Map Extract

To deduplicate business leads, normalise first, then match on keys in order of trust: the map platform's place ID, the company registration number, the website domain, the phone number in +44 form, and only then a fuzzy match on name within the same postcode. Decide up front whether branches of a chain are one lead or several, merge matched rows by clear rules that keep every source, and check a sample by hand before trusting the result.

This step comes after normalising each field, which is covered in the B2B lead list cleaning checklist. Lists built from Google Maps need it most, because splitting searches to get past the result cap produces heavy overlap; see Google Maps only shows 120 results.

Step 1: decide what one record means

Before matching anything, answer one question: are you selling to the business or to each location?

  • One row per business. Right for software, accountancy, insurance or anything bought centrally. Five branches of "Aire Valley Vets" become one record with five locations.
  • One row per location. Right for services delivered on site, such as window cleaning, waste collection or equipment servicing, where each branch manager may buy separately.

The answer changes which keys can merge rows. A shared domain merges branches under the first model and must not under the second.

Step 2: normalise the key fields

Matching fails on formatting differences, so clean first:

FieldNormalise toExample
WebsiteRegistrable domain, lowercasehttps://www.AireValleyVets.example/leeds becomes airevalleyvets.example
PhoneE.1640113 496 0000 becomes +441134960000
PostcodeUppercase, one spacels61aa becomes LS6 1AA
NameLowercase match key, punctuation removed, & to and, legal suffixes strippedSmith & Sons Plumbing Ltd. becomes smith and sons plumbing
Company numberEight characters, leading zeros kept1234567 becomes 01234567

E.164, the ITU's international numbering plan, gives every phone one canonical form, which is why it makes a good key. Keep the original values in separate columns; the match key is for matching, not for display. Name cleaning in more depth is in cleaning a B2B lead list, and the free phone number formatter converts a phone column to E.164 in one go.

Step 3: match on keys in order of trust

OrderKeyStrong becauseWrong merge riskMissed match risk
1Platform ID (Google place ID or CID, Bing or Apple listing ID)Assigned to one listingVery lowSame place can have several IDs; only works within one platform
2Company numberOne per legal entityOne company may trade several brands or sitesMany small firms are sole traders with no number
3Registrable domainUsually one per businessBranches and franchisees share it; platform domainsBusiness changed domain
4Phone (E.164)Usually one per business or siteCall centres, shared switchboards, tracking numbersDifferent numbers on different listings
5Email addressSpecificAgency or platform addressesOften missing
6Fuzzy name plus postcodeWorks when nothing else existsSimilar names on one streetRebrands, abbreviations

Two notes on the top of the table.

Place IDs. Google says place IDs "uniquely identify a place", but also that "it is possible for the same place or location to have multiple different place IDs", that they "may change over time", and that it recommends refreshing IDs "more than 12 months old" (place ID documentation). Use place ID to collapse repeats from overlapping searches, then let the later keys catch what it misses.

Domains to exclude from matching. Never match on facebook.com, instagram.com, booking platforms, marketplaces, directory sites or website builder hosts. Hundreds of unrelated businesses share them. Keep a list of these domains and skip them in the domain step.

Compare within blocks

Comparing every row with every other row is slow and produces false matches. Compare only within a block: the same postcode district for name matching, the same domain for branch checks, the same phone for phone matching. On a list of 5,000 rows, blocking by postcode district turns millions of comparisons into a few thousand.

Fuzzy name rules

Use fuzzy matching only for rows that no exact key has matched, and only within a block.

Build the match key:

  1. Lowercase.
  2. Replace & and + with and.
  3. Remove punctuation.
  4. Strip legal suffixes: ltd, limited, llp, plc, cic, & co.
  5. Strip location words that you also hold in the address, such as the town name.
  6. Sort the words alphabetically, so "Plumbing Smith and Sons" matches "Smith and Sons Plumbing".

Then compare keys with a string similarity measure (token-based measures cope well with reordered words) and set a threshold by testing it on your own data.

Name AName BSame postcode?DecisionWhy
Smith & Sons Plumbing LtdSmith and Sons PlumbingYesMergeSame key after cleaning
Bright Smile Dental LeedsBright Smile DentalYesMergeTown word stripped
Bright Smile Dental LeedsBright Smile Dental BradfordNoDo not mergeDifferent locations; possibly branches
A1 TaxisA1 CarsYesReviewCould be one firm or two rivals
The Coffee HouseCoffee House KitchenYesReviewGeneric words; similarity is misleading
Kirkgate PlumbingKirkgate Plumbing & HeatingYes, same phoneMergePhone confirms

Generic names ("The Coffee House", "Express Cleaning") are the main source of wrong merges. Require a second piece of evidence, such as the phone or the street, before merging them.

Branches and chains

Google's guidelines say "there should only be one profile per business", and in practice a chain has one listing per branch, each with its own address and usually its own phone. Signs you are looking at branches rather than duplicates:

  • Same domain, different addresses and different phone numbers.
  • Same name with a town or district suffix.
  • The website has a "locations" or "find a branch" page.

Franchises are a special case: each franchisee is often a separate legal entity with its own company number, trading under a shared brand domain. For outreach purposes they are separate businesses, even though the domain says otherwise. A company number found in each franchisee's page footer settles it.

A practical structure keeps both views: a business record keyed on domain or company number, and location records linked to it, each with its own place ID, address and phone.

Merge rules

When rows match, decide field by field which value survives.

