← Back to AI Insights
Gemini Executive Synthesis

Inaccurate reporting of web search execution, conflating requested capability with actual usage.

Technical Positioning
Transparent and accurate reporting of AI provider capabilities and usage, clearly distinguishing between requested and executed searches.
SaaS Insight & Market Implications
This bug directly undermines the credibility of NiubiGEO's AI brand visibility and competitor reports by misrepresenting provider-native web search execution. Reporting a search as 'used' merely because it was 'requested' creates a false impression of data grounding and provider activity. Clients rely on these reports to understand how AI models interact with external information sources. This inaccuracy distorts the perceived analytical depth of responses, making it impossible to discern when a provider genuinely leveraged web search versus when the capability was merely enabled. This flaw impacts trust in the platform's data integrity and the actionable insights derived from its audits.
Proprietary Technical Taxonomy
provider-native web search SearchExecution.used webSearchEnabled alwaysOn search tool plugin search call/query provider-native citation

Raw Developer Origin & Technical Request

Source Icon GitHub Issue Sep 4, 2026
Repo: Albert-Weasker/niubigeo
[Bug]: Web search is reported as used without execution evidence

### NiubiGEO version or commit

fd30617 (`Add provider web search and intent analysis layer`)

### What happened?

The new provider-native web search path records both the request and the execution status, but `SearchExecution.used` is currently derived from `webSearchEnabled || alwaysOn`.

For providers that are not always web-grounded, turning the option on makes `used` true as soon as NiubiGEO sends a search tool or plugin. The response may still contain no search call/query and no provider-native citation. `AuditRunner` then copies this value to the run, and the report shows `Provider-native web search`.

That makes the report describe a requested search capability as an executed search. It is especially confusing when a provider chooses not to call its available search tool for a particular prompt.

### Steps to reproduce

1. Use a provider with optional native search, for example OpenAI or OpenRouter.
2. Enable Web search and run a prompt whose provider response contains no recorded search action/query and no provider-native citations.
3. Inspect `result.search` or the HTML report.
4. `requested` is true, but `used` is also true and the report labels the answer as `Provider-native web search`.

The same behavior can be reproduced with the provider adapter tests by returning a normal completion without a search event or provider-native citation.

### Expected behavior

`requested` and `used` should remain separate:

- `requested` should mean NiubiGEO sent the provider-na...

Developer Debate & Comments

No active discussions extracted for this entry yet.

Adjacent Repository Pain Points

Other highly discussed features and pain points extracted from Albert-Weasker/niubigeo.

Extracted Positioning
Inadequate sentiment and recommendation analysis for Chinese language content.
Comprehensive and culturally nuanced sentiment analysis across multiple languages, accurately interpreting Chinese phrases for precise brand visibility insights.
Extracted Positioning
Inaccurate citation counting due to insufficient deduplication of URLs from different sources.
Accurate and non-redundant reporting of source citations, ensuring canonical URLs are counted once per response, prioritizing native provider citations.
Extracted Positioning
Suboptimal multi-provider audit planning, where a single missing API key blocks all other configured providers.
Robust and resilient multi-provider audit execution, gracefully handling individual provider configuration failures while allowing others to proceed.
Extracted Positioning
Inaccurate GitHub repository citation classification due to a broad substring check.
Precise and unambiguous attribution of GitHub citations, distinguishing between target repositories and similarly prefixed third-party repositories.
Extracted Positioning
Inaccurate brand name detection in prompt targeting, leading to misclassification of unbranded prompts.
Accurate brand visibility and awareness metrics, differentiating between standalone brand terms and substrings for precise reporting.

Frequently Asked Questions

Market intelligence mapped to Inaccurate reporting of web search execution, conflating requested capability with actual usage..

What problem does Inaccurate reporting of web search execution, conflating requested capability with actual usage. solve?
Based on our AI analysis of the original developer request, its primary technical positioning is: Transparent and accurate reporting of AI provider capabilities and usage, clearly distinguishing between requested and executed searches.
What are the foundational technologies related to Inaccurate reporting of web search execution, conflating requested capability with actual usage.?
Our proprietary extraction maps Inaccurate reporting of web search execution, conflating requested capability with actual usage. to adjacent architectural concepts including provider-native web search, SearchExecution.used, webSearchEnabled, alwaysOn.

Engagement Signals

0
Replies
open
Issue Status

Cross-Market Term Frequency

Quantifies the cross-market adoption of foundational terms like plugin and completion by tracking occurrence frequency across active SaaS architectures and enterprise developer debates.