Merge & Dedupe Real-Estate Leads for Your CRM

Follow Up Boss uses lowercase column names. This is the exact map the blueprint forks with at Step 8.

{
  "First Name": "first name",
  "Last Name": "last name",
  "Email": "email",
  "Phone": "phone",
  "Source": "source"
}

Sierra auto-maps these exact title-case field names. Feed your normalised lowercase headers as keys, Sierra's labels as values.

{
  "first name": "First Name",
  "last name": "Last Name",
  "email": "Email",
  "phone": "Phone",
  "source": "Lead Source",
  "type": "Lead Type"
}

Top Producer expects three distinct phone columns. Split phone first with csv-column-value-splitter, then map. See /tool/csv-column-value-splitter.

{
  "first name": "First Name",
  "last name": "Last Name",
  "email": "Email",
  "phone_cell": "Cell Phone",
  "phone_home": "Home Phone",
  "source": "Source"
}

Because the executor ignores useRegex, run csv-find-replace as a chain of literal removals instead of one bracket class. Each pass strips one character.

// Step 4a  find: "("   replace: ""  useRegex: false
// Step 4b  find: ")"   replace: ""  useRegex: false
// Step 4c  find: "-"   replace: ""  useRegex: false
// Step 4d  find: "."   replace: ""  useRegex: false
// Step 4e  find: " "   replace: ""  useRegex: false
// Result: (555) 555-1234  ->  5555551234

Step 7 dedupes on one column only. Run it twice — once keyed on Email, then re-run the output keyed on Phone — to catch rows that share a phone but differ on email.

// Pass 1  csv-deduplicator  columns: "email"
//   collapses leads that share an email address
// Pass 2  csv-deduplicator  columns: "phone"
//   collapses remaining leads that share a phone number
// (both passes keep the FIRST occurrence, case-insensitive)

The blueprint sets columns: "Email,Phone", but the deduplicator runtime reads only the FIRST listed column as the key and ignores strategy. So it dedupes on Email alone, case-insensitively, keeping the first row. Two leads with the same phone but different emails survive as separate rows. Workaround: run `csv-deduplicator` twice — first keyed on email, then re-run the output keyed on phone.

Both find/replace nodes set useRegex: true, but the workflow executor passes only caseSensitive to the engine — useRegex is dropped, so the engine escapes the pattern and matches it literally. [()\s\-\.] therefore matches the literal string, not a character class, and nothing changes. Fix: replace each regex node with literal-value passes (one char per pass), or pre-process in the standalone `csv-find-replace` tool, which honours useRegex.

If two rows match on the dedup key and only one carries a Do-Not-Contact flag, Step 7 keeps whichever appears first — it has no strictest-merge logic. Subsequent outreach is a TCPA violation: $500 per call/text, trebled to $1,500 for willful. **Manually verify**: filter opted-out rows with `csv-column-filter` before Step 7 and merge them back after, keeping the opt-out version on conflict.

Step 3 keeps only letters, digits, and spaces — so it strips . and @ from EVERY cell, including emails, turning jo@x.com into joxcom. The blueprint runs it before any email-keyed step, which is why dedup keys on lowercased emails that may already be mangled. If you need intact emails for matching, exclude email columns from Step 3 or move name-cleaning to a name-only column subset.

`csv-merger` throws CSV Merge requires at least one file input when nothing is on the node. A single file round-trips fine (union of one), but you get no dedup benefit — the point of the workflow is multiple sources. Drop every export at once.

This blueprint ships scheduleCron: null, and the fork copies that value, so a fresh fork never fires automatically. To automate (e.g. nightly lead merges), set a schedule_cron expression on your draft. The Cloudflare cron tick then enqueues a source: cron run via workflow_run_queue when isDueSince says it's due.

The from-blueprint route returns 401 unauthorized if it can't resolve a session email or access-token subject. Forking needs auth (any tier); running the resulting workflow needs only the free tier since all 7 nodes are free.

Follow Up Boss caps imports at 90 MB. Split the workflow output by date range or lead source using `csv-row-limiter` or `csv-row-splitter` before upload.

Sierra's Import Duplicates mode fills blank fields on existing leads but does NOT reassign agent ownership even if your CSV lists a different Assigned Agent. Bulk reassignment requires the Sierra UI or API — the workflow can't force it via headers.

555 1234567 could be Canadian, Australian, UK, or a US number missing the area code. Step 4 (once fixed to strip separators) reduces it to digits but can't infer the country. Manually inspect rows whose phone length falls outside 7–11 digits before trusting them as a dedup key.