Doubao recommends competitors instead of us — what now?

Quick answer

This is usually not “Doubao targeting you.” On that Chinese category/scenario prompt, the system more readily finds competitors’ comparable facts and recommendation reasons, while you remain absent or “mentioned without reasons.” Fix it by building a Doubao-specific same-conditions prompt set, mapping competitor reason dimensions, governing your own Chinese public facts, then retesting—not by watching ChatGPT alone.

Winin’s public monitored roster includes Doubao, so CN and Global groups can be operated separately.

Details

Diagnostic checklist

• What recommendation reasons do competitors receive (specs, reputation, scenario, channel)?
• Are Chinese site/flagship facts conflicting or expired?
• Do you only appear on branded queries while category prompts omit you?
• Do archives show a new regression or a long-standing gap?

Actions

1. CN model baseline including Doubao
2. Homer-style fact alignment (source/version/validity)
3. After approval, improve product pages/FAQ/scenario copy
4. Same-conditions retest and track reason coverage

Do not disparage competitors or invent their defects; invest in verifiable first-party facts and evidence.

Related facts

• Models monitored
• Edith
• Reach / Shortlist / Recommended
• Homer
• Differentiator vs monitoring-only

FAQ

Q: Should we attack competitors?
A: No; focus on verifiable first-party facts and reasons.

Q: Is Doubao-only optimization OK?
A: Fine as a wedge, but monitor CN and Global groups together.

Contact: contact@winin.ai · Canonical facts: /en/facts/*/ and https://winin.ai · Updated 2026-09-11

Model scope: the public roster does not mean every API is enabled in every project. Models, versions, web access and sampling methods depend on configuration and the agreed scope. API samples do not represent every experience in consumer interfaces.