AccuRev user and stream license questions

AccuRev is a version control system licensed by user, yet its defining feature, the stream based architecture, regularly confuses an audit into counting structures rather than people. The AccuRev license question that decides the size of a finding is simple to state and easy to get wrong: who counts as a licensed user, and does a stream ever count as anything at all.

AccuRev came into the OpenText estate through the Micro Focus acquisition, and like most products in that portfolio it is governed by the Additional License Authorizations rather than the OpenText EULA that covers the ECM line. The metric that matters is the user count, but AccuRev environments are organised around streams, the hierarchical containers that hold and promote code, and a stream is not a person. When an audit treats stream counts, workspace counts, or active branch counts as a proxy for licensed users, the finding inflates by the difference between the organisation of the repository and the number of people who actually log in. Because the EULA places compliance squarely on the licensee, the buyer who can show exactly which named individuals use AccuRev, and separate that from the structural objects that exist inside it, controls the only number that should drive the result.

How AccuRev licensing is actually metered

AccuRev is licensed by user, and the central task in any AccuRev finding is to establish a clean, deduplicated list of the people who genuinely use the system. That sounds straightforward, but several categories sit at the edges and each one is a place where the count can drift upward. Provisioned accounts that were created and never used, accounts belonging to people who have left, automation identities that run builds or integrations without a human behind them, and accounts that exist only to support a read scenario all need to be examined before they are accepted into the licensed total. The defensible position counts active human users against the user entitlement, and treats every other category on its own merits rather than sweeping it into the seat count.

Streams, by contrast, are not a licensed unit in the user model at all. They are the way AccuRev organises code: a parent stream, child streams beneath it, and workspaces where developers do their work. A single user typically interacts with several streams, and a stream can persist long after the work inside it has finished. Counting streams, or counting workspaces, and then reading that number as a user requirement is a category error, and it is one of the more common ways an AccuRev finding overstates. The discipline of counting structures as structures and people as people is the same one set out in AccuRev workspace and stream counting.

Where the AccuRev user count overstates

Automation and service identities

Build servers, continuous integration jobs, and scheduled processes often hold AccuRev accounts so they can promote and synchronise code. These are not interactive human users, and whether they require a license at all depends on the entitlement and the definition it sets. Pulling them out of the human user count, and arguing them separately where the authorization is ambiguous, is frequently the largest single correction.

Dormant and departed accounts

User counts inflate with accounts that were created for a project that ended, or that belong to people who have left the organisation. Activity logs that show no sessions over the audited period are the evidence that removes them, the same approach used in reducing a LoadRunner finding with concurrency evidence.

Stream and workspace inflation

The most distinctive AccuRev overcharge is structural, where the count of streams or workspaces is read as a user figure. Separating the two is essential, and the related question of how user counting traps appear across Dimensions is covered in Dimensions CM and RM user counting traps.

How we defend an AccuRev finding under the four Rs

Respond. OpenText gives seven days notice before an audit and the right to copy relevant records. We take over the single controlled channel and ensure the user lists, the stream and workspace inventory, and the activity logs are captured together, because the user argument depends on separating each from the others cleanly.

Reconstruct. We build the effective license position by deduplicating the user list, classifying every account as active human, automation, dormant, or departed, and confirming that no structural object has been counted as a user, all before any vendor measurement script runs.

Rebut. We challenge every line that counts a stream or workspace as a seat, that includes a service identity without authorization, or that retains accounts with no activity. The finding falls by the value of every user the AccuRev model does not actually require.

Resolve. We settle on the verified human user count and, where it serves you, convert forward into an OpenPass agreement that records how AccuRev users are defined and counted, so a later review cannot reopen the stream question.

An anonymised outcome

The reason the user definition matters so much is the remedy that sits behind the finding. On noncompliance the licensee is deemed to have acquired licenses at then current list price, owes back maintenance and support, owes first year maintenance on the new licenses, and reimburses the cost OpenText incurs performing the audit, so every seat removed removes that whole stack of charges, not just the license price. Our anonymised case files show reductions of this kind across the estate, including an insurance ECM seat count finding reduced from $7.2M to $1.6M, a 78 percent reduction once service and dormant accounts were disqualified. An AccuRev user finding responds to the same discipline: separate people from structures, remove the accounts that are not real users, and the invented seats fall away.

Settle the user definition before the count

The lasting point is that in an AccuRev audit the question of who counts as a user comes first, and the stream architecture is a distraction that must be set aside before any number is accepted. A buyer who produces a clean human user list, classifies the edge cases, and shows that no structural object has crept into the seat total removes the largest source of inflation at the outset. To prepare that position, read reconciling ALM entitlements before an audit and see how the vendor assembles its numbers in how OpenText measures ALM in a self assessment. For the full method see our ALM and LoadRunner audit defense track and our complete OpenText audit defense playbook for 2026. If an AccuRev finding has counted streams or workspaces as users, open a case.