Use caseInsurance and banking
Two insurers merged eighteen months ago. Their claims teams share a floor, yet taking over the other side's file still means asking for an email.
Both sources are scanned with the same measuresOne filing plan, signed by both sidesRights are mapped once and never widened
6.6 million documents in FileNet, 2.9 million in SharePoint, described with the same measures. 18,240 policy numbers exist on both sides.
source · discovery scans, both sourcesSome documents need a person · each one is named, per source
The volume is manageable. What slows everyone down is that the two companies never filed the same way: different classes, different folder logic, policy numbers that overlap, security groups in two directories. Copy both into one repository as they are, and you get two filing plans under one roof.
Decisions come before any document moves. Both sources are scanned with the same measures, the business signs one filing plan, each source gets its own mapping onto it, and the colliding numbers get a rule. Then the content moves, one line of business at a time.
The work runs in a fixed order, and skipping a step costs a file: measure both sides the same way, agree one plan, settle rights and numbers, then move and reconcile each source on its own. Ellen and Stefan start with the plan, Rania settles the rights, and Lucas ends up opening a claim from the other company.
Both sources are scanned first, then the two filing logics are laid side by side and merged into one plan.
Ellen Brody, the group content architect, points the operations engine at the FileNet object store of Arvelle and at the SharePoint sites of Corbin. The same scan runs on both, and nothing moves.
Stefan Kowalski, the records manager, runs three workshops with the claims and policy managers of both companies. The shared plan files by line of business, then by policy, then by claim.
Where the two sides named the same thing differently, the workshop picks one name and records the old ones as aliases. Each ruling goes into a decision log.
Each source gets its own mapping onto the same target classes. FileNet properties map one to one in most cases. SharePoint needs more rules: the claim number sits in a column on some libraries and only in the folder name on others.
Ellen writes a rule that reads the folder name when the column is empty.
Two directories, two security models, two numbering schemes. Each could expose or lose a file, so all three are settled before the pilot.
Rania Mansour, the identity and access administrator, takes the groups that grant access in each source: FileNet rights on classes and folders, SharePoint permission groups on sites and libraries. She maps both onto the shared roles of the repository.
The inventory also listed the folders whose rights differ from what they inherit. Each one is mapped explicitly.
Both companies numbered policies with eight digits, and 18,240 numbers exist on both sides for different customers. Stefan and the policy teams agree on a rule: a group policy key made of the entity code and the old number.
The old number stays as a searchable property, so adjusters can still type the number a customer gives on the phone.
Ellen runs the pilot on the 2023 auto claims, both companies in the same batch. Each source is extracted through its own connector, transformed with its own mapping and loaded into the same staging space.
The counts are shown per source, so a problem on one side does not hide in the total.
The pilot is reconciled per source, an adjuster tries it on real files, and the remaining lines move in waves.
The reconciliation report has one part per source. Each document has its own line, with its FileNet or SharePoint identifier next to the new one. Ellen shares each part with the former IT team that knows that data best.
The Corbin claims team gets the list of 752 files. For most of them an adjuster knows the claim from the letter itself, and fills in the number in the list. The failed objects are reprocessed with that value, without rerunning the pilot.
Lucas Ferreira used to work at Arvelle. He takes over claim CLM-23-008112, opened at Corbin. He types the old policy number he was given on the phone, gets both policies that carry it, and opens the Corbin one.
Its adjuster report sits in the claim folder, filed the same way as his own.
With the pilot signed, Ellen and Stefan plan the remaining lines: homeowners, health, then policy contracts and archives. Each wave reuses the mappings, the group key rule and the rights mapping, and gets its report per source.
The two old systems go read-only line by line, as their content is signed.
A batch is judged twice, once per source, and lands in one of three states. The state decides whether the next line of business moves.
Every document is filed under the shared plan, with rights no wider than in its source.
After that: the wave is signed and the source goes read-only for that line.A missing claim number, an unmapped group. Each one is named, with the source it came from.
After that: the team that knows that source fixes it, and only those documents run again.The counts do not agree, or a rule filed many documents in the wrong place.
After that: the next wave waits until the cause is found and a person decides.It does not choose the filing plan, the group key or who may read a bodily injury file. Across the ten steps on this page, every one of those choices is made by the group's team, logged and signed.
The engine scans, applies each source's mapping and reports per source. The control is in the decisions made before the pilot, so it holds for nine million documents.
Drawn from the Arvelle and Corbin consolidation. Two other companies, with other sources and another plan, will draw their own.
Where these numbers come from. They are counted on the scenario described on this page, with example data. They are not a customer result.
Two companies share one repository only when IT, records and access management agree. Each of them signs for something different:
The merger is signed, and two platforms still run side by side with their own costs.
Two filing logics, two retention habits, and your signature on both.
Two directories, where one mistake opens files to the wrong company.
With one line of business. Scan both sources, run the first workshop on its filing plan and pilot it, while every other line stays where it is.
Both old systems keep running until their content is signed, so the teams work as usual while the plan is built.
Let's talk
Send us the class list and the folder logic of each entity. We will scan both with the same measures and prepare the first workshop on a shared plan.