Private equity and portfolios
Legacy data after system consolidation: keep, archive, license or delete?
By SourceX Editorial · Updated
Short answer
Legacy data after a system consolidation should be routed record set by record set to one of four outcomes: keep it live in the surviving system, archive it read-only, license a prepared subset, or delete it. Four questions decide the route, in order: legal hold, retention schedule, rights status and likely value. Never delete before the first three are answered.
Key takeaways
- Route each record set, not each system: one help desk can hold tickets, recordings and attachments that belong in different outcomes.
- A legal hold overrides every other answer; held records are archived intact, not deleted or left behind in a dying system.
- Licensing is a layer on top of archiving: the full history is kept, and only a prepared subset is licensed.
- Delete only after the hold, retention and rights questions are answered and the approval is logged.
- Decide before the old vendor contract ends, because exports and admin access usually end with it.
What are the four outcomes for legacy data?#
Legacy data after a consolidation has four possible outcomes. Keep means records move into the surviving system and stay in daily use. Archive means a complete, read-only export is kept with its structure and links intact. License means a prepared subset of archived records is licensed to an AI developer under contract. Delete means records are disposed of under the retention schedule, with approval logged.
The outcomes are not exclusive. A retiring help desk might see open tickets kept, the full history archived, a prepared subset licensed, and old call recordings past their retention period deleted. That is why routing works at the level of record sets rather than whole systems.
| Outcome | What happens to the records | Typical record sets |
|---|---|---|
| Keep | Migrated into the surviving system and used daily | Open tickets, active customer accounts, current price lists |
| Archive | Exported in full, read-only, with IDs and attachments | Closed tickets, completed jobs, historical orders |
| License | A prepared subset of the archive licensed under contract | Linked history such as tickets with resolutions or jobs with callbacks |
| Delete | Disposed of under the schedule, with approval logged | Expired records with no hold, no right to reuse and no value |
The four routing questions, in order#
The four routing questions must be asked in order, because an earlier answer can override a later one. A record set with obvious value but an active legal hold still goes to archive first, whatever anyone hopes to do with it.
Likely value has two parts that are easy to confuse. Operational value asks whether staff still need the records to serve customers, handle warranty claims or answer audits. Licensing value asks whether the records show real work tied to outcomes, such as a ticket thread together with the fix that closed it, in enough depth to interest an AI developer. A record set can rate high on one and low on the other, so note both.
- Legal hold: is any part of this record set subject to a litigation hold, regulatory inquiry or audit request? If yes, archive it intact and stop.
- Retention schedule: does the company's schedule, or a contract or regulation it follows, require the records to be kept? If yes, keep or archive them.
- Rights status: given customer contracts, privacy notices and vendor terms, may the company keep and reuse these records?
- Likely value: are the records still needed to run the business, and do they hold linked history an AI developer might license?
The decision tree as a table#
The decision tree reduces to a handful of answer patterns, and most record sets fall into one of the rows below. Write the answers and the resulting outcome next to each record set in the consolidation plan, because the record of why a set was archived or deleted matters as much as the outcome when a regulator or an acquirer later asks.
| Legal hold | Retention required | Rights to reuse | Likely value | Outcome |
|---|---|---|---|---|
| Yes | Any | Any | Any | Archive intact and revisit when the hold lifts |
| No | Yes, in active use | Any | Operational | Keep in the surviving system |
| No | Yes, not in active use | Unclear | Any | Archive and park for a rights review |
| No | Yes or no | Clear | Linked history | Archive, then assess a prepared subset for licensing |
| No | No | Clear or unclear | None | Delete with logged approval |
| No | No | Contract requires return or deletion | Any | Return or delete as the contract requires |
Why route by record set, not by system#
Routing by record set matters because one system usually holds records with very different obligations. A field service platform can contain customer contact details, job notes, invoices, technician photos and recorded calls, and each carries its own retention period and rights position.
A system-level decision tends to fail in one of two directions. Either everything is kept because one record set must be, which carries personal data forward for no reason, or everything is deleted because most of it looked stale, which destroys linked history that had real value. Splitting the system into record sets avoids both.
Check for deletion that is already automated before routing anything. Zendesk admins, for example, can create ticket deletion schedules that keep deleting archived tickets matching their criteria, and deleted tickets cannot be restored. Pause rules like these in every legacy system until each record set has an outcome.
