Creating high-quality Strava segments for cyclocross races can be challenging. Many of the factors that create difficulties for Strava’s segment-matching algorithms are present in spades in cross races. These include segments having start and end points close together, loop segments with rides that traverse the segment multiple times, and tightly knotted segments which pass close to the start and end points. These characteristics pretty much define the typical cross course, making cross segments inherently challenging. Adding to this difficulty is the fact that cross courses are usually changed for every day of racing, which means that a segment really only applies to a single race day.
There are several things you can do to help create better cross segments.
1. Start with good data. Location data from a dedicated GPS unit, like the Garmin Edge series, is usually better than data that comes from an iPhone or an Android phone. Many Garmin units offer the ability to record data points every second (as opposed to the three-second recording interval in phones, or the even longer intervals you can get in Garmin units with the “Smart Recording” feature turned on). Having closely-spaced data points helps a lot in cross races, which contain lots of small-scale twists and turns. Good GPS data from a cross race will show the course crisply and smoothly, with very repeatable laps. Bad data will be jagged and angular, with poor repeatability across laps. The better your source data is, the better it will match other riders tracks when they ride the same course.
2. Choose the start point carefully. For courses with an off-course staging area that leads into the main course, make sure the starting point is on the main course, not in the staging area, or immediately after the staging area. If your starting point is in the run-in to the main course, it will only pick up the first lap, and will miss subsequent laps.
3. Don’t overlap the start and end points. Putting the start and end points right on top of one another can confuse Strava’s segment-matching algorithm. Leave a bit of space between them — ideally, 50 meters or so. You won’t get measurements over this short gap, but you will get better data matching for your laps.
4. Keep the start and end points away from other sections of the course. This is particularly important for the start point. If other sections of the course loop around and pass close to the start point, Strava will often pick that later section as the start point, resulting in laps that only partially match the segment. Ideally, you should have the start point located just past the finish line, and the end point just before the finish line, but for many courses this isn’t possible if other parts of the course pass close by the finish line. In those situations, the best thing to do is to put the segment start and end points in an isolated section of the course, if you can find one. You may have to experiment a bit to get this to work well — create the segment, see how well it matches your ride (and other riders’ rides), and delete it and try again if it’s not matching correctly.
5. Name your segment well. For many races, just about any ride in the general vicinity of the course will match almost every segment that has ever been created for that race, resulting in a confusing list of segment matches. If your race is called “MonsterCross”, don’t call your segment “MonsterCross”, because people will have no idea what particular course the segment applies to. “MonsterCross 2012 – Day 1″ or “MonsterCross – Sep 30, 2012″ are much better names.
6. Hide old and lousy segments. When you upload your ride and find that it matches a long list of irrelevant segments, hide the ones that don’t matter (click the “Hide” button that appears when you hold your mouse cursor over the segment). This will remove the segment from your list of segment matches, and will also give Strava a bit of negative feedback about the segment. If enough people choose to manually hide a segment, eventually Strava will automatically hide the segment by default, making everyone’s lists of segment matches a little bit cleaner.