Ashba Botanics teaches the agent a community's vocabulary
Speaking a community's language without getting it wrong
- Client
- Ashba Botanics
- Platform
- Shopify
- Market
- India
- Window
- 24 Jul to 7 Sep 2026
What Ashba Botanics has live
The single channel. Carries a chat in a category with its own technical vocabulary.
Trained on the brand's own voice, ingredient philosophy and curl-type guidance rather than generic haircare copy.
Guided curl-type matching for shoppers who can describe their hair but cannot name a product.
Product-aware prompts on a catalog where every item is part of a routine, not a standalone purchase.
Ashba Botanics makes clean, sulphate-free haircare built specifically for Indian curly and textured hair. Founded by Asha Barrak, who started the Right Ringlets blog in 2014 and built the Indian Curl Pride community to more than 80,000 members before launching the brand in 2019, it holds PETA cruelty-free certification, an EWG green rating and ECOCERT ingredients. Sold direct from ashbabotanics.com, Ashba runs its storefront chat on the YourLio Website Agent.
Challenge: a community that will notice immediately if you get it wrong
Curly hair is not a product category, it is a subculture with a technical vocabulary. Shoppers describe themselves by curl type (2C, 3A) and ask about frizz, definition, volume collapse and whether a leave-in needs rinsing. The recorded themes across the window are exactly that: lightweight volume without collapse, frizz control paired with definition, curl type suitability for 2C to 3A, and application guidance for serum and leave-in conditioner.
Ashba's founder built an 80,000-member community before she built a product. That is the brand's entire moat, and it is also the risk: a generic chatbot answering a 3A shopper with generic haircare copy does more damage than no chatbot at all.
The recorded objections show where the chat actually gets hard. Shoppers seeking frizz relief worry that a routine will not be enough for the fast results they want. And many give only a fragment ("Wavy", "Fine") which signals uncertainty rather than a specification, and needs clarifying before any recommendation can be made.
There were operational ones too: an out-of-stock travel pack with no restock timeline, and a trust risk where orders showed delivered but customers said the parcel had not arrived.
- “What products do you have for frizz?”
- “How should I use this for my curls?, Curl Moisture Milk”
- “Do I have to wash this conditioner after use?, Leave-In Conditioner”
- Worry that a routine will not deliver frizz relief fast enough
- Shoppers offering only a fragment ("Wavy", "Fine") instead of a curl type
- An out-of-stock travel pack with no restock timeline given
Solution: train the voice, then let a quiz do the narrowing
AI Sales Chat was trained on Ashba's own language (the curl-type system, the ingredient philosophy, the application sequence, and the way the founder writes) rather than on generic haircare content. In a category this literate, brand voice is not a presentation layer, it is the accuracy layer.
AI Quiz and Discovery handles the shoppers who cannot specify. Someone who types only "Wavy" is not being unhelpful; she genuinely does not know her curl type. The quiz asks adapting questions and returns a routine, and it completed 99 sessions in 46 days on a small traffic base.
The most common questions in the window were application questions, not selection questions, "Do I need to wash it off after use?", "How should I use this for my curls?" That is a routine category, and the agent has to teach the routine, not just sell the bottle.
Email AI and Segmentation are the obvious next step for this account and are not yet live. Every quiz completion is a curl type and a concern captured, a lead-generation and list-building asset the brand is not currently using.
Strategy: how Ashba runs it now
The highest resolution rate of the fifteen brands here, on the smallest traffic base. Small and specific turns out to be a good shape.
Train the vocabulary before the catalog. Product Info led at 443 chats. In a category where shoppers self-identify as 2C or 3A, an agent that does not know the system cannot answer the first question, no matter how well it knows the price list.
Use the quiz where shoppers can't self-specify. 99 quiz completions from several hundred chats. When a shopper offers only "Wavy" or "Fine", the right move is a structured set of adapting questions, not a guess.
Teach the routine, don't sell the bottle. The most repeated questions were about application, whether to rinse, how to layer, when to apply. Curly haircare is bought as a sequence, and the six most-discussed products appeared on 45 or 46 of 46 days, which is what a routine looks like in the data.
Chat intent mix, 24 Jul to 7 Sep 2026
| Intent | Chats | Type |
|---|---|---|
| Product Info | 443 | Application and suitability |
| Product Discovery | 178 | Curl-type matching |
| Product Browsing | 146 | Catalog navigation |
| Guided quiz | 99 | Structured discovery |
| Checkout Enquiry | 40 | Pre-sale |
| Order Tracking | 15 | Post-purchase |
| Offer / Rewards | 15 | Pre-sale |
Results, 24 Jul to 7 Sep 2026
| Metric | Result | Note |
|---|---|---|
| Resolved without human handoff | 95.8% | highest of the fifteen brands here |
| Chat engagement | 18.4% of sessions | 24 Jul to 7 Sep 2026 |
| Chats referencing a product | 66.0% | |
| Guided quiz completions | 99 | 14.8% of chats |
| Ended in an order | 9.9% | on a premium, routine-led catalog |
| Pre-sale share of chats | ~97% | order and return intents under 4% |
Revenue amounts, order counts and average order values are withheld at the brand's discretion; only rates and ratios are published. Email AI and Segmentation are described as a recommended next step and were not live in this window.
Is this replicable for a brand like yours?
The conditions this deployment ran under, so you can judge how close your own store is.
| Dimension | Ashba Botanics |
|---|---|
| Channels live | Website Agent only |
| Category vocabulary | Curl types 2C to 3A, porosity, definition, plopping |
| Community | 80,000+ member group predating the brand |
| Purchase shape | Routine of 3 to 5 products, not single items |
| Public price band | ₹699 to ₹1,199 (listed on ashbabotanics.com) |
| Traffic base | Small and highly qualified |
| Market | India |
Steal this playbook
- 01
Train the vocabulary before the catalog
In a literate category, an agent that does not know the domain system cannot answer the first question.
- 02
A small, specific catalog resolves better than a broad one
The highest resolution rate here came from the smallest brand.
- 03
Use a quiz when shoppers can't self-specify
A one-word answer is uncertainty, not a specification.
- 04
Teach the routine
Application questions outnumbered selection questions. Sequence is the product.
- 05
Every quiz completion is a lead
Curl type plus concern is a segment. Ashba is capturing it and not yet using it, Email AI and Segmentation are the obvious next step.
Questions about this deployment
By training it on the brand's own language and domain system rather than generic category content. For Ashba that means the curl-type framework, the ingredient philosophy and the application sequence. In a literate category, brand voice is the accuracy layer, not a presentation choice.
Significantly improved engagement and personalized interactions. Conversions are up. A powerful AI solution that truly delivers.
More brands running YourLio
All case studies
Published 2026-09-08 · Last updated 2026-09-08 · Window 24 July 2026 to 7 September 2026 (46 days), measured from YourLio daily snapshots. Revenue amounts, order counts and average order values are not published.