves a model from “plausible” to “trustworthy”.

Weights vs. instructions: what actually re-runs

This is the single most important thing to understand before changing a live model.

Two different kinds of change live on this page, and they behave completely differently.

Weights and names are arithmetic and labelling. IRIS already knows every dimension score for every asset. Changing a weight just re-does the sum. Nothing is re-read, nothing is re-analysed, and no publish is required - the moment you save, index scores recalculate.

Instructions change the judgement itself. Score meanings, missing-information behaviour, evidence guidance, and purpose all change how IRIS decides a score in the first place. Existing assets were scored under the old instructions, so their scores are not automatically correct any more. These changes require a publish, and existing assets need a re-extraction run before they reflect the new thinking.

The form tells you which case you are in. Change a weight and the footer reminds you that extraction is not re-run; change an instruction and it tells you the change applies the next time extraction or scoring runs.

Ordering dimensions

Drag a row by its grip handle, or use the row menu's Move to top / Move up / Move down / Move to bottom.

What ordering does not do is change any number. It carries no mathematical meaning — weights alone decide influence. Moving a dimension to the top does not make it count for more, and a 5% dimension sitting at the top of the list still contributes exactly 5% to the overall score.

Reordering behaves like a weight or a name change: it takes effect when you save, and it never requires a publish or a re-extraction.

Testing before you publish

Open the Test it! tab to test your current configuration against real assets and see, per dimension, what IRIS assessed, what evidence it found, and its reasoning - plus the weighted total.

If a Test required badge appears next to the button, the Publish button stays disabled until you have tested since your last change. Read the reasoning rather than just the numbers: a correct score reached by the wrong reasoning will not stay correct.

What to look for

• Everything scoring mid-range. Usually means score meanings are too vague to discriminate, or too many assets are hitting a Neutral / unknown fallback.
• A dimension that never varies. Either the instructions do not match what your assets contain, or the dimension is not worth its weight.
• Reasoning that cites the wrong evidence. Fix it with “What should IRIS ignore?” rather than by rewriting the score meanings.

Publishing

While you have unpublished changes, the page shows a banner: your changes are saved, but they do not affect IRIS scoring until you publish. Use Revert to Published to discard them and return to the live configuration.

Confirming Publish tells you exactly what it will do:

• Existing assets will require a re-extraction run.
• New submissions will use the new scores.
• Content updates to existing assets will use the new scores.

Names and weights, as above, never need this step.

Before you publish: a checklist

Scoring configuration is per use case. Repeat this setup for each use case you run - Ideation, Continuous Improvement, Project Management, Technology Scouting, and Needs Management each keep their own dimensions, weights, and publish state.