Ending a deadlock on a $25M feature

A new feature for one of Mastercard’s fraud products carried a revenue target of about $25M. Product and engineering couldn’t agree on what to build first, and after three months they were only speaking through scrum leads. I didn’t own the product. I proposed a workshop anyway, and it ended with the two leads presenting one plan together.

Situation

Acquiring banks wanted finer control over fraud thresholds, like holding newly onboarded merchants to stricter rules for a while. Product and engineering had debated the MVP for three months with almost no customer input. A week before product’s deadline, engineering presented a different solution as if it were already decided. After that the two teams only talked through their scrum leads. Morale was about what you’d expect.

My job was portfolio strategy, how each product served customers across the set. Nobody was wrong exactly. They were arguing from assumptions nobody had tested with a customer.

What I did

I flagged it to leadership and proposed a two-day, in-person workshop with VPs from product, engineering, commercialization, technical product and sales. On paper I was there to ground the MVP in customer outcomes. The real goal was to lock scope and lower the temperature.

Beforehand I collected every spec, brief and engineering proposal and merged them into one rough service blueprint: user flows, pages, components, sales touchpoints and system interactions. I sent it out ahead of time. Everyone disliked it, which was useful. It gave the room permission to start over.

We rebuilt from the customer journey, starting with what an acquiring bank needed to do. Every idea went on the wall, competing ones side by side, and old grievances went in a parking lot so they were heard without taking over. Then we plotted the conflicts on two maps: impact against certainty, and ease of use against feasibility. High-impact, low-certainty items became research questions. Easy-to-use, hard-to-build items became tech spikes. What was left, the certain and feasible, became the MVP.

I cleaned up the blueprint that night. The two open items were settled in the first half hour of day two.

What changed

Scope was locked on day one, with a beta for 10 customers who had asked for the feature. The product and engineering leads, who hadn’t been on speaking terms for months, co-presented the plan to about 60 people. The feature stayed on its delivery timeline. The SVP told me he hadn’t been able to get those two in the same room. For months afterward, the teams pulled the blueprint back up whenever an old idea crept back in.

What’s still hard

This only works if a sponsor lets you put a draft in front of senior people that’s wrong on purpose.