Does agent recovery data qualify? An eight-point scorecard for failed-handoff records

Short answer

An introducer scoring agent recovery data checks eight things: company size, rights to the records, English share where it applies, whether failures are logged with a time, whether the follow-up is linked to the original failure, how the fix is confirmed, whether bot and ticket histories are kept for the same years, and who can authorize a licensing discussion. Each scores 2, 1 or 0. Twelve or more of 16, with no zero on a gate, is a strong case for an introduction.

Visual briefFull-size diagram
Eight criteria tiles light up one by one, each scored 2 or 1, while a pointer slides along a horizontal meter with not yet, borderline and strong bands and settles at 13 of 16 for a fictional energy retailer.
Illustrative score for a fictional 520-person energy retailer. The scoring scheme is a partner aid, not SourceX's review.

This page is part of the guide to agent recovery data, which follows a failed chat transfer through to a verified fix. Its job is a qualification scorecard: a way to turn a conversation with the head of customer operations into a decision about whether to introduce.

Reading the scorecard

Each criterion scores 2 (strong), 1 (borderline) or 0 (weak), for a maximum of 16. Four criteria are gates: company size, rights, authorized contact and, for companies outside the U.S., Canada and the EU, English share. A zero on any gate means "not yet", whatever the total.

TotalWhat it meansWhat the partner does
12 to 16, no zero gateStrong caseIntroduce the decision-maker
7 to 11, no zero gateBorderlineIntroduce and name the weak criteria
0 to 6, or a zero gateNot yetGive the owner the weak criteria to check

This scheme was built for this page as a partner aid. It is not how SourceX reviews a company, and a score is not a forecast of acceptance or of any reward.

Eight criteria and what each answer sounds like

Three criteria test the company, four test the recovery records and one confirms who can say yes.

CriterionStrong (2)Borderline (1)Weak (0)
Company size (gate)Clearly 50+ full-time employees at peakNear 50 after excluding contractorsUnder 50 full-time employees
Rights (gate)Company holds its own conversation and ticket historyBot or help desk vendor terms need checkingRecords held by an outsourcer with no clear rights
English share (gate)Company in the U.S., Canada or the EU, or 70%+ English elsewhereMostly English, share not yet measuredWell under 70% English outside the U.S., Canada and the EU
Failure loggedRouting or status events stored with a timeFailures visible in reports but not stored per caseNoticed only when a customer complains again
Follow-up linkedTicket stores the original conversation or case IDLinked in recent years onlyA new, unconnected ticket in another channel
Fix confirmedSettled refund, reason code and a repeat-contact checkReason code only"Agent marks it solved"
Histories overlapBot logs and tickets kept for the same yearsBot logs kept for a shorter periodBot logs not kept at all
Authorized contact (gate)CEO, with the COO or head of customer operationsOwner known but not yet approachedUnclear, or a vendor controls the decision

The four record criteria map onto what agent developers look for. Appen's service page, checked October 18, 2026, describes a trajectory analysis service that classifies where agents go wrong and, for failed runs, has experts write the correct continuation. A company whose people recovered real failures, and recorded it, holds the natural version of that continuation, but only when failure, fix and confirmation are linked.

Help desk features make strong answers more likely. Zendesk's developer reference explains that every ticket update stores an audit with events such as field changes and comments, which lets the order of a recovery be rebuilt. Its messaging documentation describes conversation control being passed to an integration or released back to a default state, which is the kind of event that marks a failed handoff. Any system with equivalents can score well.

Three fictional companies, scored (Illustrative)

The companies and scores are invented to show how the card reads in practice.

Criterion520-person energy retailer, Canada150-person travel booking firm, India90-person software firm, UK
Company size222
Rights221
English share202
Failure logged221
Follow-up linked110
Fix confirmed111
Histories overlap110
Authorized contact222
Total13, strong case11, not yet (English share gate)9, borderline

The energy retailer logs every failed transfer and links follow-ups for its billing queue, though not for its outage line, and keeps bot logs three years against seven for tickets. The travel firm has good records, but roughly half its chats are in Hindi; as an Indian company it needs at least 70% English, so it scores zero on a gate and is not ready to introduce, whatever its total, unless the owner can show that the data in scope meets the 90% rule. The software firm is small but qualifying; its chat vendor keeps logs for a limited period and follow-ups are opened as fresh tickets, so the recovery chain is mostly broken. The graphic on this page plays the energy retailer's score.

Five questions behind the scores

These come straight from the scorecard. Ask them and write down answers, never files.

  1. Which failures does the team routinely recover from? "Released bot handoffs, failed payments and missed callbacks, a few hundred a month" is strong. "We fix things as they come up" is weak.
  2. Is the failure itself logged, with a time? Decides the failure-logged score.
  3. Is the follow-up linked to the original failure? A conversation or case ID on the ticket is the single most important field.
  4. How is the fix confirmed? A settled refund or a repeat-contact check beats a solved status.
  5. Who owns these records, and who can approve a licensing talk? The COO or head of customer operations usually owns them; the CEO approves the licensing talk.

Keep the conversation at metadata level: what exists, how much, for how long and who controls it. For company-level basics that apply to any topic, use the company qualification checklist; the introducer agreement guide covers your own paperwork.

Limits of a score

A score summarizes answers; it cannot test them. It does not show how many recoveries would pass an acceptance check, how customer data will be de-identified, or whether a buyer wants this task type. Industry interest in recovery data is real, but it does not mean every support archive qualifies.

The company test comes from the program terms: 50+ full-time employees at peak (contractors excluded), documented operating history, rights to license and an authorized contact. 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.

If the score is borderline because links break, the evidence gap analysis shows how to describe each gap. Related reading: what a computer-use agent is, lead qualification notes data for another kind of customer-facing record, and a business introducer's guide to operational data licensing.

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 a client scores 7 or more with no zero on a gate, register as a partner, confirm the company against who qualifies, and read the program terms before you introduce the COO or CEO.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
  4. Step 4You get your rewardthe greater of $50,000 plus 5% and 15% of the actual supplier payout, with no cap. A reward becomes payable after SourceX has paid the referred company under its signed agreement and the payout conditions in the program terms are met, including finance approval and your tax and payment details.

Common questions

Does the company need a chatbot to score well?

No. A failed phone transfer, a dropped callback or a payment that bounced and was later fixed are all recoveries. The scorecard tests whether failure, fix and confirmation were recorded and linked, whoever caused the failure.

Why is English share a criterion here?

Because support conversations are often multilingual. For companies outside the U.S., Canada and the EU, the program requires at least 70% of the data to be in English, so a support archive split between languages can fail that test even when everything else is strong.

Can a partner ask for a sample recovery to check the score?

No. Partners ask about systems, counts, retention and ownership, and score from the answers. Transcripts and tickets stay with the company until SourceX reviews them under an agreement.

Sources

  1. appen.com/agentic-ai-trajectory-analysis
  2. support.zendesk.com/hc/en-us/articles/5514406080538-About-Sunshi…
  3. developer.zendesk.com/documentation/ticketing/reference-guides/tic…
  4. refer.sourcex.si/terms

Free resources

Know a company with valuable proprietary data?

Become a referral partner, get your link and introduce the company’s CEO (in an owner-run company, usually the owner). Referral partners can join from most countries. We can't work with partners in countries or regions under U.S. sanctions or on U.S. restricted-party lists. The companies you introduce should be based in the U.S., Canada and the EU, or in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with at least 70% of their data in English.

Refer a company →

I own a business

Explore licensing your company's data directly to SourceX for AI training and evaluation. Start a short assessment; no uploads needed.

Start an assessment