Sign in to Hub Book a demo
Legal · Stowe Family Law

Every review from three platforms, credited to the right lawyer and office every hour, without the spreadsheet.

A member of staff kept every client review from three platforms in a hand-maintained spreadsheet, guessing which lawyer and office each one belonged to.

Talk to us about something similarAll case studies
ClientStowe Family Law
SectorLegal
WhenLive
Used byOperations staff working the queue, the firm's BI team reading a SQL view, Prolink admins granting access
Plugs intoTrustpilot, Google via Uberall, Review Solicitors, Power BI, Azure Data Factory, Prolink Hub
AINone. This one runs on rules, not models.
What we built

Collect, match, attribute, confirm. People only touch the ambiguous ones.

An hourly collector pulling new reviews from Trustpilot, Google and Review Solicitors, de-duplicated by platform ID with the original response retained.
A seven-strategy name matcher that scores each candidate by confidence and credits every person a review thanks.
Office, sub-region and region attribution that falls back from the matched lawyer to the Google listing to an office name spotted in the text.
A Review Queue where unresolved reviews are assigned, pinned to an office, confirmed or dismissed. Manual decisions always win.
A per-review timeline of collection, matching, assignment, confirmation, dismissal and restoration.
A read-only reporting view with a stable merge key, so the firm's Power BI pipeline reads a contract rather than the app's tables.
matching

Seven ways to find a name

Exact full name, nickname ('Liz' finds Elizabeth), initials, fuzzy misspellings, and a unique first name when only one Yvonne works at the firm.

attribution

Three-level office fallback

A review counts in regional reporting before any lawyer is identified.

reporting

The view as a contract

A stable row key and last-modified column for incremental BI merge. Raw payloads deliberately kept out.

How it plugs in

Three review platforms in, one clean table out.

Reviews are matched against the live employee list synced daily from the firm's Azure warehouse. The BI team never logs in; they read the view.

The applicationTrustpilotGoogle via UberallReview SolicitorsPower BIAzure Data Factory
job Collection cron at ten past every hour, all three platforms, then the matcher, every run logged with counts and errors
job Automatic flagging (no match, low confidence, several candidates, unknown location) pushes work into the queue
job On-demand historical backfill and a re-run for flagged reviews that resolves itself as the matching improves
What changed

The spreadsheet is gone. Reviews are collected, credited and region-stamped automatically every hour, and human effort goes only to the genuinely ambiguous cases.

one detail only this project hasMatching is deliberately algorithmic rather than AI: exact text, initials, nickname tables and edit distance. It can be explained to an auditor line by line.
Figures from the project record: the matcher scans against about 526 current fee earners; five help articles are published to the Hub wiki
Something similar?

Which spreadsheet does someone at your firm update by hand every week?

Book a walkthroughRequest a callback