Oracle's CEGBU portfolio runs on three colliding metrics (per user, per record, and per project value) that quietly inflate cost as contractor headcount rises. This guide names where the risk and leverage sit, and tells you exactly what to do before your next renewal or audit.
Oracle's CEGBU portfolio runs on three colliding metrics (per user, per record, and per project value) that quietly inflate cost as contractor headcount rises. This guide names where the risk and leverage sit, and tells you exactly what to do before your next renewal or audit.
Oracle's Construction and Engineering Global Business Unit (CEGBU) is where most licensing teams stop paying attention after they have counted their Primavera P6 desktop seats. That is a mistake. The portfolio that matters at scale (Primavera Unifier, Aconex, P6 EPPM, and Primavera Cloud) runs on three fundamentally different licensing metrics that do not reconcile against each other, and each one behaves badly as a project estate grows. Per user, per record, and per project value are not interchangeable, and Oracle's contract language ensures the buyer absorbs the friction between them.
In 25 years negotiating CEGBU deals, the single most reliable pattern I have seen is this: the customer sizes the deal against full-time employees, signs, and then discovers at audit that contractors, joint-venture partners, and external subcontractors were consuming entitlement the whole time. Project-heavy estates are contractor-heavy by definition. That is precisely where the meters spin fastest and where Oracle's audit remedy hurts most. This guide decodes each metric, quantifies where the money leaks, and gives you the moves to close the gaps before Oracle finds them.
The core problem with CEGBU licensing is that Oracle does not use one metric across the portfolio. It uses three, and they were designed for different eras and different products. When you run an integrated project estate (P6 for scheduling, Unifier for cost and controls, Aconex for document collaboration), you are simultaneously counting users, counting records, and tracking project value, and each of those counts can grow independently. A single project ramp-up can trigger cost increases on all three at once.
| Product | Primary metric | Licensing models | Where cost inflates |
|---|---|---|---|
| P6 EPPM | Application User / Hosted Named User | Perpetual + support, or SaaS | Contractor named users, indirect API access, peripheral access |
| Primavera Unifier | Per user, by module | Perpetual (~$10,450/user list) or SaaS | Module sprawl, records used outside licensed context |
| Aconex | Project Value (PVA) | Subscription only | Rising project value true-ups, guest users counted as named users |
| Primavera Cloud (OPC) | Per user per module | Subscription only, 5-user minimum per module | Minimum seat blocks, module stacking |
Notice what happens if you have a program of ten projects, each with fifty external subcontractors touching documents in Aconex and viewing schedules in P6 EPPM. The Aconex Project Value rises as scopes expand, triggering true-ups. The P6 EPPM named user count rises as contractors log in. If those contractors also touch Unifier records, you owe a third license. Integration between the products does not merge the counts; it multiplies them. This is the structural reality that makes CEGBU one of the more expensive corners of the Oracle estate to run sloppily.
Integration does not merge your licenses. A user who touches P6 and Unifier needs two, one per product, and Oracle's audit team knows exactly how to prove it.
P6 EPPM is licensed per user, but Oracle has quietly evolved the definition. Historically P6 licenses were sold as named or concurrent. Today Oracle's commercial model applies to Application Users, meaning each license maps to an individual person with access, not to a floating pool of concurrent sessions. That change matters for project estates because concurrency used to let you share a smaller pool across shifts and contractors; named user counting removes that cushion entirely.
A useful piece of value: a P6 EPPM license includes both Primavera P6 Professional and the Primavera P6 Web Access client, so you are not double-buying the desktop and web experiences. On older price lists, P6 Enterprise (EPPM) carried a base license around US$2,750 per application user with annual license and support around US$605, while the standalone P6 Professional perpetual sat near US$3,520 per user. Under the UK G-Cloud 14 framework, Oracle listed P6 EPPM Cloud Service at £220 per Hosted Named User per month with a 25-user minimum, plus £36 per Hosted Named User per month for P6 EPPM Web Services. Treat these as reference points, not current list; Oracle no longer publishes an authoritative public price list and its last formal CEGBU Global Price List dates to November 2016.
The volume discount curve under the G-Cloud framework is worth memorizing as a negotiation anchor: 10 percent at 101 to 200 users, 15 percent at 201 to 500, 20 percent at 501 to 1,000, and 25 percent at 1,001 and above. Note how modest these are. Oracle's public discount tiers are a floor, not a ceiling. In practice we routinely negotiate materially deeper on CEGBU renewals when the customer holds a credible alternative, but you have to build that leverage before you ask.
