IT Procurement Platform
Sales had already sold this customer on Power Platform before I arrived. What was actually being tested was the platform itself — their first enterprise Power Platform project.
The challenge
They knew their own procurement process well but had no journey maps, no view on premium versus standard licensing, and no idea what integrating with Dell and their ITSM platform would actually involve. IT procurement sponsored it.
The real brief sat underneath the stated one. This was their first enterprise Power Platform project — what was actually being tested was the platform. That framing shaped every decision that followed: keep the commitment low enough that a disappointing result cost them very little.
The approach
Licensing was the genuine design constraint, not technology. Premium connectors license per user, and licensing every employee who might order a laptop would have killed the business case before it started. So I designed the first release on SharePoint with all third-party integration brokered through a service account — moving the premium licensing requirement off the user base entirely. A single premium licence covered the integration layer and end users stayed on standard. Effectively zero incremental licensing cost for a business-wide capability.
I ran the journey mapping the engagement hadn't come with, investigated the integration surface across both vendor platforms, worked with their data team to understand their data estate, and built the concept out to wireframes before anyone committed.
Architecture
The built solution synchronises Dell's catalogue daily, enforces conditional product relationships and ordering limits, routes approvals, places the order back into Dell, raises a ticket in Halo so the service desk knows a device is coming, polls order status daily, and writes serial and tracking numbers back onto the ticket.
Because it ran as a proof of value, a post-implementation review was built in with no commitment attached. That review surfaced SharePoint's data modelling limits against a multi-request-type model and a reporting requirement it couldn't serve — alongside early evidence the platform had earned more investment. Moving to Dataverse meant premium licensing across the user base. The data model became a star schema — a central request table holding what every request shares, with each request type as its own related table, so adding a new request type became adding a table and its relationships.
Outcome
Order cycle time went from two weeks to around four hours. The solution is still running, has expanded into mobile services, telecommunications, software and printers, and was published as a customer case study.