From spreadsheets to real-time pricing decisions

Confidential work

This case study describes design solutions for a B2B SaaS platform. Detailed UI mockups and user data are protected under NDA and available to discuss in person or via secure channels.

Capacity and Overbooking

The Problem

Airlines were losing money on every flight. Not because of bad pricing – because nobody knew how many seats they actually had to sell.

Three different capacity numbers lived in three different systems. Physical seats (what the plane has), restricted seats (regulatory limits), virtual seats (what you can sell). Revenue analysts spent 4+ hours every day trying to reconcile them in Excel.

Worse: the overbooking system had a 30% variance. Analysts would set "115% target" and the system would actually book 108% or 121%. Completely random. That gap cost airlines regulatory fines and missed revenue every quarter.

The Goal

Real-time capacity visibility. Predictable overbooking. Zero manual errors.

What I built

I designed an overbooking wizard that shows how adjustments affect sellable capacity in real time. You could see: "This flight currently sells 300 seats, but with 3.5% overbooking, we can book 311 – an increase of 11 seats."

For overbooking adjustments, I added a simple wizard:

1. Filter flights (JFK-LAX, next 30 days)
2. Select rows you want to change
3. Enter target percentage (115%)
4. Preview rounded capacity (+12 seats per flight)
5. Add mandatory comment ("Summer peak demand")
6. Confirm

The system logs everything to an Influence History audit trail.

Research process

I shadowed 5 revenue analysts for a week. Watched them work. Counted clicks. Timed tasks. Asked questions.

Key finding: They didn't trust the system. They would set an overbooking target, then manually check 10-15 flights in another tool to verify it actually applied. Because it usually didn't.

The problem wasn't the algorithm – it was that the UI never showed them what actually happened after they hit "Submit."

Results *

* All since Q1 2022, across 5 airlines.

Revenue analysts went from "I hope this worked" to "I know exactly what changed."

Network Origin-Destination

The Problem

Revenue teams were analyzing the wrong thing. They'd look at individual flights: Oslo-Rome direct, 78% full - looks good, and miss the bigger picture: that route was bleeding money across five connecting options.

Eery team worked at the leg level (single flights), not the market level (regions level, network). Data lived in three separate systems. Pulling an OD report? Four days, if you were lucky. By the time analysts got their spreadsheets, the market had already shifted.

Network planners couldn't answer basic questions like "Should we fly Barcelona-Frankfurt?" without spending a day reconciling booking data, competitor fares, and last year's numbers. 15-20% of that data was wrong.

Why OD vs Leg matters

Here's the thing airlines got wrong for years.

One's tactical. One's strategic. We needed to give teams the network view.

What I built

An OD analytics platform that finally showed the full picture. Pick a city pair-say, Barcelona-Frankfurt, and see:

The filter + aggregation pattern

After multiple design iterations, we killed the drill-down idea. Why? It trapped people in hierarchies. Supporting multiple cascades is not the right thing to do in tables. Instead:

Same pattern we used for Advanced Filters. Same mental model across products.

Making it work on iPads

Network planners don't sit at desks all day. They're in airport meetings, staring at tablets. So we made the table responsive: collapsible columns, horizontal scroll for details, tap-to-expand sparklines.

One-click export? CSV with your exact view-filters, sorts, aggregation level-baked in. Filename even includes the date and criteria. No more manual cleanup before sending to stakeholders.

Results *

* Measured across 5 airlines, 2021-2023.

Network planners stopped managing individual flights and started optimizing markets. That's how Cyprus Airways went from reactive scheduling to strategic capacity allocation.

In press

Promotion Management

Work in progress

This section of the case study is currently being developed and will be available soon.

Results *

* 18 promotional campaigns, 2021-2023.

The new workflow eliminated IT dependencies, reduced campaign setup from a full workday to under 2 hours, and gave marketing teams confidence through real-time validation. Airlines reported 12% higher promotional uptake due to faster time-to-market during demand windows.

In press

Advanced Filters

Work in progress

This section of the case study is currently being developed and will be available soon.

Results *

* All since Q1 2022, across 5 airlines.

The new filter system transformed analyst workflow from a frustrating bottleneck into a fluid, keyboard-first experience. Teams now create 3× more complex analyses in 80% less time, with saved filter sets becoming institutional knowledge shared across departments.

What I learned

Talk to users early

I wasted 3 weeks building a promotion preview feature that showed "estimated revenue impact." Users ignored it. Why? They didn't trust forecasts. They wanted to see affected flights-actual inventory they could verify. I redesigned it to show flight lists instead of projections.

Validation is a feature, not a bug

Analysts initially complained that the system "blocked" them too much. I explained: "Would you rather find out about conflicts now or after the promo goes live?" They got it. Validation went from "annoying" to "life-saving."

Performance matters

A beautiful dashboard that takes 15 seconds to load is useless. I spent as much time optimizing render speed as I did on visual design.

Documentation scales

I built filter patterns once, documented them, and saw them reused across 4 other product teams. Good design work compounds.

Appendix

Capacity and Overbooking

Objects and Actions