What this case study is about
Avni Wellness is a menstrual-care brand. AutoAgent worked with the brand on Generative Engine Optimization (GEO): making its products understandable to AI systems that answer consumer questions directly.
Traditional search gives consumers a list of options and leaves most of the research to them. Conversational AI changes that: a consumer can describe their situation and ask what they should choose.
So the opportunity for Avni was not simply ranking for a product keyword such as "menstrual cup". The opportunity was to make Avni's products understandable in relation to the things an AI system has to reason about before it can answer.
- Consumer problems
- Use cases
- Product categories
- Evidence
- Comparisons
- Recommendations
AI recommendation requires more than product-name recognition; it requires contextual information about the product, use case, evidence and alternatives.
The new consumer decision journey
A consumer might once have searched "best sanitary pads for sensitive skin" and then worked through the results themselves.
Traditional journey
- Search query
- Search results
- Websites
- Product pages
- Reviews
- Comparison
- Purchase
In that journey the consumer is responsible for doing most of the research.
AI models such as ChatGPT and Gemini change the journey. A consumer can instead ask: "I have sensitive skin and get irritation during my period. What sanitary products should I consider?"
The consumer is no longer simply asking "Where can I find information?" They are asking "Based on my situation, what should I choose?"
Why AI search feels different
Traditional search gives the consumer a set of options. Conversational AI attempts to interpret the question and construct an answer.
That makes the interaction feel more like a conversation than a conventional search session.
Research into conversational AI has found that factors including social presence, conversational relevance, personalization and rapport can influence how users interact with AI systems and how willing they are to disclose information. Studies have also explored how users can perceive conversational agents as social or relational actors rather than simply information-retrieval interfaces.
This matters particularly when the subject is personal - as it is in menstrual care.
AI is becoming part of the research process
This does not mean people automatically trust everything AI says. They don't. But the interaction creates a different research behaviour: instead of opening ten tabs, a consumer can keep asking.
One continuous conversation
- "Give me the best options."
- "Why would you recommend that one?"
- "What are the disadvantages?"
- "Compare it with another product."
- "Which one would you choose for my situation?"
The research process happens inside the conversation. That creates a new opportunity and a new distribution channel for brands whose information is easy for AI systems to use.
What AI needs before it can recommend a product
A recommendation is usually the final output of several smaller understandings. Consider a simple consumer question: "What should I use for heavy periods at night?"
The chain AI has to connect
- Heavy periods
- High-protection requirement
- Overnight use
- Relevant product category
- Product
- Evidence
- Comparison
- Recommendation
If a brand's website only says something like "our product is best for heavy flow", the AI still has to work out the rest on its own.
Questions the AI is left to answer: Who is this for? What problem does it solve? When should I recommend it? Why this product instead of another one? That is where many product pages fall short.
From product content to problem content
Take the example of a menstrual cup.
Weak AI-readable relationship
- Avni
- Menstrual Cup
Stronger relationship
- Avni
- Menstrual Cup
- Reusable Period Protection
- Pad-Free Periods
- Beginner Users
- Sleep / Swimming / Exercise / Travel
Now imagine the consumer asks: "What menstrual product can I use while travelling?"
The product is no longer just associated with the keyword "menstrual cup". It is associated with a specific use case. The same principle applies across a product catalogue.
Why comparison content matters
AI recommendations are rarely made in isolation. Consumers ask which is better, what the difference is, whether they should choose A or B, and whether something is worth it. A product therefore needs information that helps AI understand its position relative to alternatives.
| Comparison built for Avni | Consumer question it helps answer |
|---|---|
| Reusable pads vs disposable pads | "What's the difference?" |
| Period panties vs sanitary pads | "Should I choose A or B?" |
| Menstrual cup vs sanitary pads | "Which is better for my situation?" |
| Different Avni products for different menstrual-care needs | "Which one is right for me?" |
Decision-oriented comparisons published as readable, structured content rather than images of tables.
The objective is not to say "Avni is always better". It is to make the decision criteria explicit - for people and for AI systems alike.
How AI recommendations require evidence
A brand shouldn't expect AI to recommend a product based only on marketing claims. AI needs accessible information such as:
- Product specifications
- Materials
- Usage instructions
- Product reviews
- Certifications and testing where applicable
- Comparisons
- FAQs
- Brand information
- Category expertise
- Consistent information across the website
These signals give AI systems more accessible context from which to understand and evaluate a product. They do not guarantee a recommendation.
The information graph behind a recommendation
There is an important part of AI visibility that consumers never see. Before an AI answer can suggest "consider Avni", the system needs enough information to construct that answer.
What the consumer sees
- Recommendation
What sits behind it
- Question
- Intent
- Category
- Brand
- Product
- Evidence
- Comparison
- Recommendation
GEO is not simply about getting mentioned. It is about making the brand easy to understand and easy to justify.
AI-referred traffic increased by 100%
The change was also visible in Avni's analytics.
Source: Avni Wellness Google Analytics 4. Exact comparison period to be added.



Traffic alone wasn't the most important observation
- Increased engagement time from AI-sourced visitors
- Stronger engagement overall
- Higher conversion performance compared with other organic sources
From ranking for keywords to owning problems
This is the broader lesson from the case. Traditional SEO often starts with: "What keyword should we rank for?"
AI discovery increasingly requires another question: "What problem should AI associate this brand with?"
For a brand like Avni, those problems could include:
- Sensitive-skin menstrual care
- Reusable menstrual protection
- Heavy-flow period protection
- Pad-free periods
- Period-product cleaning
- Period-stain removal
- Sustainable menstrual care
The brand becomes discoverable through the problem, not only through the product name.
Understand → Match → Prove → Recommend
The Avni project led us to a simple framework for AI recommendation readiness.
Understand
- Who is the brand?
- What does it sell?
- What category does it belong to?
Match
- Which consumer problem does it solve?
- Which user is it relevant to?
Prove
- What evidence supports the product?
- How does it compare with alternatives?
Recommend
- Can AI confidently explain why this product fits the user's situation?
AI doesn't need another brand description. It needs enough context to make a decision.
What brands can learn from Avni
- 1Build content around problems, not only products.
- 2Make use cases explicit.
- 3Give AI enough evidence to evaluate products.
- 4Make comparisons and decision criteria clear.
- 5Maintain consistent brand and product information across the site.
- 6Structure content so both people and machines can understand it.
- 7Measure AI-referred traffic and downstream engagement, not only traditional search traffic.
Questions about AI-driven product discovery
Want your brand recommended by AI?
AutoAgent helps brands become understandable, comparable and citable across ChatGPT, Gemini, Claude and Perplexity.
