What a standalone automation run log loses: one failure read before and after its tickets and article history

Short answer

A single-system export of a failed automation keeps what failed, when and with which error. It loses what people did about it: the invoices a clerk matched by hand, the judgment behind the fix, the macro and article that turned the fix into a rule, and the later tickets showing the rule worked. Read alone, a failure the business absorbed by hand for weeks looks like a short glitch that fixed itself. The missing context lives in the helpdesk and the knowledge base.

Visual briefFull-size diagram
A run log row showing eleven invoice-routing failures and then success has blank fields for business impact, manual fix, rule change and proof the rule worked; ticket and article pieces slide in to fill each one.
Illustrative. The run log says what failed and when; the helpdesk and knowledge base say what people did about it.

This page is part of the failed automation to policy repair guide. It reads one fictional case twice: from the automation platform's run history alone, then with the tickets and article revisions around it.

Before: the invoice-routing failures as the run log tells them

Illustrative. A fictional 520-person building-materials wholesaler routes supplier invoices from an accounts payable inbox to approvers through an automation platform. A reviewer sees only the run history export. Every value is invented.

Field in the run historyValue
WorkflowSupplier invoice capture and approval routing
Failed runs11 over three weeks
Error"PO number not found" at the extraction step
Affected senderOne supplier's billing address
Week 6 onwardInvoices from that supplier route normally
Inputs and outputsAvailable only within the plan's retention window
Who did whatNot recorded

What a reader concludes

One supplier's invoices failed to parse for three weeks, then the template was fixed. Eleven failures look like eleven invoices delayed, perhaps paid late. Nothing says a clerk matched each invoice to a purchase order by hand, that two clerks applied different standards, or that anyone wrote down how to treat invoices without a readable PO.

The export may be thinner still. Workato's documentation says job history retention depends on plan, 30 days on some and configurable up to 90 days on others, and that once it expires, job details including action inputs and outputs are deleted while summaries remain. A reviewer a year later may see only that eleven runs failed.

After: the same failures with tickets and article revisions

Restore the AP service desk and the finance knowledge base, and the three weeks become a story about people.

  • Service desk, week 1. The first alert ticket shows an AP clerk finding the PO number printed in the remittance footer after the supplier redesigned its invoice, then keying the match by hand.
  • Service desk, weeks 1 to 3. Ten more tickets. Two clerks apply the same fix; one also confirms the receiving record before approving, the other does not.
  • Service desk, week 3. The AP lead saves a reply template: find the footer PO, check receipt, route to the buyer.
  • Knowledge base, week 5. The "Invoices without a readable PO" article is revised to require the receipt check. The version note cites the first ticket.
  • Automation platform, week 6. The extraction template is updated for the new layout. Only now do runs succeed.
  • Service desk, week 9. A different supplier's invoice fails the same way; a clerk follows the revised article, including the receipt check, and closes the same morning.

The reader now sees three weeks of manual matching, two clerks applying different standards, a written rule that settled which standard was right, a template fix that came last, and evidence that the rule worked on a new supplier. Zendesk help center revisions show the differences between versions, so the article change itself is inspectable; the reason for it is visible only through the ticket number in the note.

The lost-context table: what a run export drops and where it lives

Each row is context the automation platform cannot hold, the system that usually holds it and the person who can say whether it exists.

ContextRun export aloneWhere it livesWho can confirm it exists
Business impactNot includedTicket notes: invoices, discounts or accounts touchedSupport operations manager
The manual fix and its reasoningNot includedInternal ticket notesSupport operations manager
Who approved the workaroundNot includedTicket assignee, supervisor note or macro ownerSupport operations manager
Grouping of repeat failuresSeparate failed runsTicket tags, linked tickets or problem recordsSupport operations manager
The written ruleNot includedMacro, article revision or SOPKnowledge owner
When the rule changed, and whyNot includedArticle revision history and version noteKnowledge owner
Proof the rule workedNot includedLater tickets citing the articleSupport operations manager
Run inputs after retention expiresSummary onlyExported run history, if anyone kept itIntegration or IT lead
Repairs that never became a ruleInvisibleRecurring fixes with no matching articleSupport operations manager

The last row matters most. A workaround applied again and again in the helpdesk but never written into the documented process is the case where the run log and the knowledge base both look complete and the company's real rule exists only in agents' habits. Tools can create the gap without anyone being careless: in ServiceNow, an incident with the Knowledge check box selected creates a draft article on close, and the draft stays out of the knowledge base until it is reviewed and published. For how the keys rejoin the columns, see the sibling guide on failed automation to policy repair record linkage.

Why the right-hand column is the useful part

For AI training and evaluation, the run log alone teaches a model that systems fail. The connected history shows an exception, a human decision, a written rule and a test of that rule, which is the sequence agents most need and most often lack, as why AI agents fail discusses. The journal entry automation example shows the same exception-to-rule pattern in a finance workflow. For the categories themselves, see integration error handling records and internal knowledge bases.

Scoping what the company can restore

A partner learns which rows of the table the company can restore, from which year and who controls each system. You ask; you do not collect.

  • Records that exist: run history and any exports, failure tickets, macros, article revisions or SOP versions, configuration change logs.
  • Who owns them: the integration or IT lead, the support operations manager and the knowledge owner.
  • Who can authorize a licensing discussion: the CEO, often advised by the COO, CFO, CIO or general counsel, because tickets and payloads can hold customer and supplier details.
  • Company baseline: 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.
  • Unknown until SourceX reviews it: whether rights cover each system, what must be removed, which years a buyer wants and what the company would be paid.

If a client tried to sell data before and it stalled, why past data monetization failed covers what carries over. The completeness checklist turns the table into scored questions.

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 your client can restore the tickets and article history behind its failed runs, register as a partner, check the company on who qualifies and read the program terms before you introduce the CEO, who can authorize it.

  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

The company replays failed runs and they usually succeed. Is there any context to lose?

Less, but some. A replay shows the run eventually worked, not whether anyone changed data or instructions first. Note how often repairs were plain replays; that is useful scoping information for SourceX.

Can a company rebuild lost run detail after retention expires?

Usually not from the platform itself. Summaries may remain, and tickets often quote the error, so the helpdesk can partly fill the gap. Ask whether run history was ever exported before it expired.

Should I read a few tickets to see whether repairs are documented?

No. Ask the support operations manager to describe, from memory, two or three recurring manual fixes and whether each has a matching article. Ticket content stays inside the company.

Sources

  1. refer.sourcex.si/terms
  2. docs.workato.com/security/data-protection/data-retention
  3. support.zendesk.com/hc/en-us/articles/4408829321498-Viewing-arti…
  4. docs.servicenow.com/bundle/orlando-it-service-management/page/pr…

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