Maps Chain Location Consistency Audit

Audit the multi-location listing data you already have. Give the Actor a list of location records (for example an export from your own listing management, a spreadsheet converted to JSON, or an existing Apify dataset) and it reports where names, addresses, phone numbers, websites and opening hours disagree across the chain, and which records look like duplicates.

This Actor does not scrape Google Maps or any other website. It makes no requests to listing sites, uses no proxies and needs no third-party API. It only analyses the records you provide.

Who it is for

• Agencies and in-house teams that manage listings for brands with several to a few thousand locations.
• Anyone preparing a clean-up list before updating listings by hand or through an official listing tool.

What it checks

Area	Issue codes
Name	NAME_MISSING, NAME_DIFFERS_FROM_BRAND, NAME_FORMAT_VARIANT
Address	ADDRESS_MISSING, ADDRESS_SUFFIX_STYLE_INCONSISTENT, ADDRESS_CASE_INCONSISTENT, ADDRESS_NO_POSTAL_CODE
Phone	PHONE_MISSING, PHONE_FORMAT_INCONSISTENT
Website	WEBSITE_MISSING, WEBSITE_INVALID, WEBSITE_DOMAIN_MISMATCH, WEBSITE_INSECURE_SCHEME, WEBSITE_TRACKING_PARAMETERS, WEBSITE_LANDING_PAGE_INCONSISTENT
Hours	HOURS_MISSING, HOURS_UNPARSEABLE, HOURS_INCOMPLETE_WEEK, HOURS_TIME_FORMAT_INCONSISTENT, HOURS_DIFFER_FROM_CHAIN_PATTERN
Duplicates	POSSIBLE_DUPLICATE_LOCATION (with the related record index and the signals that matched)
Info	COORDINATES_MISSING, LISTING_MARKED_PERMANENTLY_CLOSED, LISTING_MARKED_TEMPORARILY_CLOSED, RECORD_UNUSABLE

Severities are high, medium, low and info. Each audited record gets a consistencyScore (100 minus 10 per high, 5 per medium and 2 per low issue, never below 0).

How "consistent" is decided

Most checks compare a record with the most common value in your data set (for example the most common website domain or hours format). A reference is only used when at least 3 records have a value and one value holds a strict majority. Otherwise that check is skipped and the run summary says so. You can also set a reference brand name yourself.

How duplicates are found

Two records are flagged when they share a place ID, or have the same normalized address and a similar name, or the same phone and the same address, or are within the chosen distance with a similar name or the same phone. A phone number shared by many different addresses (a central call-center number) is not treated as a duplicate signal on its own. Candidate pairs are found with indexes and a coordinate grid, so large inputs do not require comparing every pair.

Input

Copy the prefilled example in the Console, or use this shape:

{
  "locations": [
    {
      "title": "Harbor Bean Coffee - Central Station",
      "address": "120 Market Street, Springfield, IL 62701",
      "phone": "(217) 555-0101",
      "website": "https://www.harborbean.example/",
      "openingHours": "Mon-Fri: 7 AM - 7 PM; Sat-Sun: 8 AM - 5 PM",
      "location": { "lat": 39.8001, "lng": -89.6435 }
    },
    {
      "title": "HARBOR BEAN COFFEE",
      "address": "120 Market St, Springfield, IL 62701",
      "phone": "(217) 555-0101",
      "website": "https://harborbean.example/"
    }
  ],
  "brandName": "Harbor Bean Coffee",
  "duplicateDistanceMeters": 50,
  "nameSimilarityPercent": 85,
  "hoursPatternSeverity": "info"
}

Field	Description
locations	Array of location objects. Recognised keys: title/name, address (or street, city, state, postalCode), phone, website, openingHours/hours, placeId/id, location.lat/location.lng (or latitude/longitude), permanentlyClosed, temporarilyClosed.
sourceDatasetId	Optional ID or name of an existing Apify dataset in your account. Used only when locations is empty.
brandName	Optional reference name. Branch suffixes such as " - Downtown", "(Airport)" or "#12" are ignored when comparing names.
maxLocations	Default 2000, maximum 10000.
duplicateDistanceMeters	Default 50.
nameSimilarityPercent	Default 85.
hoursPatternSeverity	off, info (default), low or medium. Locations can legitimately have different hours, so the default does not reduce the score.

Hours can be a string ("Mon-Fri: 9 AM - 5 PM; Sat: closed"), an array of strings, an array of { "day", "hours" } objects, or an object keyed by day name.

Output

• Dataset: one item per input record with index, status, the key fields, consistencyScore, issueCount, highestSeverity, duplicateGroupId and an issues array. Each issue has code, severity, field, message and, where useful, observed, expected, relatedIndexes and signals.
• Key-value store, record SUMMARY: totals, issue counts per code, duplicate groups, the reference values that were derived from your data, and a list of caveats for this specific run.

Records with neither a name nor an address get status skipped and are not billed.

Pricing event

The code emits one pay-per-event event, location-audited, once per audited record. It is submitted only after all results have been stored, and never when the input is invalid or the run fails before results are written. No price is configured in this repository; the payable plan must be set up separately in the Apify Console. If a maximum charge limit is set on the run, the Actor audits only as many records as that limit allows and reports it in SUMMARY.

Limitations and sample-size caveats

• Results are heuristics over the data you supply. They are not checked against live listings, and the Actor cannot tell you which value is the correct one, only which values disagree.
• With fewer than 10 records, majority-based references are unreliable. With fewer than 3 records most chain-relative checks are skipped. If the majority value is itself wrong, the correct records are the ones flagged.
• Legitimate variation (sub-brands, shop-in-shop names, regional domains, seasonal or local hours) will be reported as an inconsistency. Review before changing anything.
• Address checks are text-based. The street-type style check looks at common English words (Street/St, Avenue/Ave, Road/Rd and similar); the postal-code check recognises common US, Japanese, Canadian and UK shapes and can miss other formats or mistake a five-digit street number for a postal code. There is no geocoding or address validation.
• Hours parsing handles common English notations (12-hour and 24-hour, day ranges, split shifts, "closed", "24 hours"). Other languages and layouts are reported as HOURS_UNPARSEABLE.
• Duplicate detection can flag two real shops that share a building or a very close address, and can miss duplicates with different names, addresses and coordinates. Coordinates close to the 180th meridian are not wrapped.
• Input is limited to 10000 records per run; records beyond maxLocations are ignored and the summary says so.
• No evidence is offered here about paid demand or accuracy on real-world data sets; validate on a sample of your own records first.

Privacy

Only the records you provide are processed. The Actor does not collect personal data, does not send data to third parties and does not contact any listing site. Do not include personal contact details in the input.