QUEST monitors how large language models mention and position your brand across AI-generated responses. This guide resolves scan failures, missing brand mentions, topic generation errors, and snapshot data gaps — including the automated backfill introduced in LLMV-460 for the May 12–18, 2026 incident.

QUEST queries multiple AI systems and analyzes brand mentions, topical authority, sentiment, response frequency, and AI-generated positioning. Because results depend on third-party LLM providers, asynchronous query queues, and evolving AI training datasets, visibility scores can fluctuate over time and across providers.

## ⚠️ Error 1: QUEST Scan Not Completing — "All Submissions Failed"

**What's happening**

The OpenRouter AI submission queue (or upstream LLM provider) is overloaded. This high-severity queue failure appears most often during traffic spikes or large multi-query scans. Brands with many tracked entities, competitors, or broad topic coverage require significantly more processing time.

**Steps to try**

1. Wait 30–60 minutes before retrying.
2. In the left sidebar, click Authority > PR Distribution > QUEST (AI Visibility). On the QUEST dashboard, locate the **[Analyze query]** button at the top-right of the page and click it to trigger a fresh scan.
3. Reduce the number of simultaneously tracked topics.
4. Avoid launching multiple QUEST scans at the same time.
5. Refresh the QUEST dashboard after restarting the scan.
6. Verify your internet connection remained stable during submission.
7. Check whether other AI-dependent modules are also slow.
8. Retry during off-peak hours: early mornings, late evenings, or weekends.

**Contact support if:**

- Scans fail repeatedly after 2 hours.
- Multiple projects fail simultaneously.
- Scans never move beyond "queued" status.
- The failure persists after 24 hours, or a scan never progresses past 0%.

**Important notes**

- QUEST depends on third-party LLM provider availability.
- Queue overload does **not** indicate account corruption.
- Larger scans naturally require more processing time.
- If a scan fails mid-run, LLMV-460's snapshot-gap detection automatically schedules a backfill for any missing days once the queue recovers — no manual action required.

## ⚠️ Error 2: Brand Mentions Not Appearing After a Successful Scan

**What's happening**

Your brand may be too new, insufficiently referenced online, or not yet strongly represented in current LLM training data. QUEST only reflects what AI systems already know or infer about your brand — it does not fabricate results.

**Steps to try**

1. Confirm your brand has indexed content, backlinks, citations, and public mentions online.
2. Complete your Brand Vault fully: go to **More Features → Brand Vault** and add your company description, products/services, website URL, industry, and audience information.
3. Expand your tracked topic list and use broader query phrasing temporarily.
4. Add branded and non-branded topic combinations.
5. Publish additional blog content, PR mentions, social profiles, and directory citations.
6. Run scans again after new content gets indexed.
7. Compare visibility across different LLM providers.

**Known Issue — Resolved (Competitor Mentions)**

Between approximately early May 2026 and the LLMV-453 fix date in mid-May 2026, a parser bug in the CompetitorList component caused competitor brand mentions to be silently dropped when certain AI providers (notably `meta-llama/llama-3.1-8b-instruct` via OpenRouter) returned a top-level JSON array instead of the expected wrapped object format. This has been fixed. If your competitor data still appears incomplete for that window, run a manual rescan to repopulate the affected mentions.

**Contact support if:**

- Established brands consistently show zero mentions.
- Scans return completely empty datasets.
- Competitor mentions remain missing after a manual rescan.

**Important notes**

- QUEST does **not** fabricate brand visibility.
- Newer brands naturally have weaker AI visibility footprints.
- AI Visibility (formerly LLM Visibility) evolves gradually over time.

## ⚠️ Error 3: Topic Generation Fails or Returns Irrelevant Topics

**What's happening**

QUEST generates topic suggestions using AI, based on your brand and industry context. If generation fails or returns off-target results, the AI likely has insufficient brand context — most commonly because Brand Vault is incomplete or the brand category is too generic.

**Steps to try**

1. Go to **More Features → Brand Vault** and verify all brand details are complete and accurate.
2. Add a specific industry category and a detailed audience description.
3. Manually enter topics that are directly relevant to your brand.
4. Remove overly broad topic suggestions and replace them with niche-specific queries.
5. Re-trigger topic generation after updating your Brand Vault.

**Contact support if:**

- Topic generation consistently fails after multiple attempts.
- All generated topics are entirely unrelated to your industry.

**Important notes**

- Topic quality improves as your Brand Vault information becomes more complete.
- Manually curated topics typically perform better than auto-generated ones for niche brands.

## 📅 Snapshot Gaps in Share-of-Voice and Sentiment Trend Lines

**What's happening**

A **snapshot gap** is a missing data point in your Share-of-Voice (SoV) or sentiment trend charts — caused by a queue outage, LLM provider downtime, or a scan that stalled without fully recovering. Before **LLMV-460**, these gaps were permanent and required manual intervention.

**LLMV-460: Automated Backfill (no action required)**

LLMV-460 ships an automated snapshot-gap detection and backfill mechanism. When the system detects a missing snapshot day, it automatically schedules a date-constrained Celery backfill task scoped to the affected date range. Your trend lines self-heal — no manual intervention needed.

**May 12–18, 2026 incident**

Between May 12 and May 18, 2026, the QUEST snapshot pipeline experienced a queue stall that interrupted daily snapshot generation for some projects. Affected customers would have seen missing daily data points, flat-lined visibility scores, or zero-data days in their SoV and sentiment trend lines during that window. LLMV-460's automated backfill will fill confirmed gaps within 24–48 hours depending on current queue load. If your data still looks wrong after backfill completes, contact support.

**Known limitation (LLMV-489, fix in release)**

For days within the May 12–18 window where a scan failed and an automatic retry is still pending, the gap-detection sweep may have scheduled a backfill concurrently with the live retry. This race condition can produce low-fidelity backfilled data alongside the retry result. If your data for those specific days still looks incomplete after 24 hours, contact support so an engineer can verify backfill vs. retry state for your account.

**Report Builder data alignment (RB-1252, resolved late May 2026)**

If you are embedding AI Visibility (formerly LLM Visibility) data in a Report Builder report and the numbers differ from the standalone QUEST dashboard, this was caused by a date-range and filter bug — the AI Visibility (formerly LLM Visibility) topic widget applied an unconditional +1 day offset to the end date and used an overly broad query\_type filter. The bug is now fixed. Regenerate or refresh any saved Report Builder reports created before the late-May 2026 fix to get aligned figures.

**OpenRouter platform list cleanup (May 2026)**

If OpenRouter previously appeared in your tracked AI platforms list and is now absent, this was corrected automatically as part of a data cleanup in May 2026 — OpenRouter is the upstream routing layer and was never an intentionally selectable platform in QUEST.

**What you'll see during backfill**

- Affected date ranges appear as **"Processing"** in the QUEST dashboard until backfill completes.
- Trend lines update automatically once backfill finishes — no manual refresh needed.
- Backfilled scores reflect actual LLM response data, not interpolated values.

**Contact support if:**

- Gaps from May 12–18, 2026 are still visible after 48 hours.
- New gaps appear outside a known outage window.
- Trend lines show data but values appear inconsistent.
- Report Builder numbers still disagree with the standalone QUEST dashboard after regenerating the report.

**🎯 You now know how to resolve the most common QUEST issues — from scan queue failures and missing brand mentions to snapshot gaps that self-heal via LLMV-460's automated backfill. For questions about Brand Vault setup or LLM provider coverage, contact the Search Atlas support team.**