Here is where P6 EPPM becomes an audit landmine. Per Oracle's P6 EPPM Licensing Information Manual (v24, September 2025), any user who accesses data originating from, or submits data integrating into, P6 EPPM through a third-party application using automated processes (including Access Points) must be licensed under the Primavera P6 Peripheral Access Program. In plain terms: if you have built an integration that pushes schedule data into a BI tool, an ERP, or a custom portal, and a person consumes that data downstream, Oracle's position is that person needs a P6 license too. This is the classic indirect access problem, and CEGBU estates are full of integrations.
The second landmine sits in the document repository. The v25 manual states that the P6 EPPM repository is valid only for Primavera-originated documents. Create repositories, folders, or workspaces manually outside the Primavera applications, and you trigger full-use licensing on the underlying components. We cover the related database exposure in depth in our analysis of the Oracle Database hiding under P6 EPPM and its restricted-use limits, because the same repricing mechanic applies to the embedded database engine that ships with the product.
Unifier is licensed per user, but critically it is sold in modules. Oracle separates functions such as Unifier Project Controls and Unifier Facilities and Asset Management, and each module is licensed per user independently. The buyer advantage here is real: you can purchase licenses only for the specific module you intend to use, so a team doing cost controls need not buy the facilities management module. The buyer risk is equally real: module sprawl. As Unifier gets adopted across departments, users end up needing multiple modules, and the per-user cost stacks.
On list pricing, one reseller reference puts Unifier at roughly US$10,450 per user license with annual support at 22 percent, around US$2,299 per user per year. That is a premium product, and it means every unnecessary Unifier named user is a five-figure entry over a typical contract term. Unlike Aconex, Unifier supports both perpetual and subscription (SaaS) licensing, which gives you a genuine strategic choice at renewal. We walk through that choice in detail in our companion piece on the Unifier named user metric, modules, and the records trap.
The records trap is the Unifier equivalent of the P6 repository trap. Unifier is a records engine at its core (business processes, forms, cost sheets), and Oracle's licensing language ties the entitlement to records used within the licensed context. Use those record structures outside the licensed module scope, and you risk a full-use reprice on the broader deployment. The lesson is the same across CEGBU: the metric is not just the user count, it is the user count operating strictly within the entitled context.
At roughly $10,450 per user list plus 22 percent support, every unnecessary Unifier named user is a five-figure liability across the term. Audit your module assignments quarterly.
Aconex is the most opaque of the four products, and the most dangerous for contractor-heavy estates. It is subscription only (no perpetual option exists), and Oracle does not publish any pricing; quotes are available only on request. On large projects, Aconex is usually priced against Project Value rather than user count, which sounds simple until you read the consumption mechanics.
Per the Oracle Aconex Cloud Service Descriptions (January 2025), Project Value is consumed toward the Project Value Allowance (PVA) on a straight-line basis over the delivery duration. If a project extends beyond the Services Period end date, only the proportion of Project Value prior to expiry consumes PVA. That is manageable. The trap is the next clause: on any increase in Project Value during the Services Period, the customer agrees to execute an order with Oracle to reflect that increase. In construction, project values rise constantly through change orders, variations, and scope creep. Every material increase becomes a contractual trigger for a new order, and Oracle holds all the pricing leverage at that moment because your project is already live and mid-delivery on Aconex. You cannot credibly walk away from your document management platform while a project is running.
Rising project value is a contractual true-up trigger in Aconex. Oracle prices the increase while your project is live and you have zero exit leverage.
The second Aconex bomb is the guest user rule. In the Aconex Enterprise model each user can access any project managed under the order, and Aconex Guest Users are explicitly considered Hosted Named Users. Read that again. On a document collaboration platform whose entire purpose is bringing external parties (architects, engineers, subcontractors, suppliers) into a shared workspace, every one of those external guests counts against your license entitlement. This is the single largest source of contractor-driven cost inflation in CEGBU, and it is invisible until you count actual accounts against what you purchased. Our deep dive on Aconex per-project versus enterprise licensing works through which model minimizes this exposure for your project profile.
Oracle Primavera Cloud (OPC) is Oracle's newer subscription platform, and it is meaningfully cheaper per user than legacy P6. Most implementations start around US$100 per user per month with minimum seat requirements, roughly one-third the cost of a full Primavera P6 Professional license. A common structure is a five-user annual starter pack at US$130 per user per month, with additional licenses at US$1,560 per year each. Another reference cites US$7,800 per year for five users. Support and updates are included in the subscription, so unlike perpetual P6 you do not pay a separate 22 percent maintenance line.
The key OPC constraint is the minimum: you must purchase at least five user licenses per module. That per-module minimum matters if you adopt several OPC modules, because the floor multiplies. For a small team using one module, OPC is genuinely cheap. For a broad rollout across many modules, the minimums stack and the economics shift. Whether OPC beats P6 EPPM at your scale is a real calculation, not a slogan, and we run the break-even math in Primavera Cloud vs P6 EPPM: which model costs less at scale.