Keep the identifiers that tie record sets together. A ticket archive without customer or product IDs, or a job archive without invoice references, loses most of its value for audits and for licensing alike.
Illustrative: a software holdco retires three help desks#
Illustrative: a fictional vertical software holding company consolidates three acquired businesses onto one help desk. The legacy systems are a Zendesk instance, a Freshdesk instance and an on-premises ticketing tool from the oldest acquisition.
The portfolio CTO routes each record set. Open tickets are migrated and kept. Closed tickets with resolution notes and linked Jira issues are archived in full and flagged for a licensing assessment. Call recordings in the on-premises tool fall under a legal hold from a customer dispute, so they are archived intact. Chat logs from one instance have no retention requirement, and the rights review finds the acquired company's privacy notice limits their reuse, so they are deleted with the general counsel's logged approval.
Because the archive kept ticket, customer and product IDs, the later licensing assessment works from metadata alone, and the prepared subset comes from the archive without reopening any legacy system.
Who owns each routing decision?#
Each routing question has a natural owner, and the consolidation plan should name them before the cutover date is set. Do not let the systems integrator or the new vendor decide; their job is the migration, while the company decides what the records become.
| Question | Owner | Input needed |
|---|---|---|
| Legal hold | General counsel or outside counsel | Open matters, preservation notices, audit requests |
| Retention schedule | CFO or controller | Company schedule, contract terms, applicable regulations |
| Rights status | Counsel with the business owner | Customer contracts, privacy notices, vendor terms, acquisition agreements |
| Likely value | Portfolio CTO or CIO with operations leads | How records are used, how record sets link, depth of history |
| Final sign-off | Company CEO | The completed routing table |
How SourceX approaches legacy archives#
SourceX works on the license outcome only, and only once the legal hold and retention questions have answers. The Supply step of the SourceX five-step transaction reviews archive metadata, such as record sets, date ranges and linking IDs, without receiving files, and the Rights step checks the contracts and notices the records were collected under.
Large archives stay in the company's own storage or ship on encrypted drives; SourceX does not host multi-terabyte datasets. The supplier approves the subset before anything is released, and the SourceX Evidence Packet records the provenance and permitted use of what was licensed.
Frequently asked questions
Can archived data be licensed years after the system is retired?
Yes, if the archive kept records complete and linked and the rights position still allows it. A clean export with attachments, IDs and the terms that governed collection can be assessed long after the original system is gone. An archive reduced to flat reports or PDFs is much harder to use.
What format should a legacy archive use?
Use open, structured formats that keep relationships between records, such as database dumps or CSV and JSON exports with IDs, plus original attachments in native formats. Store a data dictionary describing tables and fields beside the export. A proprietary backup that needs the old software to open it makes a poor archive.
Is deleting legacy data ever the right call for value?
Sometimes. Records with no operational use, no retention requirement and no right to reuse carry storage cost and privacy exposure without upside. Deleting them, with logged approval, reduces risk and makes the remaining archive easier to govern. The mistake is deleting before the routing questions are answered.
Who should approve deletion of legacy records?
The company's CEO should sign off on a routing table that counsel and the CFO have reviewed. Log each approval with the record set, the date and the reason, so the company can later show that deletion followed its schedule and did not sidestep a hold or a pending request.
What if the old vendor contract ends before the routing is finished?
Negotiate a short extension or a read-only access period, and take a full export of every record set immediately as a safety copy. Routing can continue from the export. Losing admin access before the export is the one outcome that removes every option at once. Check the vendor's post-termination rules early: Atlassian's documentation, for instance, says data from a deactivated paid cloud site is kept for 60 days and then permanently deleted.
Sources
- Zendesk admins can create ticket deletion schedules that delete archived tickets after a set period. Deleted tickets cannot be restored, and the schedules keep deleting any tickets that match their criteria. Source
- Atlassian's support documentation states that after a cloud site is deactivated, data is retained for 15 days for trials and 60 days for Free, Standard, Premium or Enterprise plans, and reactivating within this retention period restores the product's data and preferences. Source
Related resources
- InsightCan a distributor license its pricing and quote history?
- InsightHow to design a data retention policy that keeps records licensable
- QuestionShould companies sell or license their data?
- QuestionDo AI labs buy legal documents?
- InsightShould you delete old data or keep it for AI?
- SolutionEnterprise data: the records of how organizations actually work
See if your company qualifies
A short company assessment. No data uploads are needed.