A lead list is only as good as its weakest row. One wrong domain, one role address, one bounce and your sending reputation takes a hit that can take weeks to repair. The good news is that a clean list is not an accident — it is a process, and the process is repeatable.

Here is the exact workflow I use for every client list, whether it is a few hundred contacts or several thousand.

1. Define the buyer before you collect any data

Most bad lists are built backwards. Someone opens a database, filters by industry, exports twenty thousand rows and then wonders why the reply rate is under one percent.

Start with a short written definition: the job titles that make the decision, the company size band, the industry and the geography. Add a disqualifier list too. If you cannot explain who should not be on the list, your filters are too broad and you are paying for noise.

  • Primary titles plus the ones you would accept
  • Company size by headcount, not by revenue alone
  • Countries and languages you can actually support
  • Trigger events: funding, hiring, expansion, new regulation

2. Collect from more than one source

No single database has every field. Some are strong on company data, others on people, others on contact details. The skill is in combining them.

My usual pattern is: one source for the company identity and website, a second for the person and their role, and a manual check on the company site to confirm the naming pattern. Where two independent sources agree, confidence is high. Where they disagree, the row gets verified by hand or dropped.

3. Verify every address

Verification happens in three layers. Syntax catches typos and pasted text. A domain and MX check removes dead companies. The mailbox probe is the strongest signal, and it is the slowest, so it runs last.

Anything that cannot be verified is flagged in its own column rather than silently included. Clients appreciate that honesty, and it protects the sending domain.

4. Clean, standardise and de-duplicate

  • Remove duplicates by domain first, then by person
  • Normalise country names, phone formats and capitalisation
  • Split full names into first and last
  • Map job titles into a consistent seniority field

Deduplicating by name alone produces false merges. Always match on the strongest available key.

5. Deliver in a format the team can use immediately

Google Sheets or Excel with frozen headers, one row per decision maker, no merged cells and no colour coding used as data. If the client uses a CRM, match the import columns exactly so nobody has to remap anything.

The result

A two thousand row list that took an extra day to verify will outperform a twenty thousand row list that bounces. Accuracy compounds: every clean campaign makes the next one easier, because the sending reputation is an asset you build rather than repair.

Common mistakes that undo a good list

Even teams that follow the workflow above lose rows to a handful of recurring errors. Catching them early is cheaper than repairing a campaign.

  • Buying on row count. A cheap file with a low verification rate costs more in damaged reputation than it saves in licence fees.
  • Trusting one source. Any single database goes stale quietly, and you discover it only when replies collapse.
  • Skipping the sample. Ten rows reviewed by hand reveal the real quality of a supplier before you commit.
  • Keeping unverified rows in the main send. Move them to a separate segment and test in small batches.
  • Forgetting consent and local rules. B2B outreach is regulated, and an honest opt out protects everyone.

After the list is built, the work is not finished. Normalising fields, deduplicating properly and packaging the file so a colleague can import it without remapping is covered in turning scraped data into a clean file.

Treat verification as a process rather than a one-time event. A list checked in January will have drifted by spring, so re-check it before any large send and keep the bounce rate low. The mechanics of each check are explained in how to verify email addresses.

If you would rather not build this process yourself, start with a free sample of ten rows for your exact target market and judge the quality before you commit.