RaceGoLive, a live results app: what I've learned

Why I built live tracking for road races and what race day taught me

Sports events

You cross the finish line, pull out your phone and look for your time. Your family, who saw you go past kilometre 8, wants to know if you've finished yet. And thousands of people are doing the same thing at the same moment.

I time races for RaceChip Aragón. Every year I do the timing for the 10K, the Half Marathon and the Zaragoza Marathon, as well as races like the 10K Bomberos Zaragoza. Getting results out live is one of the toughest parts of the job, and that's why I built RaceGoLive.

The whole race, live, on your phone

RaceGoLive is a web app for following a race live. You just open a link, with nothing to install and no sign-up. The map shows the course with runners moving between checkpoints, and below it the overall, gender and age-group results update every few seconds.

Every athlete has a profile with split times, pace and position. You can find them by bib number, name or club, save favourites or create groups to follow several runners at once. Your family knows where you are at every moment, whether you're at kilometre 8 or at the finish.

RaceGoLive sits alongside the timing, not inside it. The timing company keeps using its own system, which produces the official results on site as usual, and RaceGoLive pulls those times through an API (currently from Copernico) to build a live page branded for each race. It includes screens for the race announcer and the finish line, views for streaming in OBS or vMix, and space for sponsors. It has already powered live coverage of the 10K Bomberos Zaragoza and the 10K Zaragoza. If you time races and want to offer it to your organisers, everything is on the partners page.

A runner deserves more than a list

I wanted to give runners something more polished. A results list does the job, but runners deserve more. Watching their race unfold on the map, comparing their pace with their friends' and seeing their position the moment they cross the line makes results part of the experience.

So I built live tracking with a clean, modern design, made for the phone, because that's where people follow the race. On the day of the 2026 10K, 90% of visitors were on a phone.

What I've learned

First my bib, then my family's

People look up their own bib first, then the bib of someone they're waiting for. That's why search, favourites and groups matter more than any stats panel. And since most people arrive phone in hand, the app has to load without asking them for anything.

Almost a million queries in a single day

On 25 October 2026 we timed the Zaragoza 10K and 5K, with more than 6,600 registered runners. That day the official race website reached 47,714 page views, and thousands of people followed the race on RaceGoLive from their phones.

At the peak, more than 16,000 requests arrived in five minutes and almost 88,000 in an hour. Every open phone polls for updates every ten seconds, and every search or filter adds another request. Behind the scenes, the database handled almost a million queries that day, because every request recalculated the standings across more than 6,600 results.

The live page slowed down and threw errors while the race was still on. The timing itself was never affected, because our system runs independently of the app and we delivered the official results on site as reliably as ever. For RaceGoLive, though, it was a trial by fire, and it taught me the most valuable lesson of the project. The number of visitors matters less than the cost of each request.

Compute once, serve many times

The decisive improvement was to compute each index once per data snapshot and share it across every request. We went from twenty-one queries to two. Add a longer cache, shared reads and a pause on per-visit statistics, and response times dropped sharply.

Measure (95th percentile)BeforeAfter
Results list119 s0.5 s
Live race status69 s0.6 s
Runner profile66 s0.4 s
Event page50 s0.7 s

Rehearse with a thousand spectators before the start

Until then I had only tested how times came in from the timing system. Four days later I ran a load test that simulates up to a thousand spectators at once browsing the 5,181 runners of the 10K, pausing two to five seconds between clicks. I also tried a scenario where every request missed the cache. Every response came back correct, with a median of 23 to 103 milliseconds, on a single local process with no CDN in front.

That simulator is now part of my race-week routine, and I'll run it before every race.

Don't let the crowd slow down the live feed

In RaceGoLive, the engine that collects times from the timing system and the website spectators use shared the same data layer. When spectators overloaded it, the live page stopped receiving new times for a while, even though the official timing carried on unaffected. Today the engine handles those peaks without going down.

Ready for the next big race

With the lessons from the 10K, RaceGoLive is ready for whatever comes next.

If you're a timing company and want to offer live coverage to your race organisers, I explain how in live results for timing companies. If you organise a race and want a website that holds up on race day, that is part of what I do in websites for races and sports events. Both pages are in Spanish.

Related services · Sports events