Recovery loops evidence gaps: what a review expects from failed runs and fixes, and what usually survives

Short answer

A recovery loop can only be judged if every turn of it survives: the control, each failed run, the diagnosis, the fix and the final signed check. Companies usually keep the start and the end, the legacy control and the clean final balance, and lose the middle. Failed runs vanish when import tools purge history, fixes go unrecorded, permissions are unclear and sign-offs are vague. Describe gaps in an introduction as what is confirmed, what is likely missing and who can check.

Visual briefFull-size diagram
Six columns compare the evidence a review expects for a recovery loop with what usually survives a past migration: the control is almost fully kept, while failed runs, diagnosis notes, fixes, authority records and the sign-off show dashed outlined segments where records are usually missing.
Illustrative pattern, not measured data. Column heights show a typical shape of what survives a migration, not figures from any company.

This gap analysis is part of recovery loops: five questions for introducers. The worked import example shows a complete loop of three runs. This page sets that ideal against what a company that migrated years ago usually still has, so you can predict the shortfall and say so plainly.

The evidence a review expects for each loop

A reviewer replaying a loop needs six kinds of record. The first and last frame the loop; the four in between are the loop itself.

EvidenceReviewer's question it answersIf absent, the reviewer concludes
ControlWhat was the target?Nothing can be called short or over
Failed-run logsHow far did each attempt miss?Only a success is visible
Diagnosis notesWhy did it miss?Retries look random
Fix recordsWhat changed before the retry, and who changed it?Repair and luck look the same
Authority recordWas that person allowed to change it?Controls may have been bypassed
Signed final checkDid a named person confirm the pass?The end point is unverified

What usually survives, and what usually does not

The pattern below is a typical shape, not a measurement. It is what a reviewer should expect at a mid-sized company that finished an ERP migration two or more years ago without planning to keep the trail.

EvidenceUsually keptUsually missingWhy it goes missing
ControlLegacy aging report or trial balance at cutoverRarely missingAuditors want it
Failed-run logsThe final successful runEarlier failed runsTool purges history on a schedule
Diagnosis notesScattered emailsA note tied to each failed runAnalysts fix and rerun without writing
Fix recordsThe final mappingEarlier mapping versions and the date each changedFiles saved over
Authority recordThe cutover planWho approved each master data changeAccess was granted informally under deadline
Signed final checkA final balance in the ERPA bucket-by-bucket comparison with a named approver"Looks fine" in an email

The failed-run row is the one to ask about first, because the loss is often automatic. Microsoft's documentation for Dynamics 365 Finance and Operations says each import job run records succeeded and failed counts per entity and an execution log, and that by default job history and related staging data older than 90 days are deleted. Other products have their own rules; ask rather than assume.

How each gap breaks the loop

Using the three-run example from the workflow page:

  • Failed runs purged. Runs 1 and 2 disappear. The archive shows one successful import, with no failure to learn from.
  • Fix not recorded. The change between runs 2 and 3, reversing the sign on credit memos, leaves no trace. Run 3 looks like luck. Worse, a reviewer cannot rule out that someone forced the balance with an adjusting entry instead.
  • No authority record. Nobody can say who created the 31 missing customers or whether they were allowed to. The loop may have broken the company's own controls.
  • Vague final check. Without a bucket comparison and a named approver, the loop has no verified end.

Research on agents shows why the diagnosis gap hurts most. The Reflexion paper describes agents that keep written reflections on failed attempts in a memory buffer and use them on the next try. Without the reflection, a sequence of attempts is just repetition, and the same is true of a company whose analysts never wrote down why a run failed.

GapRun it erases in the exampleWhere the evidence may still beMetadata question
Tool purged job historyRuns 1 and 2Exported error files, project shared drive"Were import error reports saved outside the tool?"
Fix not recordedThe change between runs 2 and 3Change tickets, mapping version history"Were mapping changes logged during the cutover?"
No permission recordWho loaded 31 customersAccess review, cutover RACI"Who signed the cutover plan, and is it kept?"
No final checkRun 3 sign-offReconciliation workbook, controller email"Is there a signed reconciliation for each migrated balance?"

Describing the gap honestly in an introduction

Name what the contact confirmed, what is probably missing and who can check. Three Illustrative examples:

SituationHonest wording
Tool purged, files kept"The controller says the import tool no longer shows failed runs, but error reports were emailed to a project mailbox that IT archived. The CIO can confirm what the archive holds."
Fixes undocumented"Two ERP migrations since 2019 with signed reconciliations. Mapping changes between attempts were not logged, so fixes may be hard to trace."
Implementer holds the files"The implementation partner ran the loads and kept the working files. Whether the company can obtain them depends on the engagement contract."

Avoid estimating completeness, calling a gap minor before anyone has looked, or suggesting the review will find a workaround. Vendors such as Appen market agent data built on failure-mode taxonomies and verifiers, which shows buyers care about failures. It does not show that any particular company's partial records are wanted.

Where missing evidence sometimes turns up

Some gaps close when the right place is checked. Backups and restored systems may hold old import staging tables; the page on ransomware-restored records explains why gaps in restored data should be described honestly. Process mining tools may have captured event logs from the cutover period. If an agency or implementer did the work, the guide on separating a firm's own operating records from client-owned work helps settle who holds what. In a sale or wind-down, an assignee may hold the systems as estate assets; see ABC assignee intangible asset recovery. The controller who signed the cutover usually knows where the reconciliation workbooks went.

The company still needs 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. The review packet contents list shows what the company would assemble later. More record families are in 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 you can describe a client's migration records and their gaps in three honest sentences, register as a partner, check who qualifies, and read the program terms before introducing the CFO or CIO.

  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

If the import tool purged its history, is the loop lost?

Not necessarily. Project teams often exported error reports, saved mapping versions and kept reconciliation workbooks on a shared drive. Ask where project files were archived, not what the tool shows today.

Does a missing gap disqualify the company?

Rarely on its own. Each gap reduces how many loops can be judged. Only SourceX's review with the company can say how much survives.

Should I estimate how complete a client's records are?

No. Pass on what the contact confirmed and what is likely missing, and let the review measure it. A guessed percentage tends to be wrong in both directions.

Sources

  1. learn.microsoft.com/en-us/dynamics365/fin-ops-core/fin-ops/data-…
  2. papers.nips.cc/paper_files/paper/2023/hash/1b44b878bb782e69…
  3. appen.com/agentic-ai
  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