> ## Documentation Index
> Fetch the complete documentation index at: https://docs.goosybear.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Strategy and taste

> Declare what you stand for, and let Goosy keep what he learns from your feedback — so you stop correcting the same thing every time.

Brand says how you sound. **Strategy** and **Taste** say what you are trying to
achieve and what you have already learned about achieving it — the two settings
that stop you re-explaining yourself in every request.

They sit next to each other in **Settings** for that reason: Strategy is *what
we are trying to say*, Taste is *what we found out saying it*.

## Strategy

**Settings → Strategy** is where you declare the strategy this workspace
operates against. Everything Goosy makes is then scored against it.

### Nine statements

The **Strategic statements** tab holds nine kinds of statement, each declared
separately:

| Statement              | What it answers                      |
| ---------------------- | ------------------------------------ |
| **Mission**            | Why you exist.                       |
| **Vision**             | Where this is going.                 |
| **Positioning**        | Where you sit against everyone else. |
| **Value proposition**  | Why someone should choose you.       |
| **Differentiator**     | What only you can say.               |
| **Strategic goal**     | What you are pushing for right now.  |
| **Content pillar**     | The themes you keep coming back to.  |
| **Competitive stance** | How you talk about the alternatives. |
| **Audience promise**   | What a reader gets, every time.      |

You do not need all nine. Anything you have not written reads **Not declared**
and offers **Declare**; anything you have written shows its version and offers
**Edit**.

### Versions, not overwrites

Every statement is versioned. A declared statement carries a version marker and
a **History** link, so a re-declaration is a new version rather than a silent
overwrite, and you can see what you used to say.

### The alignment report

The **Alignment report** tab scores how well the strategy you declared actually
fits your material, your audience, your market and your performance signal.

It opens with one **overall alignment** figure and when it was last calculated,
then breaks it into six dimensions:

* **Internal consistency** — do your own statements agree with each other?
* **Statement vs corpus** — does what you publish match what you said you stand
  for?
* **Statement vs audience** — does it match the people you say you are for?
* **Statement vs market** — does it hold up against the space you sell into?
* **Statement vs trend** — is it still true of where things are moving?
* **Performance reality** — does the work that performs match the strategy?

A dimension with nothing to go on reads **N/A** rather than inventing a number.

Below the scores, **Recommendations** name specific gaps and offer the action
that closes each one — for example, adding a content pillar when good topic
ideas keep arriving that no existing pillar covers.

The report refreshes weekly on Sunday night and again whenever you re-declare a
statement. **Re-run now** recalculates it on demand.

## Taste

**Settings → Taste** is headed *what you've taught us*, and that is exactly what
it is: your own notes, from your own work. The screen tells you how many are in
play, and every run reads them before it writes a word.

You do not write this from scratch. It fills up as you work, which is why the
second month is easier than the first.

### Three kinds of note

Notes are grouped by what they are:

* **What worked** — anything you would want again next time.
* **What didn't work** — anything you had to fix, or would not repeat.
* **What could be better** — one change that would have made something easier.

### Reading a note

Each note shows the lesson in your own words, then what it applies to —
**Everything**, or one kind of output such as scripts or titles — and where it
came from: **after a run**, or written here by hand.

### Retiring and restoring

**Retire** on any note takes it out of play; the next run stops hearing it. A
retired note is not deleted — it stays visible, marked retired, with **Bring
back** beside it.

### Adding one yourself

The **Add a note** form takes three things: whether it worked, did not work or
could be better; what it applies to; and the lesson in your own words. The
screen puts it well — *something you learned that we should not have to be told
twice*.

### The end-of-run question

Goosy also asks. One screen at the end of a run puts the three questions to you,
and it is always skippable.

### Where the notes go

Every script, title, description and thumbnail written for this workspace reads
the notes that apply to it, alongside your brand voice. That is the whole point:
correct something once, here, instead of correcting it in every draft.

## What to do next

* **[Production profiles](/setup/production-profiles)** — the look and shape a
  render comes out in.
* **[Brand](/setup/brand-profile)** — the voice all of this is spoken in.