FieldPrefer
Legal name and company numberThe company register
Trading nameThe business's own website, then the map listing
EmailThe business's own website, verified
PhoneThe most recent source; keep the others as alternates
Opening hours and statusThe most recent map listing
AddressThe map listing for trading address; the register only for registered office
CategoryYour own segment mapping, not any single source

And three rules that are not negotiable:

  1. Keep every source. Store all place IDs, listing URLs and website URLs on the merged record. They are your evidence and your dedupe keys for the next run.
  2. Suppression wins. If any of the merged rows has opted out, complained or is on a do-not-contact list, the merged record is suppressed. Never let a merge resurrect an opt-out.
  3. Keep conflicts visible. If two sources give different emails or phones, keep both and mark the conflict for review rather than silently picking one.

Merging lists from different tools and spreadsheets follows the same rules; to see the overlap between two lists first, use the free merge and compare lists tool.

A worked example (fictional)

Six rows from overlapping Google and Bing Maps searches for vets in Leeds:

RowSourceNameDomainPhonePostcodePlace ID
1Google, "vet in LS6"Aire Valley Vets Headingleyairevalleyvets.example+441134960101LS6 1AAChIJ_example_A
2Google, "veterinary surgery in LS6"Aire Valley Vets Headingleyairevalleyvets.example+441134960101LS6 1AAChIJ_example_A
3Bing, "vet in Leeds"Aire Valley Veterinary Surgeryairevalleyvets.example+441134960101LS6 1AA(Bing ID)
4Google, "vet in LS17"Aire Valley Vets Alwoodleyairevalleyvets.example+441134960202LS17 5AAChIJ_example_B
5Google, "vet in LS1"City Paws Vetsfacebook.com+441134960303LS1 6AAChIJ_example_C
6Bing, "vet in Leeds"City Paws Veterinary Clinic+441134960303LS1 6AA(Bing ID)
  • Rows 1 and 2 merge on place ID.
  • Row 3 merges with them on domain plus phone plus postcode (a different platform, so no shared ID).
  • Row 4 shares the domain but has a different address and phone: a branch. With one row per business it joins the Aire Valley record as a second location; with one row per location it stays separate, linked by domain.
  • Rows 5 and 6 cannot match on domain (facebook.com is excluded), but merge on phone and postcode, confirmed by the fuzzy name match.

Six rows became two businesses with three locations.

Check the result

  • Count rows before and after, and the share merged by each key.
  • Review 20 to 50 random merges by hand. Wrong merges are worse than missed ones: they send a message meant for one business to another.
  • Review near misses just below your fuzzy threshold.
  • Confirm every suppressed row is still suppressed after merging.

Excel's Remove Duplicates can handle the exact-key steps on a clean column, but it deletes later rows and keeps the first, so it loses information rather than merging it. Keep a copy before using it.

How Map Extract helps

  • Each listing keeps its platform identifiers: Google place ID and CID, and Bing and Apple source IDs, along with the keyword and location that found it.
  • Bing result pages are re-read after each page turn so recycled result cards are not counted twice, and Bing flags listings it identifies as chains.
  • The website data extractor reduces each site to a canonical domain and extracts phone numbers, company numbers and VAT numbers where the site shows them, which give stronger keys than names.
  • When matching a website to a Maps listing, a listing is accepted only if its website domain matches, with phone, name, address and shared social profiles adding evidence, and the reason for the match is stored.
  • Observations from every source are consolidated into master records, and the workflow builder saves the sources you run so the next collection is matched against the same records.

What it does not do: it cannot decide for you whether branches are one lead or several, and automatic matching can still be wrong for generic names; review held records before export.

Public business and social results the app collects are uploaded automatically to Map Extract's servers while it is licensed and running; the privacy notice explains this.

Key takeaways

  • Decide business or location before you match.
  • Normalise domain, phone, postcode and name first.
  • Match on place ID, company number, domain, phone, then fuzzy name within the postcode.
  • Exclude social, booking and directory domains from domain matching.
  • Keep every source on the merged record, and let suppression always win.

Frequently asked questions

What is the best way to remove duplicate businesses from a lead list?

Normalise the fields first, then match on keys in order of reliability: platform place ID, company registration number, website domain, phone number in +44 form, and finally a fuzzy match on business name within the same postcode. Merge matches into one record, keep every source link, and review a sample of merges by hand to catch branches or different businesses that were wrongly combined.

Should I deduplicate leads by website domain?

Domain is one of the strongest keys for independent businesses, because each usually has its own site. It fails in two cases: branches of a chain share one domain, and some listings use a social page, booking platform or website builder address instead of their own domain. Exclude shared platform domains from matching and decide whether branches should stay separate before you merge.

How do I handle branches of the same company in a lead list?

Decide whether you sell to the business or to each location. If you sell to head office, merge branches into one business record and keep the locations as a list. If each branch buys separately, keep one row per location, linked by a shared business key such as the domain or company number. Either way, do not let a fuzzy name match merge two branches by accident.

Does Excel's Remove Duplicates work for lead lists?

Only for exact matches on clean keys. Remove Duplicates deletes every later row that matches on the columns you choose and keeps the first, so it cannot merge information from two rows or catch near matches such as 'Smith & Sons Ltd' and 'Smith and Sons'. Use it after normalising, on a key column such as the domain, and keep a copy of the file before running it.

Sources

  1. Place IDs, Google Maps Platform documentation
  2. Guidelines for representing your business on Google, Google Business Profile Help
  3. ITU-T Recommendation E.164

Written by the Map Extract editorial team. Found an error? Email support@mapextract.uk and we will check and correct it. How we research guides: editorial policy.

See it working on your own searches.

Start with one keyword, one location and a small limit. Review the records, then decide on a licence.

Top