A word of caution on OPC pricing figures: the public numbers above come from resellers and regional frameworks, not a current Oracle list. Oracle deliberately keeps OPC pricing off any authoritative public list, which means benchmarking against peers and holding competitive quotes (from Autodesk and others) is the only way to know whether your OPC quote is fair. For context on the competitive landscape, our guide to Autodesk Construction Cloud licensing lays out the per-user, per-project, and Flex token models that OPC and Aconex compete against.
Every CEGBU trap discussed above funnels into one place at audit: the reprice. When an Oracle audit finds a restricted component used outside its licensed context (a repository built manually, an integration feeding data downstream, a module used beyond its scope), Oracle's position is that the use was never licensed at all. The remedy is not a small adjustment. It is full-use licensing at list price for the whole deployment, frequently with back support added on top.
Here is the counterintuitive part that catches sophisticated buyers off guard: the cheaper the restricted grant was, the larger the gap to the full-use reprice. A heavily discounted restricted-use P6 license looks like a bargain when you sign it. But if the audit reclassifies it as full-use, the gap between the discounted restricted price you paid and the full-use list price Oracle now demands is enormous. Discounting the restricted grant does not reduce audit exposure; it amplifies it. This is why we tell clients never to celebrate a deep discount on a restricted-use CEGBU license without first mapping exactly what the restriction permits.
| Trap | Where it lives | Audit remedy Oracle claims |
|---|---|---|
| Peripheral / indirect access | P6 EPPM integrations, Access Points | Full-use licensing for all downstream data consumers |
| Manual repository creation | P6 EPPM document repository | Full-use license on underlying components |
| Guest users | Aconex Enterprise | Guests reclassified as Hosted Named Users, retroactive |
| Module use outside scope | Unifier records | Full-use reprice on broader deployment |
| Rising project value | Aconex PVA | Mandatory true-up order at Oracle's pricing |
| Uncounted contractors | P6, Unifier, Aconex | Additional named users plus back support |
The scope of what Oracle examines across a CEGBU estate is broader than most licensing teams expect, spanning user account logs, integration endpoints, and project value records. We map the full examination surface in what an Oracle audit examines across the Primavera and Unifier estate. If you have any of the traps above active in your environment, read it before Oracle's letter arrives, not after.
The most common CEGBU compliance failure is beautifully simple: customers buy licenses for full-time employees and forget contractors and part-timers. Oracle audits routinely find customers with more actual users than purchased licenses. In project-heavy estates this is not an edge case, it is the norm, because construction and engineering work is delivered largely by external parties who all need access to the schedule, the cost controls, and the documents.
The counting rules depend on the metric definition in your contract. Under Hosted Named User, you must ensure no unlicensed individual has access, full stop; every account is a license obligation. Under Hosted Employee, all contingent workers and outsourced staff must be included in scope, which explicitly pulls contractors into the count. And in Aconex, as covered above, guest users are Hosted Named Users by definition. There is no metric under which external contractors are free.
This is the single biggest budgeting error we see. A program manager sizes the CEGBU deal against the internal team, and the joint-venture partners, subcontractors, and consultants who log in daily are simply never counted. Then the project scales, the account list grows silently, and the audit finds a gap that is measured in hundreds of users. Our detailed methodology for getting this right is in counting external contractors in Primavera and Aconex deployments, and it should be mandatory reading for anyone budgeting a new CEGBU program.
There is no metric under which external contractors are free. Size the deal against your total access population, not your payroll.
For CEGBU term (time-limited) licenses, Oracle's list pricing follows a predictable ramp against perpetual list: 20 percent for one year, 35 percent for two years, 50 percent for three years, 60 percent for four years, and 70 percent for five years, with support running at 22 percent of the perpetual license fee. Understand what this curve tells you. A five-year term at 70 percent of perpetual is almost as expensive as buying perpetual outright, minus the residual asset value. Term licensing only makes economic sense for genuinely short-duration needs, typically one to two years, where you pay 20 to 35 percent and avoid a perpetual commitment.
The 22 percent support line is where perpetual CEGBU estates bleed over time. On Unifier at roughly US$10,450 per user, that is around US$2,299 per user per year in support alone, indexed upward year over year. For a large perpetual estate, support fees eventually exceed the original license cost. This is precisely the calculation that pushes mature Primavera and Unifier customers toward third-party support, where the annual cost typically drops by half or more. We work through the decision framework, including the audit-rights and version-support tradeoffs, in Primavera and Unifier: should you move to third-party support.
