A meat wholesaler operating in the UK automated its B2B lead generation with two AI systems: a voice and written assistant that takes orders over WhatsApp/phone, and an automation that scans official food-registry and map data to find potential restaurant/business customers in a region. A single regional scan (Manchester, restaurant category) surfaced 60 potential customer records; 97% had a phone number and 100% had a website — all from real sources, with no fabricated contact details. Below is the setup and three measured figures.
The problem: manual lead generation and scattered order channels
The wholesaler had two separate bottlenecks: (1) finding new customers was entirely manual — reaching restaurants and businesses in a region one by one took hours; (2) orders from existing customers arrived scattered across WhatsApp, phone, and email, making order tracking harder. Both problems came from the same root: growth and operations couldn't scale together without adding headcount.
The setup: two automations, one operation
The system built has two parts. First, a voice and written order assistant — connected to the existing phone/WhatsApp line, taking orders and forwarding them into the production and invoicing process. Second, an automated lead-generation system — it scans official food-registry data (registered business records) and map data (by region and category) to collect contact details (phone, website, address) for potential B2B customers; a new region/category scan can be started from the panel with one click. Both systems were built on top of the existing line and processes — neither replaced anything from scratch.
Measured result 1: 60 potential customers from one scan, 97% with full contact info
A single panel scan in the Manchester area, restaurant category, updated or added 60 business records; 58 of them (97%) had a phone number, and all 60 (100%) had a website. For the 2 businesses without a phone number, the source data simply didn't have one recorded — no value was ever fabricated; missing data was left missing.
Measured result 2: the same scan, at low cost
The 60-result scan completed in just 3 API requests (20 results per page) — at a cost of roughly $0.10. A scan at this scale, per region/category, takes minutes rather than the hours manual research would need.
Measured result 3: the order assistant's response speed
Across 19 real conversations with the order assistant, 17 were answered in 1.5-6.7 seconds; the line handles multiple simultaneous requests without making anyone wait in a queue.
Next step
To discuss how a voice/written order assistant and automated lead generation could be set up for your business, visit the voice AI assistant service page, request a short discovery call through our contact form, or call +90 850 840 08 23 and talk live with Azra, our voice assistant.
Frequently asked questions
Where do the numbers in this case study come from?
From live measurements in the project record: the panel scan's own result count, the phone/website completeness rate already present in the source data, and the order assistant's real conversation durations. No figure was estimated.
What sources does the lead-generation system pull from?
Public, official records such as food-registry data and map/business data; information not present in the source (like a phone number) is never fabricated — it's left blank.
Which channels does the order assistant work through?
The existing phone line and WhatsApp; orders are forwarded into the production/invoicing process in structured form, with no number change required.
Can this setup be adapted to another industry?
Yes, the lead-generation side adapts to any region/category-based B2B sector (wholesale, supply, service businesses); the order-assistant side is similarly reconfigured for different order flows.