Temporal context evidence gaps: the missing dates and versions that hide which rules applied

Short answer

Temporal context is hard to assess when a record that rebuilds a past date is missing. Six gaps do the damage: old rule versions thrown away, the decisive date not kept separately, no record of when a change was issued, no link from the case to the version used, no history of later corrections, and no record of who held authority at the time. Each gap invites a specific wrong conclusion. Gaps narrow what can be licensed; they rarely end the conversation.

Visual briefFull-size diagram
Six vertical columns show, for a fictional insurance agency, how many of the last ten years carry each kind of temporal evidence; filled sections rise to the years available and orange outlined sections mark the missing years, largest for the settlement authority table and the claim-to-version link.
Illustrative owner estimates for a fictional insurance agency, 2016 to 2025.

This gap analysis is part of the guide to temporal context in an ordinary workflow, which follows one property claim decided under terms that changed partway through the year. Here the question is what a reviewer can and cannot rebuild when parts of such a record are gone, across a whole archive rather than one file.

Six kinds of evidence a review expects

To judge one decision against the rules valid at the time, a reviewer needs to rebuild two timelines: when things happened, and when the company knew. Six kinds of evidence carry those two timelines.

Expected evidenceWhich timeline it rebuilds
Old rule versions with start datesWhat the rules said on any given day
Decisive date, separate from other datesWhich day's rules apply to this case
Issue or entry date of each changeWhen the company knew the rules had changed
Link from the case to the version usedWhich rules the decision-maker actually applied
Payment and later correctionsWhether the decision stood or was reversed
Authority table in force at the timeWhether the decision-maker was allowed to decide

A database built for this can answer "what was true on that date" directly. Microsoft's SQL Server documentation describes a FOR SYSTEM_TIME AS OF query that returns the rows valid at a chosen point in time. Most archives are not built that way, so the evidence has to be assembled from documents and fields, and that is where gaps appear.

What archives usually keep, and what they lose

The pattern reflects how policy, claims and pricing systems are commonly run, not a survey of companies.

EvidenceUsually keptUsually missingUsual owner
Old rule versionsDocuments in a shared drive or document systemStart dates on older versionsPolicy administration
Decisive dateA date on every caseA separate loss or quote date in older systemsClaims or sales operations
Issue date of each changeStart datesThe date the change was entered or issuedPolicy administration
Case-to-version linkThe amount paid or price chargedA version number on the decisionClaims or finance
CorrectionsThe final figureThe first decision and the reason it changedFinance
Authority tableToday's approval limitsLimits in force in earlier yearsOperations or HR

The wrong conclusion each gap invites

The table uses the illustrative March claim. Each row removes one record and shows what a reviewer would wrongly conclude.

Missing recordWhat a reviewer sees insteadWrong conclusion it invites
Retired endorsement 2Only endorsement 3 in the fileThe $10,000 deductible looks like an error or a favor
Loss date (only report date kept)April 9 as the only dateEndorsement 3 appears to apply
Issue date of endorsement 3A start date with no entry dateCannot tell whether the change was known when the claim opened
Link from claim to the version usedA payment with no referenceCannot check whether the adjuster followed the rules
Payment and any later correctionA settlement figure with no follow-upCannot tell whether the result stood or was reversed
Settlement authority table for MayToday's authority limitsCannot check whether the approver was allowed to approve

The first two rows are the most damaging, because they make a correct decision look wrong. The others leave a decision unverifiable rather than misread. Data platforms describe the same concern from the buyer side: micro1 says (checked 2026-10-18) its quality step flags issues such as missing evidence. That shows interest in the category, not demand for any particular archive.

Ten years at an insurance agency, estimated (Illustrative)

Illustrative and fictional. A 230-person commercial insurance agency is asked, for each kind of evidence, how many of its last ten years of claims (2016 to 2025) carry it.

EvidenceYears available, out of 10Why it falls short
Superseded endorsements10Document system kept every version
Separate loss date6Old claims system had one combined date until 2020
Issue dates4Recorded only since a policy system upgrade
Case-to-version link3Adjusters cite endorsement numbers only since 2023
Payments and corrections8Earlier payment history archived in summary form
Authority table2Approval limits kept only as current settings

Read across, 2023 to 2025 is the fully checkable slice except for authority in 2023, and 2020 onward supports rebuilding which rules applied, though not always which the adjuster cited. That is a concrete description an owner can give in one paragraph. The graphic on this page draws these six columns.

Describing the gaps in an introduction

Instead ofSay
"We keep full history.""Every endorsement since 2016; loss dates kept separately from 2020."
"Adjusters follow the policy.""Settlements cite the endorsement number from 2023 onward."
"Approvals are controlled.""Current authority limits are on file; earlier limits are not."

Give years and who answered, and note any migration dates. Issue dates deserve special care: Martin Fowler's essay on bitemporal history points out that records can lag behind reality, so the date a change was entered is a separate fact from the date it took effect. When a company is winding down or changing systems, these dates matter even more; the data opportunity handoff checklist for a startup shutdown covers that case. Where rule changes were discussed in chat before being entered, the metadata-only Microsoft Teams inventory template shows how to describe those channels without opening them.

Who can fill the gaps, and what stays unknown

Policy administration owns versions and issue dates; claims operations owns loss dates and adjuster notes; finance owns payments and corrections (the resource on MCP for CFO dashboards explains why finance figures need traceable evidence behind them); operations or HR owns authority tables. The CEO, or the principal in an owner-run agency, can authorize a licensing discussion, and carrier agreements may affect rights in some documents.

The company test does not change: 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: how many cases pass, the de-identification needed, the rights review outcome and SourceX's current data needs. To rate a client first, use the temporal context qualification scorecard; for what a review asks the company to prepare, see the temporal context review packet. Claim-file records more broadly are covered in claims notes as a data type.

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 client can say which years carry each kind of evidence, as in the table above, register as a partner, check who qualifies, and read the terms before introducing the decision-maker.

  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

Is a gap during a system migration a deal-breaker?

No. A known gap with clear start and end dates is easy to scope around. It is far better to state it than to let a reviewer find it.

What if old rule versions survive only as email attachments?

They may still count, if they carry dates and can be matched to cases. Describe where they sit and roughly how complete they are; SourceX judges usability in review.

Should the agency rebuild the missing authority tables?

No. A table rebuilt today is not a record from the time. Describe the gap honestly and let the review decide how to treat those years.

Sources

  1. martinfowler.com/articles/bitemporal-history.html
  2. learn.microsoft.com/en-us/sql/relational-databases/tables/queryi…
  3. micro1.ai/blog/introducing-flow
  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