# See the recommendation gap. Act on what the evidence supports.

> First establish where a brand or product is absent, misunderstood or lacks sufficient recommendation evidence. Then act on approved changes and retest them under the same measurement conditions.

- Canonical page: https://winin.ai/en/services/
- Language: English
- Last updated: 2026-09-04

## From a model answer to work that can ship.

*01 · MONITOR*

### [Edith sees the recommendation gap](/en/products/edith/)

Move from a bounded free signal to a formal baseline that retains the question, model, raw answer and sources shown in the answer.

*02 · GOVERN*

### [Homer confirms the usable facts](/en/products/homer/)

Resolve sources, conflicts, owners, markets, versions and validity before a public claim enters the workflow.

*03 · ADVANCE*

### [Friday advances the GEO work](/en/products/friday/)

Turn confirmed gaps and approved facts into website, product-data, channel and source tasks, escalating only material decisions.

## Choose the customer decision you need to improve.

*BRAND GEO*

### [What evidence supports the brand entering consideration?](/en/services/brand-geo/)

Clarify identity, category, positioning, capabilities and the evidence that supports a qualified recommendation.

*PRODUCT GEO*

### [Which buying situations fit the product, and why?](/en/services/product-geo/)

Clarify product relationships, specifications, scenario fit, comparisons and time-sensitive commerce facts.

## Seeing the gap starts at one of two depths.

### Free initial check

Five fixed questions, one sample each and up to three active model APIs provide a limited initial signal.

[Start the free check →](/en/free-ai-presence-check/)

### Formal baseline

Confirm the real question set, market, models, repeated runs, raw answers and metric denominators before making changes.

[Discuss a formal baseline →](/en/contact/)

## Optimization work is matched to the actual gap.

Homer supplies the approved fact version; Friday then chooses the relevant work: a fact correction, website or product-page change, structured data, product data, an authorized channel update or an evidence package for a legitimate external source.

- Missing evidence pauses the public claim for review
- Material facts and every external publication require named approval
- Publication is recorded only after the real URL and content are verified
- Edith retests the same questions under comparable conditions

## Every step has a deliverable and an evidence gate.

| Step | Customer receives | Evidence gate |
| --- | --- | --- |
| Free Check | A private initial-signal report | Successful runs within the stated API scope |
| Formal baseline | Repeated runs, raw answers, sources shown in the answer and confirmed gaps | Frozen scope and comparable measurement conditions |
| GEO improvement | Approved fact, page, data, channel and source tasks | Every material statement maps to an approved fact |
| Comparable post-test | A bounded account of what changed and what remains uncertain | Verified publication and the same measurement conditions |

## Managed monitoring opens the next improvement cycle.

Monitor recommendation answers, competitor movement, fact accuracy and sources shown in the answer over time. Each traceable gap is routed to the product that owns the next decision.

- Consideration and recommendation changes by question family
- Fact errors and outdated product details
- Competitor movement and source drift
- Route each issue to Edith review, Homer fact governance, Friday execution or human judgment

## Keep progress measurable from the start.

> Winin focuses on the facts, evidence and public paths an enterprise can improve. Named owners confirm material facts and external publication, while comparable runs show what changed over time.

## Source

This Markdown document is the machine-readable counterpart of https://winin.ai/en/services/. The canonical HTML page remains the public presentation source.