# Identity resolution data fit scorecard | SourceX

Canonical: https://refer.sourcex.si/questions/does-cross-tool-identity-resolution-data-qualify
Last updated: 2026-11-04

Score the company on three gates and seven criteria. The gates are 50+ full-time employees at peak, rights to license the records and an authorized contact. The criteria test whether its systems are actually linked: operating history, cross-reference fields, kept migration maps, merge logs with retired IDs, a shared strong key, recorded link decisions and downstream confirmation. A strong fit stores another system's ID in its CRM, kept the old-to-new mapping when it replaced a tool, and logs every merge. Ask; never request the files.

This scorecard is part of the guide to [cross-tool identity resolution](/guides/concept-cross-tool-identity-resolution), which follows one hospital customer across four systems. Here the question is narrower: does a particular company hold linked records worth an introduction, and how strong is the fit?

## Scoring rules and bands

Each criterion earns 2 points for a strong answer, 1 for borderline and 0 for weak. Seven criteria make a maximum of 14. Gates are pass or fail and are not added in.

| Score | Band | What to do |
|---|---|---|
| 12 to 14 | Strong | Introduce; the company likely holds links with their basis recorded |
| 7 to 11 | Borderline | Introduce with the weak criteria named; review will size the usable share |
| 0 to 6 | Weak | Links are mostly guesswork or lost; clear up the open questions first |

Why the basis of a link matters so much: methods differ. Salesforce's documentation describes exact matching, which treats non-identical strings as non-matches, alongside fuzzy methods such as edit distance and Jaro-Winkler. The Fellegi-Sunter model, documented by the UK Ministry of Justice's Splink project, scores each pair from per-field agreement weights. A link made by an exact tax ID and one made by a weak name score are very different evidence, and the scorecard rewards companies that recorded which was which.

## Three gates before any points

| Gate | Pass | Fail |
|---|---|---|
| Size | 50+ full-time employees at peak, contractors excluded | Under 50, or mostly contractors |
| Rights to license | The company controls its CRM, ERP and help desk records, and customer or vendor terms allow reuse discussions | A parent group, franchisor or client controls the records |
| Authorized contact | A named CEO who can open a licensing discussion | Nobody with that authority identified |

Documented operating history is also required; a company with none fails outright, and depth is scored below. English share is a fourth gate outside the U.S., Canada and the EU: Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English.

## Seven criteria for linked records

| Criterion | Strong (2) | Borderline (1) | Weak (0) | Question to ask |
|---|---|---|---|---|
| Operating history | Five or more years of customer history across the main systems | History broken by a system change | Under two years | Which years does each customer system cover? |
| Cross-reference fields | CRM or help desk stores the ERP customer number | Stored for key accounts only | No system stores another's ID | Does any system hold another system's ID? |
| Migration maps kept | Old-to-new ID mapping kept for every replaced tool | Kept for the latest migration only | Deleted after go-live | When a system was replaced, was the mapping kept? |
| Merge logs | Retired ID, surviving ID and date logged on each merge | Logged in one system | Duplicates deleted with no trace | When two records merge, is the retired ID logged? |
| Shared strong key | Tax ID, contract or customer number on most records in each system | On some records | None | Is there a key that appears in every system? |
| Link decisions recorded | Linked and held-for-review statuses kept | Links kept, no status | Links made silently | Are uncertain matches marked, or just merged? |
| Downstream confirmation | Payments, contracts or disputes tie back to the linked record | Payments only | Nothing confirms a link | Does anything later prove a link was right? |

Ask in that order. If no system holds another's ID and no mapping was kept, the later answers carry little weight. The [company qualification checklist](/resources/company-qualification-checklist) covers the general screen behind these rows.

## Three Illustrative companies, scored

Each company is fictional, and the answers came from one call with an operations or finance lead.

| Criterion | 540-person staffing firm, two acquisitions | 150-person software reseller | 85-person dental clinic group, Australia |
|---|---|---|---|
| Operating history | 2 | 2 | 1 |
| Cross-reference fields | 2 | 1 | 0 |
| Migration maps kept | 2 | 1 | 0 |
| Merge logs | 2 | 1 | 1 |
| Shared strong key | 2 | 2 | 1 |
| Link decisions recorded | 1 | 0 | 0 |
| Downstream confirmation | 2 | 1 | 1 |
| Total | 13 of 14, strong | 8 of 14, borderline | 4 of 14, weak |

**The staffing firm** absorbed two regional agencies and kept the client mapping files from both integrations. Its CRM stores the billing system's client number and logs every merge. It loses one point because uncertain matches were merged by a coordinator without a status, so a reviewer cannot tell confirmed links from judgment calls.

**The software reseller** has a shared customer number in its ERP and CRM, but the help desk is keyed on email domain and the old billing tool's mapping was kept only in part. Borderline, with the help desk and pre-2020 billing named as open items.

**The clinic group** has patient and payer records in separate practice systems with no cross-references, and merges are done by deleting duplicates. Patient records also bring health privacy questions that the rights review would have to weigh. It passes the gates, but the fit is weak until linkage evidence turns up.

## When to raise it, and what the score cannot show

Identity questions come up naturally during consolidation work, so [CRM consolidation after an acquisition](/guides/crm-consolidation-after-acquisition) is a good moment to ask, as is the diligence work described for [cross-border M&A advisors](/for/cross-border-m-and-a-advisor-referral-program). A company about to retire a tool should keep its mapping files; [exporting data before cancelling SaaS](/resources/export-data-before-cancelling-saas) covers why.

The score cannot show how many links hold up when checked, whether linked records were collected under terms that permit joining them, whether any buyer wants them or what the company would be offered. Mercor, for example, says on its data page (checked October 18, 2026) that it works with enterprises to anonymize and contribute operational data to AI labs; that describes one company's business, not demand for any particular archive. Only SourceX's review settles fit.

Next, the [evidence gaps page for identity resolution](/questions/cross-tool-identity-resolution-evidence-gaps) shows how to describe weak rows, and the [review packet contents list](/resources/cross-tool-identity-resolution-review-packet) shows what the company would prepare. More record families are under [data types](/data-types).

## Next step

> **Refer a company. No cap on rewards.** Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.

If the company clears the gates and its linkage answers land it in the strong or borderline band, [register as a partner](/register), confirm [who qualifies](/who-qualifies) and read the [program terms](/terms) before you introduce its decision-maker.

## Sources

- https://moj-analytical-services.github.io/splink/topic_guides/theory/fellegi_sunter.html
- https://help.salesforce.com/s/articleView?id=sales.matching_rules_matching_algorithms.htm&language=en_US&type=5
- https://www.mercor.com/data/
- https://refer.sourcex.si/terms
