Rebuilding a $14B checkout around pricing honesty

Verizon had not redesigned online ordering in fifteen years. The site was a $14 billion consumer division revenue channel running on a stack that predated the iPhone. Cart abandonment was 73 percent. The business needed a number, not a redesign. The job was to find the number, build the case for the work that would change it, and ship the work without breaking the channel that was already running.

I started by mapping the entire customer journey across four sales channels (web, app, kiosk, call center) instead of just the surface I had been hired to fix. Three things showed up in the first month.

First, the checkout was twelve steps. Industry standard at the time was four to five. Every additional step was bleeding ten to fifteen percent of the funnel.

Second, pricing was hidden until the final screen. Promotional discounts, bundle rules, taxes, regional fees, none of it was visible until the customer had already entered their payment information. We were optimizing for conversion at the exact step where the customer learned what the product actually cost.

Third, and this was the hardest one, the team had been measuring the wrong number. Conversion at checkout was the surface metric. The actual problem was diagnostic: customers were arriving at checkout with the wrong cart configuration because the configurator above checkout had taught them to expect a price that checkout then revealed was not the real price. The bleed was not in checkout. It was in the lie checkout was being asked to deliver.

Before converging, the team explored three directions.

Direction A: Optimize the existing twelve-step flow. Keep the structure, reduce friction within each step. A/B test button placement, form field order, microcopy. This was the path of least organizational resistance. I rejected it because the data said the step count itself was the problem, not the content within the steps. Optimizing twelve steps would have produced a faster twelve-step checkout. It would not have closed the gap to the industry standard of four to five.

Direction B: Progressive disclosure checkout. Collapse twelve steps visually into an accordion or single-page layout, but keep the same data collection underneath. This tested well in wireframes with internal stakeholders but failed in user testing. Customers could see all twelve sections at once, which made the checkout feel longer, not shorter. The cognitive load did not change. The presentation did.

Direction C: Structural reduction to four steps with upfront pricing. Kill eight steps. Surface real pricing in the configurator before checkout begins. The riskiest option organizationally because it required every adjacent team to give up ownership of their step. This was the direction I committed to, and the one I had to argue for publicly.

I made three opinionated calls that followed from Direction C.

Reduce checkout from twelve steps to four. Aggressive and against precedent, the kind of thing that gets pushed back on by every adjacent team that owns one of the steps. I argued the case publicly and committed.

Surface real pricing upfront. Show promotional discounts, bundle rules, and regional fees in the configurator, not in checkout. The instinct in telecom was to hide the full price as long as possible. I argued that the cart abandonment was the cost of that instinct.

Decouple the migration. The old checkout was running. Don't replace it. Run the new one in parallel, route a small percentage of traffic, escalate based on a hypothesis-driven testing protocol rather than raw conversion delta. We knew what we believed. We tested whether we were right, not whether we were lucky.

The redesigned checkout shipped to four sales channels: web, mobile web, native app, and assisted (call center reps using the same flow). The work covered the entire funnel from configurator to confirmation, not just the four checkout steps. It included a unified pricing model surfaced through a single component, a cart-aware configurator that prevented the mismatch problem at the source, and controlled migration tooling that let the team escalate from 5% traffic to 100% over six months without taking the channel down once.

[1/5]· The shipped checkout flow: eligibility, plan configuration, review, and confirmation

The 72 percent conversion lift was not a launch-day number. It was the result of a six-month controlled migration with four distinct testing phases.

Phase 1 (5% traffic, weeks 1–4). The new checkout launched to a randomized slice of customers. Conversion was up 28% immediately, but support ticket volume spiked. Customers were completing checkout faster but calling in to confirm their order had gone through. The new flow was so much shorter they did not trust it had worked. We added an explicit order confirmation step with a receipt summary, a step we had originally cut. Conversion held. Support tickets dropped 40%.

Phase 2 (15% traffic, weeks 5–8). The pricing transparency component was the variable. Customers who saw real pricing in the configurator had 34% lower cart abandonment than customers who saw promotional pricing. But bundle customers, the highest-value segment, initially showed lower conversion. Investigation revealed the real price for bundles included fees that the promotional price had been absorbing. We worked with the pricing team to restructure the bundle display so the value was visible, not just the cost. Bundle conversion recovered and exceeded the control by week 7.

Phase 3 (40% traffic, weeks 9–16). At scale, two new patterns emerged. First, mobile web conversion lagged desktop by 18 points. The four-step flow was designed for a desktop viewport. We redesigned the mobile checkout as a single scrolling page with sticky pricing, a structural deviation from the desktop flow that tested 22% better than the direct responsive port. Second, the assisted channel (call center) needed a different information hierarchy. Reps were reading the same screens customers saw, but they needed the data in a different order. We built a rep-facing view that mirrored the customer flow but surfaced eligibility and pricing data upfront.

Phase 4 (100% traffic, weeks 17–24). Full migration. The team shifted from testing to monitoring. We instrumented every step transition and set up anomaly detection on three metrics: step-to-step drop-off, time-to-completion, and pricing-mismatch rate (customers who saw one price in the configurator and a different price in checkout). The mismatch rate, which had been 23% under the old flow, dropped to under 2%.

Conversion lifted 72% through the new checkout. Attributed revenue across the consumer digital portfolio was $200M+ over the two-year measurement window, traceable to the redesign through the controlled migration data. Upfront pricing transparency, the call I had argued for hardest internally, was adopted as a telecom industry standard within eighteen months. Within five years, every major US carrier was showing real prices in the configurator. The instinct that had been protected for fifteen years did not survive the data.

The thing I would do differently is start the conversation with finance and operations earlier. I built the design case first and brought finance in second. The work shipped, but the first six months of stakeholder management would have been smoother if I had started with the people who owned the revenue model and worked outward from there. The lesson, which I have applied at every role since: in mature companies, design's job is to convince finance that the design instinct is also the financial instinct. That conversation goes faster when you start it on day one.