How to Build a Prospect Database From Scratch

Most prospect databases die within a month of being created. Someone exports a few hundred rows, blasts the first fifty, and the sheet quietly turns into a graveyard of stale emails and forgotten follow-ups. The difference between that and a database that feeds your pipeline every week is not the number of rows. It is structure: which fields you track, how honestly the statuses reflect reality, and whether every active row has a next action with a date on it.

This guide walks through building one from scratch — the exact column layout, the statuses worth using, where the data comes from, and the maintenance routine that keeps the whole thing alive.

Start with a narrow definition, not a big list

Before you collect a single contact, write down who belongs in the database. "Companies that might buy" is not a definition. "Dental clinics in Lisbon with an active website and their own brand, not franchises" is. A tight definition does two things: it keeps junk out of the sheet, and it makes every later step easier — messaging, prioritising, follow-up — because the rows resemble each other.

Pick one niche and one city or country for the first build. You can clone the structure for a second segment later. Mixing segments in a single sheet is the fastest way to make your filters useless and your reply rates unreadable.

The column layout that actually works

You do not need forty columns. You need about eighteen, grouped so your eye can scan a row in two seconds. Here is a layout that has survived real outreach campaigns.

Identity: who they are

• Company name — as written on their website, not the registered legal entity.
• Niche — one short label, from a fixed list you define. Free-text categories rot fast.
• City and Country — two separate columns. You will filter by both, separately.
• Website — full URL. Empty is fine; a guessed URL is not.

Contacts: how to reach them

• Email — the best one you have, one per row. Park extras in the notes.
• Phone — international format with country code, always. Mixed formats break deduplication.
• WhatsApp — yes/no/unknown. A phone number is not proof of a WhatsApp account; verify before you count on the channel.
• Telegram, Instagram, Facebook, LinkedIn — one column each. For local businesses, Instagram often answers when email never will.

Workflow: what happens next

• Status — a single value from the fixed list below.
• Channel used — where the last touch happened.
• First contacted and Last touch — dates, filled the moment you send anything.
• Next follow-up — the date you will act again. The most important column in the sheet.
• Notes — short, dated, newest on top: "12 Jun — asked for pricing, decision after summer."

Optional: priority

A simple A/B/C column beats any elaborate scoring model at this scale. A = perfect fit and reachable, B = fit with gaps, C = long shot. Sort by it every morning and work the A rows first.

Statuses that mirror reality

Statuses fail when they describe hopes instead of events. Use a short list where each value is triggered by something that observably happened:

1. New — imported, nobody has looked at it.
2. Qualified — a human checked the row against your definition and kept it.
3. Contacted — first message sent. Sent, not drafted.
4. Replied — any answer, even "not now".
5. In conversation — an actual exchange about their needs.
6. Won — they became a customer.
7. Lost — a clear no, with a one-word reason: price, timing, competitor, silence.
8. Not a fit — your mistake at import, not their rejection. Track it separately; it measures the quality of your sourcing.

Two rules keep this honest. One status per row, always. And no status ever moves backwards without a note explaining why.

Follow-up dates: the column that pays for everything

Most replies do not come from the first message. They come from the second and third touch, sent at a sane interval to people who did not answer the first. That entire mechanism lives in a single column: Next follow-up.

The rule: any row in an active status — Contacted, Replied, In conversation — must have a date in that column. No exceptions. If there is no next step, the row belongs in Lost or Not a fit, not in limbo.

A cadence that works for cold B2B outreach: first follow-up 3 days after the initial message, second after 7 more, a final one after another 14. Three attempts, then out. Each morning, filter the sheet by Next follow-up ≤ today and work only that view. This one habit is the difference between a database and a to-do list you are afraid to open.

Where the data comes from

Good rows come from combining sources, because no single one holds everything:

• Map platforms — the widest coverage of local businesses: name, address, phone, site.
• Business registries — confirmation a company legally exists and is active.
• Web search — the company site itself, where the real email and social links live.
• Social profiles — often fresher than the website, sometimes the only channel that gets answered.

Collecting this by hand runs three to five minutes per complete row — for two hundred rows, that is most of your working week, and phone formats and duplicates will still come out inconsistent. This step is exactly what is worth automating. JustLeadIt finds companies by niche and city across maps, business registries and web search, collects the public emails, phones, Telegram, Instagram, Facebook and LinkedIn profiles in one pass, and checks which phone numbers actually have WhatsApp. The export lands in XLSX or CSV, so it maps straight onto the column layout above. New accounts get two free searches on JustLeadIt — enough to build your first hundred rows in minutes instead of days and see whether your niche definition holds up.

Verify before you write to anyone

An unverified database punishes you twice: bounced emails damage sender reputation and push the next campaign into spam, and messages sent to phone numbers without WhatsApp simply evaporate while your sheet claims "contacted".

Before the first outreach pass:

• Drop rows with no working contact channel at all.
• Deduplicate by phone and by website domain — the same business often appears under two names.
• Mark which numbers really have WhatsApp instead of assuming.
• Spot-check ten random rows manually. If three are wrong, your source or your definition needs fixing before you scale, not after.

Keeping the database alive

B2B contact data decays at roughly 25–30% a year. People change roles, businesses close and move, phone numbers rotate. A database left alone for six months is half fiction. The fix is not heroic cleanups; it is a small routine:

• Daily — work the follow-up view. Log every touch the moment it happens; memory-based logging is fiction by Friday.
• Weekly — clear rows stuck in New longer than a week: qualify or delete. A backlog of unqualified rows creates the illusion of a full pipeline.
• Monthly — recheck bounces, close out third-follow-up silences as Lost, and run a fresh search on your segment to catch newly opened businesses. New openings are usually the most responsive rows in the sheet.
• Quarterly — archive Lost and Not a fit older than six months into a separate tab. Keep the working sheet under a thousand rows; beyond that, filters and attention both degrade.

A realistic first-week plan

1. Day 1 — write the niche definition, build the sheet with the columns above, fix the status list.
2. Day 2 — collect 100–200 rows for one niche and one city.
3. Day 3 — verify, deduplicate, assign A/B/C, qualify everything out of New.
4. Days 4–5 — first outreach pass across the A rows, statuses and follow-up dates recorded as you go.
5. Week 2 — work the follow-up view daily. This is where the replies actually arrive.

A prospect database is not an asset you build once; it is a habit with a spreadsheet attached. Get the columns right, keep every active row dated, and feed it fresh rows monthly — and it will quietly out-produce any list you could ever buy.