Sport & surf: seven quantities, eight disciplines, one catalog
A “5” on a sports listing can be a 5 m2 wing, a size 5 ball or an L5 racket grip. Three products, three aisles, one number.
The same number means different things from one aisle to the next
This is what sets sport apart from every other vertical. In fasteners, an M8 is always a diameter. Here, a surf and multisport retailer sells boards, wings, wetsuits, balls, rackets and shoes — six families that share no attribute at all. A number only takes its meaning from the product written next to it.
A general-purpose engine doesn't draw that distinction: it treats “5” as a word, and returns the wing, the ball and the racket on the same page. The sport rule pack holds 26 rules, written to resolve exactly this ambiguity (the detail of each attribute).
Volume, sail area and thickness kept apart
The sport rule pack activates when you pick your industry, at indexing time. It produces these annotations, each of which becomes a usable filter:
| Annotation | What it isolates |
|---|---|
VOL_75 | a board's volume in liters — the sizing selector |
SURF_5M2 | the sail area of a wing or a kite |
NEO_4_3 | a wetsuit's thickness, torso and legs |
TAILLE_T5 | a ball's standard size |
POIDS_300G | a racket's unstrung weight |
CORDAGE_16X19 | the string pattern, mains and crosses |
GRIP_L2 | the grip size |
To these are added the product family, the discipline (surf, wing, kite, foil, handball, basketball, running, tennis) and the material — carbon, epoxy, EPS or neoprene. Between two boards of identical volume, the material separates two prices by a factor of three. This pack pairs more than any other: six combinations come out of it, one per family — PLANCHE_32L, AILE_5M2, FOIL_95L, RAQUETTE_300G, BALLON_T3, COMBI_4_3. That follows directly from how heterogeneous the aisle is: where fasteners have a single buying pair, six families sharing no attribute call for six.
Five references, exactly as a supplier writes them
| Indexed reference | What the engine extracts |
|---|---|
| Wing Foil Board DUOTONE 75 L | FAM_PLANCHE · DISC_FOIL · VOL_75 · FOIL_75Lwing foil board, 75 liters |
| Aile de wing NAISH Freeride 5 m2 | FAM_AILE · SURF_5M2 · AILE_5M2wing, 5 m2 |
| Combinaison neoprene ONEILL 4/3 mm homme | FAM_COMBINAISON · MAT_NEOPRENE · NEO_4_3 · COMBI_4_34/3 wetsuit, 4 mm torso, 3 mm legs |
| Ballon de handball MOLTEN T3 | FAM_BALLON · DISC_HANDBALL · TAILLE_T3 · BALLON_T3handball, size 3 — adult |
| Raquette tennis BABOLAT 300g 16x19 L2 | FAM_RAQUETTE · DISC_TENNIS · POIDS_300G · CORDAGE_16X19 · GRIP_L2tennis racket, 300 g, 16×19 string pattern, L2 grip |
A retailer's catalog aggregates imports from different suppliers, keyed in by different people, over ten years. The same volume is written 75L, 75 L, 75 litres or volume 75 L in it: all four spellings produce the same annotation.
75 liters of wing are not 75 liters of surf
This is what the second pass exists for. Volume alone selects nothing: a 75 kg rider looks for 30 to 35 liters in a shortboard, and 90 to 110 liters in a beginner wing board. What designates a product is the volume + discipline pair, not the number.
So the engine first annotates each attribute on its own, then reads its own annotations back to form the buying pairs: PLANCHE_75L, AILE_5M2, RAQUETTE_300G, BALLON_T3. A search for “wing 5m2” then returns wings only — not wing boards, which have no sail area.
What the pack does not recognize
Two attributes of this industry are deliberately absent, and it is better to know it before indexing than to find out afterwards.
Feet-and-inch dimensions. A board described as 6'2" x 19 1/4" x 2 1/2" loses its dimensions as it is read: normalization only accepts the period, comma, slash and hyphen as internal separators, so the apostrophe and the double quote disappear before the first rule runs. The fraction survives, the foot and the inch do not. The board stays indexed and findable by its name, brand and material — but its dimensions do not become a filter. Volume in liters plays that role instead, and it is the selector riders actually use.
Shoe sizes. 42 2/3 produces the token 2/3, which is exactly a wetsuit thickness or a string ratio. The ambiguity is not only between brands: it sits inside a single catalog. Recognizing shoe sizes would cost false positives on two other families, and that price was not judged worth paying.
The taxonomy already built for you
The sport pack isolates volume, sail area and size as filterable attributes, not as text. You can then offer a “boards from 70 to 90 liters” or “4/3 wetsuits” page without building the taxonomy yourself.
Does a naming scheme specific to your catalog still escape the pack (an in-house range, a shaper's code)? Add your own Custom Rule from the console, describing what it should recognize — with no regex to write.
Indexing your catalog
The getting started guide covers the full setup, from signup to your first indexed catalog. The documentation details every endpoint.
Watch the breakdown happen. The table above is the engine's output, not hand-written. The same mechanism runs on the home page, with no signup — on a hardware or fashion catalog, not sports gear. What you see there is the annotation mechanism itself, the one the sport pack applies to your volumes and sizes.