Integration cutover reconciliation: what to inventory before a legacy application sunset

Short answer

Before a legacy application is switched off, inventory the reconciliation trail of its cutover: every version of the field mapping, the import and validation logs for each load, the count reconciliations, and the exceptions that someone accepted in writing. Note where each record lives, how long it is kept and who owns it, because run logs in ETL tools can expire within weeks. A systems integrator records only metadata and introduces the implementation director plus the executive who owns the data.

Visual briefFull-size diagram
A legacy status code that meant credit hold before 2019 and pickup hold after it, followed by bars showing 6,400 unexplained holds falling to 1,250 after mapping v2 and to zero after the remaining lines are accepted as exceptions in v3.
Illustrative reconciliation for a 380-person parts distributor. Line counts are fictional.

Why the reconciliation trail disappears at sunset

Cutover reconciliation is the work of proving that what landed in the new system matches what left the old one, and explaining every difference. The proof is spread across tools that are each retired or cleaned up on their own schedule: the legacy application and its database, the ETL or integration tool, the project tracker and the shared drive where mapping specifications were kept.

Run history is the first to go. Microsoft documents that Azure Data Factory only stores pipeline run data for 45 days and recommends routing diagnostic logs through Azure Monitor to keep it longer. A project that did not set that up has already lost the detail of its early trial loads, whatever the sunset date. Ticket history is next: Jira's data pipeline export writes issue history to a separate file with field names and from and to values, but includes only history after the export's start date, so a narrow export window drops the early decisions.

Mapping specifications are the quietest loss. Teams often edit one document in place, so the version that produced a bad trial load is overwritten by the version that fixed it. Without the earlier version, nobody can show why the numbers changed.

Illustrative: one legacy code, two meanings

A fictional 380-person parts distributor is retiring its legacy order system. Its status code H meant credit hold until 2019 and pickup hold after that, a change nobody documented.

  1. Mapping v1, trial load on Mar 2. Every H is mapped to credit hold. Reconciliation shows 6,400 unexplained credit holds in the new system.
  2. Mapping v2. After the finance team spots the pattern, 5,150 lines dated from 2019 are remapped to pickup hold, leaving 1,250 unexplained.
  3. Mapping v3. The remaining 1,250 lines cannot be classified from the data, and the implementation director signs them off as accepted exceptions. Unexplained holds fall to zero.

The chart above shows the three versions. If only v3 survives the sunset, the history shows a clean mapping and loses the reason it took three attempts: the undocumented change of meaning and the decision to accept what could not be resolved.

What should be inventoried before a legacy application sunset affects integration cutover reconciliation history?

List record classes, date coverage, the owning system and the real access deadline, which is the day the legacy application, its database and the project tooling lose their licenses or servers. Preservation is the company's decision; the inventory only makes the decision informed.

A metadata-only checklist for the implementation director:

  1. Mapping specifications: how many versions exist, and are old versions kept or overwritten?
  2. Load and validation runs: which ETL tool, how many trial and production loads, and does run history still exist or was it exported?
  3. Reconciliation outputs: which counts were tested (lines, balances, open items), and where the results are stored.
  4. Cutover tickets: which project tracker, roughly how many defects and exceptions, and whether approvals sit on the tickets.
  5. Legacy reference data: code tables and their documented meanings, including undocumented changes like the code H example above.
  6. Access deadline: the actual decommission dates for the legacy application, the ETL environment and the tracker project, not the go-live date.
  7. Authority: who in the company decides what is kept, and whether any customer or supplier contracts restrict reuse of order data.

Do not suggest keeping data longer than the company's policies and contracts allow, and do not copy anything for safekeeping yourself. When the sunset follows an acquisition, the M&A integration records page and the buy-and-build IT integration guide cover who typically controls the systems at that stage.

Where each record usually lives, and how long it lasts

Use this as a prompt for the implementation director, not as a description of any one client.

RecordUsual homeWhat can shorten its lifeWho to ask
Mapping specificationsShared drive or wikiEdited in place, old versions overwrittenImplementation lead
Load and validation run historyETL or integration toolTool run retention, such as 45 days in Azure Data FactoryIntegration engineer
Count reconciliationsSpreadsheets or reportsSaved on personal drivesFinance or data lead
Cutover defects and accepted exceptionsProject trackerProject archived after go-liveProject manager
Legacy code tablesLegacy databaseDatabase switched off at sunsetLegacy system owner

Tracker exports are a question for the project manager: Jira, for example, can export search results, including to CSV, and whether a given export would carry the approvals is for the project manager to confirm. The right-hand column matters as much as the rest. Each record has a different custodian, and the inventory is only complete when each of them has answered.

Does the company qualify, and what stays unknown?

The introduced company needs 50+ full-time employees at peak (contractors excluded), a documented operating history, rights to license the records, 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 90% of the data is in English.

What nobody knows until SourceX reviews the opportunity: whether there is dataset demand for these records, the rights review result, what must be excluded or de-identified, and the company's offer. SourceX licenses approved business data directly from companies for AI training and evaluation and manages agreement, rights review, data preparation, delivery and payout. Partners sign the Channel Partner Agreement before submitting introductions; the first valid introduction within the 90-day window is credited when the company submits a verified application. Illustrative only: if a company's Actual Supplier Payout were $250,000, the default reward would be $62,500; at $800,000 it would be capped at $100,000. 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. More sectors are on the industries hub.

Go deeper on integration cutover reconciliation

Next step

Refer a company. Earn up to $100,000. Default reward: 25% of Actual Supplier Payout, capped at $100,000 cumulatively per referred company. Eligibility and payment conditions apply, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.

If your client's reconciliation trail survives and the sunset date is set, register as a partner, confirm the company against who qualifies, and read the program terms and reward rules before sending the brief.

  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 rewardYour 25% of the actual supplier payout. 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

Are the mapping specifications mine or the client's?

It depends on the statement of work. Many integration contracts assign deliverables to the client, but some reserve tools, templates and methods to the integrator. Check your own contract before describing those documents as the client's records.

Is a clean cutover with no exceptions less interesting?

Usually, yes. A load that matched on the first pass shows little judgment. The valuable records are the ones where a mismatch was found, explained, fixed or accepted, and then checked.

Can I mention the legacy system's vendor in the introduction?

Yes, as a plain system name. It helps SourceX understand what exports are possible. Do not include logins, extracts, screenshots or table dumps.

Sources

  1. refer.sourcex.si/terms
  2. learn.microsoft.com/en-us/azure/data-factory/monitor-programmati…
  3. confluence.atlassian.com/adminjiraserver/data-pipeline-export-schema-…
  4. support.atlassian.com/jira-software-cloud/docs/export-search-results

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 90% 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