The big picture
Every Mentova analysis follows the same five-stage loop: generate prompts, run them across AI models, extract brand mentions from the responses, compute visibility metrics, and surface opportunities. Understanding this flow helps you set realistic expectations and interpret your results correctly.
Stage 1 - Prompt generation
You define a topic (a search theme or question area relevant to your brand) and Mentova generates a set of realistic prompts from it. These prompts simulate the kinds of questions real users actually ask AI assistants on that theme.
The number of prompts per topic depends on your plan. You can also write or edit prompts manually if you want to test specific phrasings. If you selected an avatar, the prompt generation incorporates that persona's context so the questions feel authentic to that audience profile.
Stage 2 - Running the campaign
When you launch a campaign, every prompt is sent to every AI model you selected. Each prompt × model combination is called a run. Mentova handles the API calls, retry logic, and response storage automatically.
Campaign runs are queued and processed asynchronously - you do not need to keep the app open. You will receive a notification (and optionally an email) when the campaign completes.
Stage 3 - Brand mention extraction
Each AI response is processed by Mentova's extraction pipeline. The pipeline identifies every brand, company, or product name mentioned in the text using a combination of:
- Pattern matching against known brand names and aliases.
- Named entity recognition (NER) heuristics to catch brands not yet in the list.
- Deduplication and normalization so variant spellings (e.g. "Adobe Inc." and "Adobe") are counted as one entity.
Entities appearing in only one run are filtered out as likely noise. Your own brand is always retained.
Stage 4 - Visibility metric computation
With mention data collected, Mentova computes the core metrics for every entity found:
- Mention Rate - percentage of runs where the entity was mentioned.
- Share of Voice - the entity's share of all brand mentions across runs.
- Average position - where in the response the brand typically appears (earlier = more prominent).
- Sentiment - the tone of the mentions.
These metrics are aggregated across all completed campaigns and cached for fast dashboard loading. A snapshot is also stored per campaign so you can track trends over time.
Stage 5 - Opportunities and sources
From the visibility data, Mentova identifies two types of insights:
- Opportunities - topic angles and content formats where competitors appear but you do not. These are ranked by potential impact and shown in the Opportunities page.
- Sources - websites cited in AI responses related to your topics. Domains that cite competitors but not you are highlighted as gaps worth addressing.
How long does it take?
Most campaigns complete within a few minutes. Larger campaigns (many prompts × many models) may take longer as runs are processed in parallel queues. The campaign status indicator on the dashboard updates in real time.
Common pitfall
Running a single campaign gives you a snapshot, not a trend. Visibility metrics become meaningful when compared across multiple campaigns over time. Enable scheduled campaigns (daily, weekly, or monthly) to build a time-series view automatically.