Airbnb Page Performance Score (PPS): Multi‑Platform Metrics, Weighting, and Evolution

Airbnb places a strong emphasis on page performance as part of its commitment to craftsmanship and its mission of creating a sense of belonging for everyone. Fast page experiences are critical for business growth and user satisfaction.

To achieve a shared understanding of “fast,” Airbnb needed a system that could handle platform‑specific performance indicators for Web, iOS, and Android. Product engineers face challenges in prioritizing metrics, while managers need a way to compare and report progress across platforms.

Airbnb therefore built a new measurement system called the Page Performance Score (PPS) that aggregates multiple real‑user performance metrics into a single 0‑100 score for each page on each platform.

Early performance measurement

Initially, Airbnb used a single metric called Time To Airbnb Interactive (TTAI) to measure the time from page start to visible and interactive content. While TTAI helped establish a performance culture and address latency, it proved limited because different platforms have different baselines, and a single number could not capture the full user experience.

Introducing the Page Performance Score (PPS)

PPS breaks down performance into three concepts:

Page – a user journey on Airbnb.

Performance – multiple metrics collected for each page.

Score – a daily 0‑100 value derived from the metrics on each platform.

This aggregation allows easy comparison across pages and platforms and improves upon the single‑metric approach.

Metrics

Metrics are user‑centric and fall into two main categories:

Initial load metrics – measure time from page start to content becoming visible.

Post‑load metrics – measure responsiveness after the initial load.

Initial load metrics

Web: Time To First Contentful Paint (TTFCP) and Time To Largest Contentful Paint (TTLCP). Native: Time To First Layout (TTFL) and Time To Initial Load (TTIL).

Post‑load metrics

Web: Interaction to Next Paint (INP), Total Blocking Time (TBT), Cumulative Layout Shift (CLS). Native: Scroll Thread Hangs (STH), Additional Load Time (ALT), Rich Content Load Time (RCLT).

Formula

After measuring the metrics, Airbnb converts them into a single PPS using a formula derived from Lighthouse. Each metric is mapped to a curve based on good, medium, and poor thresholds, producing a score between 0 and 1. The weighted sum of these scores yields the final PPS.

PPS = curve(metric_1) * weight_1 + curve(metric_2) * weight_2 …

For the Web platform, the weighting example is:

PPS = curve(TTFCP) * 35% + curve(TTLCP) * 15% + curve(INP) * 30% + curve(TBT) * 15% + curve(CLS) * 5%

Evolution of PPS

Moving from a single metric to PPS required organizational change, including training teams to think beyond seconds. Old TTAI ranges were mapped to the new PPS ranges to ease the transition.

Because PPS is designed to be extensible, new user‑centric metrics (e.g., Cumulative Layout Shift introduced by Chrome in 2019) can be added without changing the 0‑100 scale.

Weighted Average Score (WAS)

Beyond individual pages, Airbnb tracks a Weighted Average Score across the organization by weighting each page’s PPS by its traffic. This provides a single number to prioritize high‑impact pages.

Conclusion

With PPS, engineers and data scientists have a suite of user‑centric performance indicators that enable clear comparison across pages, platforms, and teams. The system supports continuous iteration: metrics can be swapped, weights adjusted, and targets tightened while the 0‑100 scale remains stable.