CEGBU compliance is not a project, it is a discipline. Because the meters (users, records, project value) all move continuously in an active estate, an annual true-up is not enough. Best practice, which we endorse from repeated audit-defense experience, is to track every individual with access to Primavera systems, routinely audit user accounts across P6 EPPM, Unifier, and Aconex, remove accounts for departed staff and expired project partners, and mandate a quarterly review of all Primavera user accounts against licenses held.
None of this is glamorous, but it is the difference between a routine renewal and a seven-figure audit surprise. The customers who get burned are always the ones who treated CEGBU as a set-and-forget purchase. In an estate where project value and contractor headcount move every month, static entitlement management is a guaranteed loss.
First, before any new purchase or renewal, count your true access population across all four products, including every contractor, guest, and partner. Size the deal against that number, not your headcount. Underbuying to save on the initial order is the most expensive false economy in this portfolio, because the gap resurfaces at audit as a full-use reprice with back support.
Second, decode your contract metrics precisely. Know whether each product is Hosted Named User, Hosted Employee, per-module, or Project Value, and know exactly what context restrictions attach to any discounted restricted-use grant. Remember that the cheaper the restricted grant, the larger your audit exposure, so never accept a discount on a restricted license without documenting the boundary in plain language.
Third, model the metric choices deliberately. OPC versus P6 EPPM, Unifier perpetual versus SaaS, Aconex per-project versus enterprise: each has a break-even that depends on your user count, module spread, and project duration. Do the math with real numbers before Oracle frames the options for you. Hold a credible competitive alternative (Autodesk in particular) because Oracle's published discount tiers of 10 to 25 percent are a floor, and the deeper concessions come only when you have leverage.
Fourth, install the quarterly governance discipline described above and never let it lapse. Fifth, if your perpetual support bill has grown past the point of value, run the third-party support numbers seriously. For a broader view of Oracle's licensing tactics across the estate, from the audit playbook to the metric definitions Oracle applies elsewhere, the Oracle Java per-employee licensing analysis shows the same repricing and definition-shift patterns playing out in a different product line. The tactics rhyme across Oracle's portfolio, and recognizing them early is your best defense.
The CEGBU portfolio is not unmanageable, but it is unforgiving of casual buyers. Three metrics, four products, and a contractor population that never stops growing add up to an estate where discipline pays for itself many times over. Count everything, decode every restriction, model every choice, and govern quarterly. Do that, and the audit letter becomes a formality rather than a crisis.
Aconex is subscription only and, on large projects, is priced against Project Value rather than user count, consumed against a Project Value Allowance (PVA) on a straight-line basis. Two clauses drive cost inflation: any increase in project value during the services period obligates you to execute a new order at Oracle's pricing, and Aconex guest users are explicitly counted as Hosted Named Users. In a document platform built for external collaboration, that guest-user rule makes contractor headcount a direct licensing cost.
Yes. Integration between products does not merge licensing. A user who accesses data in both P6 EPPM and Unifier requires two licenses, one for each product. This surprises many buyers who assume an integrated platform means unified licensing, and it is a common audit finding in connected CEGBU estates.
Oracle's P6 EPPM licensing manual requires that any user who accesses P6-originated data, or submits data integrating into P6 EPPM, through a third-party application using automated processes (including Access Points) must be licensed under the Primavera P6 Peripheral Access Program. In practice this means downstream data consumers reached through your integrations may need P6 licenses, which is a classic indirect-access exposure that Oracle audits actively pursue.
Often yes on a per-user basis. OPC typically starts around $100 per user per month, roughly one-third of a full P6 Professional license, with support and updates included so you avoid the separate 22 percent maintenance line. The catch is the minimum of five user licenses per module, which stacks if you adopt several modules. Whether OPC wins depends on your user count and module spread, so run the break-even before deciding.
When an audit finds a restricted component used outside its licensed context, Oracle's remedy is full-use licensing at list price for the whole deployment, often with back support. The gap between the discounted restricted price you paid and the full-use list price is the exposure, so the deeper the original discount, the larger that gap. Never accept a discount on a restricted grant without documenting exactly what the restriction permits.
Count all of them. Under Hosted Named User you must ensure no unlicensed individual has access. Under Hosted Employee, all contingent workers and outsourced staff must be in scope. In Aconex, guest users are Hosted Named Users by definition. Size any deal against your total access population including contractors, partners, and guests, because Oracle audits routinely find more actual users than purchased licenses in project-heavy estates.
Oracle Cloud at Customer enterprise licensing framework. Buyer side framework across OCI Dedicated Region, Exadata Cloud at Customer, autonomous database.
Gated with a work email on the download page. No sales follow up you did not ask for.
Get the White Paper →
500+ enterprise clients. 11 vendor practices. Industry recognized. One conversation can change what you pay for the next three years.
One buyer side briefing a week. Renewal signals, audit moves, and the levers that work. No vendor spin.
