Business complaint pattern analysis

Access and scope

Set CRAWLORA_API_KEY to your Crawlora key. Prefer the listed Crawlora MCP tools when connected; otherwise use the bundled scripts/crawlora.sh, which sends x-api-key to https://api.crawlora.net/api/v1. Read reference/endpoints.md for exact methods and parameters. Check application code as well as HTTP status; payloads are inside data. Stop on authentication errors, back off on 429, and retry a transient upstream failure once. Bound calls/pages by the requested scope and credit budget. Retain source IDs, URLs, source dates, and observation/crawl times separately.

Compare reported complaint themes or response patterns for selected businesses. Establish business/location scope, reporting/date window, and comparable complaint categories. Keep complainant allegations, business responses, customer replies, and BBB's presented status as distinct attributed evidence.

Resolve the entity and retrieve complaint evidence

1. Use BBB dataset facets/search or live search to identify exact business profiles. Match business ID, name/alternate names, location, website, and HQ relationships. A multi-location HQ profile can aggregate branches; do not assign its count to each location or combine it with branch counts as independent complaints.
2. Stored BBB profiles do not embed full complaints. Use the returned profile URL or verified HQ URL with live business/complaints/more-info routes, not a guessed complaint-page slug. Preserve crawl time versus live collection time.
3. total_complaints has its own reporting window, typically the presented three-year window; closed_last_12_months is a different measure. Retain the returned period_text rather than inventing a common time span across firms. There is no date filter/pagination parameter on the complaints route: filter returned itemized dates locally and mark bounded coverage.
4. complaints_list_incomplete=true means real aggregate totals can accompany unavailable itemized detail. Do not substitute an empty list for zero complaints or fabricate theme/status counts. A genuine total_complaints=0 can be normal, but it only describes that source/window, not all customers' experience.
5. Preserve complaint ID, date/type/status, narrative and response-thread roles. BBB statuses include Resolved, Unresolved, Answered, Unanswered, Unpursuable; an answer is not automatically a resolution or admission. Keep BBB's REMOVED redactions intact rather than reconstructing identities.

scripts/crawlora.sh /datasets/bbb-businesses/facets facet=category
scripts/crawlora.sh /bbb/search query="plumbing" location="Austin, TX" page=1
# Use a verified returned profile URL, for example:
# scripts/crawlora.sh /bbb/business/complaints url="$BBB_PROFILE_URL"
# scripts/crawlora.sh /bbb/business/more-info url="$BBB_PROFILE_URL"

Report patterns with defensible denominators

Deduplicate by complaint ID; multiple response rounds are one complaint with several observations. Code themes with uncertain/multi-label states and cite short attributed excerpts. Show itemized sample size, aggregate totals, missing lists, and date filters separately. A sample resolution/response proportion uses that sample's comparable statuses, not the total number of customers or a mismatched aggregate window. Response intervals need actual role/date pairs; a rounded date is not an exact elapsed-hour measurement.

Return identity/coverage, complaint and response evidence tables, theme/status counts, source reasons-for-rating where relevant, and unresolved questions. The main profile's reasons list can be shallow boilerplate; more-info supplies the fuller per-factor list. Reviews, accreditation, and letter grades are separate measures. Counts without comparable customer/transaction volume cannot rank a business's complaint rate or prove fraud, quality, liability, or consumer outcomes. Do not file complaints, contact businesses, publish accusations, or initiate a legal action from a research request.