Skip to content
Marcus Gonzalez

Prime — Pricing Platform

Prime is live for 1,700 advisors. The pricing tool now handles comparisons they once had to do in Excel.

The old tools put every possible price in one table, so advisors copied the few that mattered into Excel. Prime follows the way they sell: shortlist products, then compare the upfront cost with the monthly payment. Keeping the comparison inside Prime lets advisors stay focused on the customer instead of switching tools mid-call.

Fig. 00

From catalog to conversation.

The table — show everything
Pricing: full catalogone table

expand one row, lose the rest

The sale — shortlist, then compare
Prime: shortlist, then comparetwo moves
compare upfront cost vs monthly payment
Pay pointslower rate, lower monthly
No pointshigher rate, cash in hand
the comparison stays in the conversation
Before and after: a dense pricing table that shows the whole catalog at once becomes a two-part experience — shortlist the products, then compare what the customer pays upfront against the monthly payment.
Context01

Prime had to replace every capability of a business-critical pricing platform. Functional parity did not require preserving the interaction model advisors already worked around.

Chase depended on one vendor to produce every customer-facing loan price across marketing offers, digital self-service, and loan production. If that platform failed, Home Lending could not market offers, quote rates, or price loans. Prime had to reproduce every capability, but my team could redesign how advisors used it.

The open design question was how advisors would use the tool.

Fig. 01

Replacing the dependency was fixed; reconsidering the experience was the opening.

  1. The dependency3 surfaces

    One third-party platform produced every customer-facing loan price.

  2. The riskbusiness-critical

    Chase did not control a platform Home Lending could not operate without.

  3. The mandatethe open question

    Replace the functionality one-to-one. How advisors use it was left to design.

Three stations on one line: one third-party platform produced every customer-facing price; Chase could not operate without a platform it did not control; the mandate was to replace the functionality one-to-one, which left the experience as the open design question.

I led four designers across the United States and India and facilitated the discovery sessions that translated advisor needs into requirements for the new pricing engine.

Problem02

The inherited interaction model organized pricing around the full catalog instead of the decisions advisors made with customers. Advisors moved the few relevant choices into Excel, slowing the conversation and splitting their attention.

Each loan product branches by term, amortization schedule, backing agency, and point option. The interface treated completeness as usefulness: it displayed every combination but gave advisors no practical way to narrow and compare the choices that fit the customer.

Fig. 02

Three generations of pricing tools preserved the same interaction model.

  1. 01the vendor's tool
  2. 02the internal tool
  3. 03the direct predecessor

the same show-everything table

1

advisors kept Excel open beside all three

1

Expanding one row to see its point options pushed the other candidates out of view, so advisors retyped the relevant choices into a spreadsheet and compared them there.

Three generations of pricing tools — a third-party platform, an internal tool, and Prime's direct predecessor — all resolve into the same dense show-everything table, and advisors built the usable comparison in Excel beside all three.
Research03

Research revealed a two-stage sales process hidden by the full catalog. A one-week sprint tested whether an interaction built around that sequence could replace the table.

Advisors first narrow the catalog around par, the rate where the customer neither pays nor receives points. They then compare how paying more up front changes the monthly payment. We turned that sequence into prototypes and revised them between tests. By the end of the week, the team had enough evidence to continue with the two-part design.

Fig. 03

One week from inherited assumption to tested model.

  1. 01Inherited assumptionShow every product and price in one table.the table
  2. 02Observed selling modelShortlist around par, then compare cost and payment.from research
  3. 03Prototype and reviseTurn the sequence into a two-step interaction, revised between tests.one week
  4. 04Advisor evidenceAdvisors completed the decision without the full table on screen.human
  5. 05ProceedEnough evidence to continue with the two-part design.decision
Five stages: the inherited assumption to show every product in one table, the observed selling model of shortlisting around par then comparing cost and payment, prototypes revised across the week, advisor evidence that the two-step model works, and the decision to proceed.
Decisions04

Three generations of pricing tools had made the table feel inevitable. I used the sprint to prove a two-step model with advisors and work through its performance constraints with engineering.

Build around the decision, not the catalog

Stakeholders expected another table. I led the team in translating the two-step sales process into product tiles and a side-by-side view that kept the complete decision inside Prime.

Price the likely products before the full catalog

During the sprint, engineering showed that pricing every possible combination could take up to a minute. I worked with the team to price the most likely choices first while the complete catalog loaded in the background, so advisors retained access to every option without delaying the customer conversation.

Fig. 04

Demand concentrates; access does not disappear.

~84%

Conventional fixed-rate loans

~60%

30-year fixed selections

~93%

Selections between zero and two points

Full catalog firstup to a minute
1Common choices firstunder two seconds
1

A minute of loading interrupts a live customer conversation. Prime shows common choices in under two seconds while the complete catalog continues loading in the background.

Three concentration figures from pricing history — about 84% conventional fixed-rate loans, about 60% 30-year fixed selections, about 93% between zero and two points — drawn as normalized proportion fields, above a load-time contrast: the full catalog nearly fills its track at up to a minute while the common choices load in under two seconds.

Prove the new model with advisors

Because the table had strong internal support, a plausible interaction was not enough. We tested the prototype with advisors, who completed the two-step workflow without keeping every price on screen at once.

Looking back05

Everything I led on Prime is live: the pricing experience reached 1,700 Home Lending Advisors during the week of July 6, 2026, and the team is watching adoption and product mix before asking advisors whether they still keep Excel open.

What I would do differently: test the two-part workflow against the show-everything expectation earlier and with the stakeholders who held it, not only with advisors. The advisor validation settled the design; the belief that every price had to stay visible cost us a sprint we could have spent on the comparison itself.

When does showing everything stop being a feature and start being the thing people work around? If you're untangling a tool people have learned to route past, let's talk.

marcus@marcusjg.comLinkedIn

Next caseAdvisor Mobile → UnityPrime replaced a system by matching how people actually work. In the next case, learning how people actually work stopped a funded app before it was built.
What deserves mobileranked
Rate lockmobile
Full application
Document upload