Searching in French: elision, accents and hyphens

If you sell in French, there are three things about the language that break search engines and that don't exist in English. This isn't a French version of the synonyms problem: it's a different problem.

That's what this post is about. Why a store that works fine in English or Spanish starts failing the moment you open it to France, Belgium or Switzerland, and what to look at to fix it.

The apostrophe, the accents and the hyphen. In that order of importance.

1. The apostrophe: the word your search engine splits in two

In French, the article attaches to the following word when that word starts with a vowel. L'huile. D'olive. L'écran. L'imprimante.

For a search engine that chops text at punctuation, that apostrophe is a separator. And that's where the damage starts, in two directions:

• Your catalogue breaks. If your product is called Huile d'olive extra vierge, the index can end up storing d and olive separately. That stray d is useless, and worse: if your search discards words shorter than three letters, it throws it away and ends up with less information than you had.
• The search breaks. The customer types l'huile d'olive and the engine searches four fragments, two of which are noise.

And there's a third level, the one that really separates a good search engine: the same customer might type l'huile, lhuile, l huile or simply huile. All four have to land in the same place.

The check: take your best seller whose name contains an apostrophe and search for it those four ways. If any of them fails, you now know where your customers are leaking out.

2. Accents: some people type them, some don't

French carries a heavier accent load than Spanish and, above all, more variety: é, è, ê, ë, plus the cedilla ç and the diaereses.

The problem isn't typing them. The problem is that on a phone keyboard they cost a long press, and plenty of people simply don't bother. Creme for crème. Cafe for café. Francais for français.

A search engine that treats è and e as different letters is splitting its audience in two: those who type accents find things, those who don't, don't. And since the latter tend to be the ones in a hurry on a phone, you're losing precisely the people with the least patience.

The normalisation has to work both ways: someone typing creme must find crème, and someone typing crème must find a product you happened to create without the accent.

3. The hyphen: one word, two or three

French is full of hyphenated compound words, and they happen to be product names:

• porte-clés — keyring
• sèche-cheveux — hairdryer
• micro-ondes — microwave
• tire-bouchon — corkscrew
• presse-agrumes — citrus juicer
• lave-vaisselle — dishwasher

Each one gets written three ways in real life: with the hyphen, with a space, and all joined up. sèche-cheveux, sèche cheveux, sechecheveux. And your catalogue has exactly one of them.

If your search engine doesn't join and split on its own, you end up needing synonyms for every variant of every product, and that's work that never ends. The right arrangement is for the engine to handle it and for you to keep synonyms for what's genuinely yours.

And two more, less serious but real

Gender and number. noir, noire, noirs, noires. Your product page has one form and the customer types another. It's the French equivalent of singular and plural, with twice as many cases.

English living alongside French. In the same store and sometimes in the same customer: smartphone and téléphone portable, e-mail and courriel, ordinateur portable and laptop. Here synonyms are exactly right, and the choice depends on who you sell to: in Quebec the French word wins by a distance; in France, often it doesn't.

What to check in your store, in ten minutes

1. Search a product with an apostrophe all four ways: with the apostrophe, without it, with a space, and without the article.
2. Search your best seller without accents.
3. Search a compound word with the hyphen, with a space, and joined up.
4. Search a colour in masculine and feminine.
5. Open the list of searches with no results and see how many contain an apostrophe or a hyphen. The proportion usually surprises people.

If 1 or 3 fail, it isn't a matter of adding synonyms: it's that the engine isn't treating French as French. Synonyms there only plug holes one at a time.

Why this matters to us

Frankly: because France is the market where we have the most at stake, and because the published material on internal search in French is scarce and almost all of it translated from English, where none of this exists. English has no elision, no cedilla, and no sèche-cheveux.

In Findalo the apostrophe, the accents and the hyphens are handled by the engine, with nothing for you to declare. That leaves synonyms free for what's yours: how your customers name your products, which is where the real gains are.

If you'd like to check it with your French catalogue, there's a free 1-month trial, or write to us and we'll look at it with you.