B2B services
Collecting and verifying a company database for B2B sales
A pipeline that gathers organisations by category and district, deduplicates records, checks whether their websites respond, and produces a report the sales team can open.
- Client
- In-house product
- Year
- 2026
- Services
3
Pipeline stages, each independently restartable
CSV + HTML
Export formats for the sales team
0
Progress lost when collection is interrupted
Challenge
Sales did not need "a list from a directory" but a callable database: no duplicates, current websites, and a clear flag for whether the organisation has a working site at all — which is precisely what separates a warm contact from a cold one.
A naive export by category produces something unusable: the same business appears under several categories, branches duplicate the head record, and a share of the websites stopped responding long ago.
Solution
We built a three-stage pipeline. The first stage collects by category and district, respecting the source's rate limits and surviving interruptions: progress is written line by line, so a rerun does not start over.
The second normalises — addresses and phone numbers to a single format, deduplication on a stable key rather than the name, and tagging of branches and multi-category records.
The third checks the websites: parallel requests with timeouts, response classification (working, redirect, unreachable, domain does not resolve), and export to CSV plus an HTML report a manager opens without installing anything.
A routine task for any B2B sales team: turning a directory into a working contact list. The value here is created not by collecting the data but by cleaning and verifying it.
Stack
- Python
- asyncio
- httpx
- pandas
- JSONL
DEKART — DIGITAL
Tell us about the task
We reply within 15 minutes during business hours. The first consultation and estimate are free.