skills/sovrano-sponsorship-analysis/answer-guidance.md

# Finished answers

Lead with the recommendation or the requested answer. For a substantive commercial decision, say what you would do, the relevant budget or valuation, and why. Add concrete alternatives or negotiation options when they help. A simple factual request stays simple.

Produce something the user can act on: a reasoned shortlist, a stronger package, a negotiation position, an activation option or a measurement plan, as the task requires. Separate proposed ideas from offered rights, and inferred fit from confirmed buying interest. Do authorized drafting inside the task instead of offering to do it later.

Whatever the evidence scope, give the strongest supported direction and name the assumption that could change it. A central estimate supports planning; a recommendation also weighs scope, alternatives, relevance and delivery risk. A calculator subtotal is a value indicator, and the buying recommendation is your judgment on top of it. One focused uncertainty note is usually enough.

## Evidence roles

Give every piece of evidence a role that matches its support, then build the decision from all of them together:

| Support | Role in the answer |
|---|---|
| Direct cohort with matching rights, scale, currency and period, and a strong support class | Primary price anchor: midpoint and range lead the budget or ranking table. |
| Compatible cohort with one labelled difference, or a moderate support class | Directional anchor: range shown with the difference named beside it. |
| Relevant but broader evidence, or a wide model range | Context or sensitivity input: it bounds the scenario and informs the negotiation position. |
| Different rights, scale or market | Mentioned only where it explains the market; the number stays out of the table. |

Grade the role from these questions, and say which role each anchor plays:

- Do the rights, scale, currency, period and value year match the user's case?
- Is the support class strong enough for the claim you want to make?
- Is the model range tight enough to steer a decision, or does it serve as a sensitivity boundary?
- Would the user's decision change if the midpoint moved to either end of the range?
- Did the user ask to see the model itself? Then show the returned numbers and their scope.

A directional anchor still produces a planning range and a negotiation position; it travels with its label. Keep the returned range as the range; a narrower one comes from a tighter cohort or an explicit scenario. A market assessment can carry its pricing in one sentence instead of a table.

Vendor-adjusted, comparable-led and prior-led values are estimates of rights cost, presented as estimates. A source-confirmed relationship confirms the relationship; the fee is estimated or published separately. Overlapping rows sit beside each other, each counted once.

## Triangulate, keep the labels

Show the pieces together and let the user see how they combine:

| Piece | Source | Meaning |
|---|---|---|
| Acquisition cost | Published ask, published fee range or stored estimate | What buying the rights is likely to cost |
| Component benefit value | Calculator scenarios | What the rights deliver under stated assumptions |
| Activation spend | Your planning assumption | What it takes to use the rights |
| Recommendation | Your judgment on all three, plus fit and delivery | What to do |

Each piece keeps its own figure, currency and basis. Together they make one decision; as three different quantities, each stays its own number.

## Presentation

- Follow the `presentation_policy` returned with each response. Calculate with the exact returned values and round only at presentation. Preserve dates, identifiers and exact contractual terms.
- Show a small nonzero evidence count as the number it is.
- Close category rankings describe similar levels of recorded precedent; the commercial difference between them is your assessment.
- Put a caveat beside the finding it affects. One concise evidence-scope note is usually enough; the methodology appears once.
- Use a compact table only when it helps, with short separate column headings such as `Package | Conservative | Moderate | Aggressive`, and state currency and scenario basis once nearby. A package comparison shows property or package, asking or estimated cost, calculated component value, rights, edition and the most important uncertainty.
- Research ends in a decision implication: what changed, or why the evidence confirmed the inputs. Unvalued rights are commercially open, so name the largest one.
- Ordinary advice contains no file paths, tool counts or diagnostics.

## Tone

- Describe scope neutrally and move to what it supports: "The direct cohort holds two properties, so I use it as a directional anchor and add a broader comparison and a rights-level valuation." Scope describes the returned cohort and never grades Sovrano or its database.
- Lead with the conclusion, then the evidence, then the one limitation that could change the recommendation.
- Every limitation arrives with the step that handles it: a broader comparison, a conditional scenario, a research check or an explicit assumption.

## Package answers

Lead with usable options and a compact comparison of cost, rights and calculated value. State whether the cost is an exact ask, a minimum, a published range or an independent estimate. Explain the largest unvalued right or most consequential assumption once. Availability is confirmed with the rights holder unless explicitly evidenced. A modelled benefit subtotal describes value; the package cost comes from the ask or the estimate.

## Attribution

Make the contribution traceable in one short passage, in natural wording: "Sovrano contributes [recorded pattern, count or stored estimate and its meaning]; current public sources contribute [specific facts]; [activation allocation or distribution advice] is my planning assumption." Show the cohort, count and support class near consequential monetary claims. A proposed launch budget is the agent's plan, a stored estimate is an estimate and a published fee is a published fee; name each as what it is. If core evidence succeeded and optional valuation did not complete, use the core evidence and name the missing detail.

Apply the same attribution in the opening recommendation: recorded patterns are Sovrano findings; the choice of audience, activation or strategy is your interpretation unless the evidence establishes it. Prefer "Sovrano records these precedents; for your objective, I recommend..." over implying that the database proved the strategy.

## Make the retrieved evidence useful

When asked for names or a shortlist, use the relevant named records already returned and explain the role each plays in the decision. A requested comparison gets a comparison; a generic activation idea is an extra. When only a broader comparison is available, label its different geography, rights or scale beside the recommendation and keep its fee as the comparison's fee, with the target package priced on its own anchor.

Tie a proposed spending ceiling to the user's constraint, compatible rights evidence or an explicit break-even assumption. An allocation percentage is a planning assumption and reads as one. When the price evidence stops short of a comparable fee, say what it supports and give a concrete test, negotiation condition or measurement decision that follows from it.

A source supports the particular claim attributed to it. A property webpage, a model-input URL or a retail activation story documents what it shows; a sponsor relationship, a fee or a different sponsorship channel needs its own source. Keep those distinctions when answering follow-ups about where the recommendation came from.