A coordination problem disguised as a dealer problem

Allied Solutions runs RefundPlus, a service that calculates and processes refunds on cancelled GAP and other ancillary auto-loan products. It started as a manual, hand-stitched pilot for a few lender clients. It worked. Then federal regulations changed: lenders were now required to refund customers in every applicable case, on tight timelines — in some states, as little as thirty days. Demand surged past what the manual process could absorb.

Allied came to Harmonic with a sharp brief: turn a bootstrapped pilot into a profitable, automated, scalable service — without breaking the version already running. The complication was structural. RefundPlus moves data and money across four kinds of actor: the lender who originated the loan, the dealer who sold the ancillary product, the provider who underwrites it, and Allied in the middle. Each one held part of the truth. None held all of it.

Phase 1 ran inside-out workshops with Allied's operations, implementation, support, and sales teams, then outside-in interviews with lenders and dealers. The two views did not match. Inside Allied, "the customer" generally meant the lender. Outside, the experience was shaped just as much by dealers and providers — most of whom Allied didn't employ.

A second misalignment surfaced. Inside the company, RefundPlus was treated as a product: something Allied had built and sold. But it was behaving like a service — multi-party coordination, dependency on partners outside Allied's walls, failure modes visible only from the edges.

The working hypothesis going in was that dealers were where refund processing slowed down. The outside-in evidence said otherwise: most processing delays traced to providers — the entities actually cutting refund checks. Dealers contributed a different problem: many didn't know the refund regulations had legal teeth, or how short the timelines now were. Their lack of urgency was real. It wasn't the root cause anyone had assumed.

Phase 2 was collaborative. Rather than design the future state privately and present it, the team built it with the people who would run it: operations, technology, partner management, and leadership across Allied. The workshops produced four artifacts that fit together: a future-state service blueprint; a capability inventory tagging every capability by business area and feature concept; an evolution map in two views (by business capability and by feature concept, because program managers and product owners ask different questions of the same data); and moment briefs for high-stakes moments the workstreams would carry forward.

The blueprint did one thing on purpose that's easy to miss. Wherever the workshop group couldn't resolve a design question — whether Allied should own calculation logic, what data it needed from lenders at origination — the unresolved question stayed on the blueprint as an explicit open item, not smoothed over. The blueprint told the truth about what wasn't decided yet, and named who would need to decide it.

The harder, less visible part of this engagement was helping Allied see RefundPlus as a service in the first place. The shift — from "a product we sell to lenders" to "a coordinated set of behaviors across four kinds of partner, most of whom we don't employ" — didn't happen in a single workshop. It happened across multiple sessions where someone described a current-state problem and the room walked back through the blueprint together and found the actual cause was a step earlier or later in the system than anyone had assumed.

By the close of the engagement, the team's project retrospective named "the client started to think of RefundPlus as a service, not a product" as one of the outcomes. That framing shift doesn't show up on a capability roadmap. But it determines whether the future state can be built at all — because you can't design the coordination layer of a multi-actor service if you're still optimizing for the efficiency of one actor's internal process.