Case Study
Heterize: how Amazon seller onboarding scaled with Hinglish voice AI
What 400,000 AI calls to Amazon sellers in India taught us about finding the sellers who are ready to start, and how OmniDimension's voice AI qualified them in under two minutes.
Published November 16, 2026


400,000+
AI calls to Amazon sellers across India
200,000
Sellers reached
12,000+
Interested sellers qualified as hot leads
5,500+
Hot leads with a regular GST number, ready to onboard

At a glance
- Customer
- Heterize Infotech, an Amazon seller onboarding and growth partner in India
- Industry
- E-commerce services
- Goal
- Find registered and new sellers who want to start selling on Amazon, and route the ready ones to the onboarding team
- Channel
- Outbound AI calls in Hinglish
- Scale
- 10+ campaigns over 13 months, each with its own pitch
Overview
Heterize Infotech helps sellers in India get started and grow on Amazon. Its lists hold hundreds of thousands of sellers: some registered but not yet live, some new, and some selling with room to grow. Each seller needs a short, personal conversation before the onboarding team can help.
OmniDimension gave Heterize an AI agent that calls each seller in natural Hinglish, finds the problem, pitches Amazon's onboarding support, checks eligibility and sends only the ready sellers to the onboarding team.
The challenge
The problem
A seller list is long, and most sellers are not ready today. The onboarding team needed answers to three questions on every call:
- Is this seller interested in selling on Amazon now?
- What is stopping them? Listings, catalogue, account setup or plain doubt.
- Are they eligible? A regular GST number, a trademark or brand authorization, and a product category.
Why calling by hand did not scale
- A human team can make a few hundred calls a day. The lists held hundreds of thousands of sellers.
- Most calls end in under half a minute, so callers spent their day on people who were not ready.
- Sellers across India answer in Hindi, English or a mix, and a formal script sounds wrong in all three.
- Eligibility checks happened late, so the onboarding team spent time on sellers who could not go live.
Key insights from 400,000 seller calls
1. Sellers who have registered convert best
2. A no takes 30 seconds, a yes gets two minutes
3. "Call me later" is a lead, not a no
4. Hinglish wins the conversation
5. Ask eligibility before the hand-off, not after
The solution: a five-step qualification engine
1. Problem discovery
The agent opens with one direct question: is the seller facing an issue with listings or account support? This finds the reason the seller has not started, in the seller's own words.
2. Benefits pitch
The agent explains the offer: three months of free account management and a dedicated account manager for listings, catalogue and sales support.
3. Eligibility screening
For interested sellers, the agent checks:
- a regular, active GST number (composite GST does not qualify)
- a trademark, brand authorization or purchase invoice
- the product category
4. Objection handling
If the seller hesitates, the agent asks what is holding them back, answers the concern, and offers a one-minute eligibility check. If the seller is busy, it books a callback time.
5. Smart routing
Each call ends with structured data: interest, hot lead status, GST, brand documents, category and callback time. Hot leads go to the onboarding team, and callbacks go back into the calling queue.
What made it work
- Natural Hinglish with simple words, never formal translated Hindi.
- One agent per campaign, so each list got the pitch that fit it.
- Plug and play: new lists went live as new campaigns without changes to the team's process.
How it works
The diagram at the top of this page shows the flow:
- Heterize uploads a seller list as a campaign.
- The agent calls each seller and asks about their Amazon problem.
- Interested sellers hear the benefits and get an eligibility check.
- Hesitant sellers get their concern answered. Busy sellers get a callback time.
- Every call writes back interest, eligibility and next step.
- Hot leads go to the onboarding team. Callbacks go back into the queue.
The impact
- 400,000+
- AI calls to Amazon sellers
- About 200,000
- Sellers reached
- About 150,000
- Connected calls
- 12,000+
- Interested sellers qualified as hot leads
- 5,500+
- Hot leads with a regular GST number, ready to onboard
- 10,000+
- Callback times captured from busy sellers
- Under 2 minutes
- Typical length of a call that ends in a hot lead
The onboarding team works only on ready sellers. About 1 in 12 connected calls became a hot lead, and each one came with eligibility already checked.
The reach no human team could match. 200,000 sellers got a personal call in their own mix of Hindi and English.
Less manual calling. The team stopped dialling lists and started onboarding sellers.
The outcome
Heterize runs seller onboarding as a steady pipeline. The AI agent makes the first call, finds the problem, makes the pitch, checks eligibility and books callbacks. The onboarding team starts every day with a list of interested, eligible sellers instead of a raw phone list.
Lessons for e-commerce enablers and marketplaces
- Call the sellers who already started. Registered sellers convert to ready leads at twice the rate.
- Let a no end fast. Short calls with uninterested people keep large campaigns affordable.
- Treat every callback as a lead. Record the time and call back.
- Speak the seller's language. Natural Hinglish beats a formal script.
- Qualify before the hand-off. Ask the eligibility questions on the first call.
See more: voice AI for e-commerce, multilingual voice AI, voice AI for Indian languages, and OmniDimension for e-commerce.
FAQs
See what OmniDimension can do for your leads
Book a 15-minute walkthrough with our team.