Regex Case Insensitive and Word Boundary in Search Console

In the last lessons we used ordinary English words in our regex: we wrote (price|cost|how much) and ^(how|can|why). But the problem starts when you look at real Search Console data. You might find "optimise images" and "optimize images", "wifi router" and "wi-fi router", "iphone 17 colours" and "iphone 17 colors". So the very thing you're trying to isolate, users can type in more than one way. Today we won't learn a different regex. It's the same regex we know, but we'll learn to use it with the way people actually type, without turning every pattern into a huge list of options.

Square brackets []

You have: optimise images, optimize images, compress images, image size. The difference between "optimise" (UK) and "optimize" (US) is one letter, s or z. Do I have to write both words in full every time? No. When we write [sz], it means: pick one character from the ones inside the brackets. So we can write optimi[sz]e instead of (optimise|optimize).

A character class can also hold a space and a hyphen. People type the same product as "wifi", "wi-fi" and "wi fi". The middle is a hyphen, a space, or nothing at all. So: [- ] picks a hyphen or a space, and the ? from lesson 4 makes it optional: wi[- ]?fi.

The same idea works for other spelling pairs. The ? on its own covers a letter that's sometimes there: colou?r matches "colour" and "color", and e-?mail matches "email" and "e-mail". But careful: we're not correcting spelling, and we're not saying "every s is a z" in general. We're using a character class or an optional letter in one particular place, inside one particular word, because we noticed a real variation.

() versus []

If I have whole words (optimize, compress, speed up), I use (optimi[sz]e|compress|speed up) style alternation. If I have a different character in the same spot (optimise, optimize), I use optimi[sz]e. And we can combine them: (optimi[sz]e|compress|speed up) is a group of words, and inside one of them we used a character class to pick the letter.

Now let's try to match the whole word

Say you want to segment the "pro" models: iphone 17 pro, macbook pro. You write pro. Look at what comes back, because the match is partial:

"product", "promo" and "improve" all contain the letters p-r-o. That's the classic problem with short words in English: app is inside "apple" and "happy", ai is inside "email" and "repair". We need to say: the word pro on its own, not pro inside another word.

This is where \b comes in: a word boundary. It doesn't match a character; it matches the edge between a word character and something that isn't one (a space, a hyphen, the start or end of the text). The RE2 engine Search Console uses defines \w, a word character, as A to Z, a to z, 0 to 9 and the underscore. So \bpro\b means: pro, with no letter or digit glued to either side.

Now "product", "promo" and "improve" are gone. But notice "pro-level camera phone" stayed: a hyphen isn't a word character, so it counts as a boundary. Is that what you want? Maybe, maybe not. That's exactly the kind of case to check against your real data, not to assume.

Another way to say "the whole word", using only what we already know: the word comes at the start of the query or after a space, and is followed by a space or the end of the query: (^|\s)pro(\s|$). It's stricter than \b: it drops "pro-level" too, because a hyphen isn't a space.

Finding queries with characters outside basic English

Some queries carry accents or characters from other languages: "café near me" next to "cafe near me", "crème brûlée recipe" next to "creme brulee recipe". Inside square brackets, a ^ at the very start flips the meaning: any character that's not in the set. And \x00-\x7F is the range of basic English characters (ASCII). So [^\x00-\x7F] means: one character that isn't basic English.

And if you want the queries that do contain English letters? [A-Za-z]: a character from A to Z or a to z.

Two words in order, with anything between them

Say you want the queries that mention an iphone and then a case: "iphone 17 pro case", "iphone silicone case". First a small new symbol: * means zero or more (compared with +, which means one or more). With . we can write .*, meaning "any number of characters, even zero".

So iphone.*case means: iphone, then anything, then case. The order matters: iphone has to come first.

"case for iphone 17" didn't show up, because there case comes before iphone. If you want both orders, combine them: (iphone.*case|case.*iphone). That's useful when you want to understand how people word things: do they lead with the product or with the accessory? Is there a CTR difference between the two phrasings?

Do I have to turn every variation into a clever regex?

No. If you try to compress every letter, every variation and every typo, you can end up with a pattern nobody understands. Sometimes (tech zone|techzone|tech-zone|teck zone) is clearer and safer than a very clever regex. The test isn't "who wrote the shortest regex?" but "is the pattern clear, and does it produce the right segment?"

The practical test

You have a brand called Smart Store, and users type it as: smart store, smartstore, smart-store, and the typo smrt store. Careful: the site also gets queries like "smartphone store", which isn't the brand.

A thinking test

If you have a brand that people type 8 different ways, what's the first thing you do? Not "write the cleverest regex possible", but "understand the real variations in the data". Then you decide: is the difference one character? Use []. Do I have different words or spellings? Use (). Is a small part optional? Maybe ?. Is a short word matching inside longer ones? Think about \b. What matters is that the segment is clear and correct.

Before the next lesson

So far we've worked mainly on queries. But Search Console lets us use regex on pages too. In the next lesson we move the same mindset to the structure of the site. We'll see how to isolate the blog, products, categories, services, different languages, PDF files, pagination, and URLs with parameters, years or IDs. And that's where regex starts helping us understand not only how people search, but also which parts of the site get the visibility and the traffic in Search Console.