A public registry gives you depth. Volume gives you coverage. If your outreach math says you need several hundred touches per closed deal (lesson 4.1.1), you need a list in the hundreds or low thousands, and you need it to cost very little per row so that most rows can turn out to be wrong without hurting you. This lesson is how to build that list in about a week of focused work, and the spreadsheet schema that keeps a thousand rows usable instead of a thousand rows of noise.
Principle
A list is cheap. A clean list is not. Spend your effort on the second.
The raw material for a volume list costs almost nothing per record. Public registries publish bulk files at no cost or a small fee. Directories and map listings show thousands of businesses by category and city. Commercial list services sell business records in bulk, and software that extracts listings from map and directory sites exists and is inexpensive per record. This textbook does not teach any particular tool: products change constantly, and many sites' terms of use restrict automated copying, so check the terms of any source before you take data from it in bulk.
What matters is the arithmetic, which is the same however you gather the rows. Work it with your own figures. As an illustration only: if a source costs $100 for 2,000 records, that is 5 cents a record. If one in ten survives cleaning and scoring, each usable row cost 50 cents. If your planning ratio says 100 touches yield about one deal worth pursuing, the data cost of that deal is trivial next to your time. Your time is the expensive input. Optimize for clean rows per hour, not cost per record.
Three rules follow:
- Stack several sources. No single source covers a market. Registries miss unlicensed and young businesses; directories miss businesses that never market themselves; map listings miss companies that serve other businesses. Four or more sources, merged, give you coverage.
- Deduplicate ruthlessly. Stacked sources overlap heavily. The same company appears under its legal name, its trade name, two phone numbers and a website with and without "www." Duplicates waste touches and embarrass you when one owner gets the same email twice.
- Score before you spend. Enrichment by hand and paid contact data cost time or money per row. Score rows on public clues first, and spend only on the ones that fit.
Case
Coastal Comfort HVAC is the fictional company used throughout this curriculum. Its row came out of a volume build like the one below, run for a Florida services buy box.
The buyer's sources: the state contractor licensing records for HVAC (public database), the secretary of state for entity details (public database), map and review listings for "air conditioning contractor" in each target city (volume), and the state chapter member list of an HVAC contractors' association (directory). After merging, the same companies appeared two, three, sometimes four times. Coastal Comfort appeared three times: as "Coastal Comfort HVAC, LLC" on the license and state filing, and as "Coastal Comfort Heating & Air" on its map listing, with a second phone number. Matching on the normalized website domain collapsed them into one row, keeping the license number from one source, the reviews from another, and both phone numbers.
The scoring pass used only public clues: years since formation, license age, single location, owner-operated signals, and a size estimate from fleet and headcount. Coastal Comfort scored in the top tier. That week's run returned 31 companies that fit the buy box; Coastal Comfort was one of them, and the only one that turned into a deal. That is not a disappointing yield. It is the outreach math working as planned.
Steps
The five-day method below is a schedule, not a promise. A first build commonly takes longer. The order is what matters.
Day 1: buy box, tracker, first source
- Translate your buy box into data filters. Industry codes (primary plus two or three adjacent), geography (counties, a metro area, or a radius), size proxies you can observe (headcount band, fleet size, license class, number of locations), inclusion signals (single location, ten or more years in business, family or founder name in the company name), and exclusions (franchises if outside your thesis, private-equity-owned, public, recently acquired).
- Build the tracker using the schema below.
- Pull your primary public database (lesson 4.2.1) into the tracker, filtered to the buy box.
Day 2: stack the sources
- Add at least three more sources: a second public database, a directory or association member list, and the volume layer (map and directory listings for your category in each target city).
- Tag every row with its source and the date.
- Aim for coverage, not perfection. You will clean tomorrow.
Day 3: deduplicate and normalize
Deduplication rules, applied in the spreadsheet:
- Normalize the website to a bare domain: strip "http://", "https://", "www.", trailing slashes and anything after the domain. Lowercase it.
- Normalize the phone to digits only, last ten digits.
- Normalize the address to street number, street name and ZIP code; treat "St" and "Street" as the same.
- Treat two rows as the same business if the domain, the phone, or the street-and-ZIP match.
- In each duplicate cluster, keep the richest row and copy over any field the others add (license number, second phone, reviews). Record every source in the source column.
Sort by each normalized column and read down the list; matches sit next to each other. A helper column that concatenates normalized domain and phone makes the matches easy to flag with a conditional format.
Day 4: score and tier on public clues
Score each row 1 to 5 on five criteria, using only what you can see:
| Criterion | What earns a 5 |
|---|---|
| Revenue fit | Size proxies land in your band |
| Geography fit | Inside your core area |
| Transition readiness | Long tenure, older license, no visible successor |
| Owner engagement | Owner visibly runs the business: answers reviews, named on the site |
| Deal simplicity | Single entity, single location, no obvious complications |
Total out of 25. A workable tiering: 20 and above is hot, 15 to 19 warm, 10 to 14 cool, below 10 pass. These cut-offs are this school's house rubric, not a validated model; adjust them once you see your own response data. Only hot and warm rows move on to enrichment (lesson 4.3.1).
Day 5: spot-check and set the rhythm
- Spot-check. Pick ten random hot and warm rows and verify each by hand: is the business open, does the license show active, is it really in your industry? If more than a couple fail, fix the filter that let them in before you go further.
- Set the refresh rhythm. Monthly: re-pull your sources, add new entrants, rerun the deduplication. Quarterly: re-check status and intelligence on the top tier. Continuously: note any press about ownership change, retirement or sale for top-tier targets.
The tracker schema
One sheet, one row per business, columns grouped by purpose:
| Group | Columns |
|---|---|
| Core | Company name, legal name, website (normalized domain), phone (normalized), address, city, state, ZIP, industry code, category |
| Provenance | Source 1, source 2, source 3, public record identifier (license, USDOT, NPI or entity number), date first found, date last verified |
| Size and fit | Size proxy used, size estimate band, years in business, locations |
| Contact | Owner first name, owner last name, title, email, email confidence, phone, phone confidence, waterfall layer that produced each |
| Intelligence | Founding year, license date, successor signal, family-owned signal, recent press, notes with links |
| Scoring | The five criterion scores, total, tier |
| Reach | First touch date, channel, last touch date, status, next action date, do-not-contact flag |
Two rules make the schema work. Every fact you add carries its source in the notes or a provenance column. And the do-not-contact flag is permanent: once an owner asks you to stop, that row never re-enters a sequence, through any channel, even after a refresh adds the business again from a new source. Check new rows against your do-not-contact list every time you merge.
Deliverable
A master tracker in a spreadsheet, with:
- Your buy-box filters written on a first tab.
- Rows from at least four sources, including at least one public database, each tagged with source and date.
- A completed deduplication pass, with the normalization rules applied.
- Every row scored and tiered, and the hot and warm rows marked for enrichment.
- A ten-row spot-check recorded, and the monthly and quarterly refresh dates on your calendar.
This lesson is education, not legal advice. Data sources carry terms of use, and outreach to the people on your list is regulated; review both with counsel before you contact owners at volume.