Methodology

How recommendations work.

Every match percentage on MarathonHelp comes from a transparent, point-based system — not a black-box model. This page documents exactly which signals matter, how they're weighted, and how ties get broken.

The Marathon Match Quiz

The quiz assigns each race a raw score out of 12 points, then converts that to a match percentage. Five signals contribute:

SignalMax pointsWhy it matters
Goal4What you want from the race — PB, first marathon, bucket-list major, or trail experience. Highest signal because it filters the catalogue most aggressively.
Timeline3When you want to race. Within 3 months, 3–6 months, 6–9 months, or long-range planning. Races outside your window can still match but score lower.
Region2Continent preference. Same continent earns full credit; adjacent continents earn partial credit.
Course2Terrain preference — road, trail, or mixed. Hard exclusions don't apply: a 5/5 road race can still surface for a trail-curious runner if other signals align.
Experience1First marathon, building, or experienced. Smallest weight because pace and ambition matter more than résumé.

Total maximum: 12 points. A race scoring 9/12 displays as a 75% match.

Tie-breaking

When two races tie on raw score, the higher overall race quality wins — defined as pbPotential + beginnerFriendly. That biases ties toward races that are objectively stronger across either dimension, rather than picking the alphabetically-first one.

Dashboard recommendations

Your dashboard surfaces three races based on your saved-races history. The scoring is simpler than the quiz — it's pattern-matching, not preference-matching:

  • ·Same region (+3)Race continent matches the most-represented continent in your saved races.
  • ·Similar difficulty (+2)PB-potential within 1 point of your saved-races average.
  • ·Similar season (+1)Race month within 2 months of your saved-races average month.
  • ·Quality bias (+0.5)Small bias toward higher-rated PB courses to break ties.

With no saved races yet, the dashboard falls back to surfacing high-quality, beginner-friendly World Majors — the safest bet for any runner without history.

What recommendations don't do

  • ·No AI generation — every reason shown is deterministic and traceable to a rule.
  • ·No personalised pricing, no sponsored placements, no “featured” tier.
  • ·No cross-site tracking. Your saved races stay in your account.

How is each race scored in the first place?

Recommendations only work if the underlying race data is honest. We document the PB-potential, beginner-friendly, elevation, and qualification logic separately.

How we score races →