Search Console regex: the filters that reveal the queries you are not tracking

Where the regular expression filter lives in the Search Console performance report

Before writing a single pattern you have to find the box. Google tucked Search Console regex inside the ordinary filter menu without any fanfare, which is why plenty of people have used the tool for years without knowing it exists. It is four clicks, always the same four.

One warning that saves a lot of frustration: the filter runs over the queries Search Console chooses to show you, not over the true total of searches that sent you traffic. Google anonymises the very low-volume tail, so there will always be a gap between the sum of the table rows and the total on the chart. That is not a bug in your pattern, it is how the report works. The time window has its own ceiling too: the report keeps up to 16 months, a limit with real consequences for seasonality analysis that we cover separately in the guide on Search Console's 16 months of historical data.

The other thing worth deciding up front is what you will do with whatever you find. A regex filter with no destination ends up as a screenshot in a document nobody opens. The natural destination is a short list of keywords tracked daily, which is exactly what a rank tracker with Search Console built in is for: Search Console tells you where you appeared according to Google, the tracker tells you what position you hold today, in your city and on your device.

Minimum viable regex syntax in Search Console: RE2 and what it refuses

Google Search Console regular expressions are not the ones in your text editor. Search Console runs on RE2, the engine Google wrote so that no pattern, however twisted, could hang a server. That linear-time guarantee has a cost: RE2 removes every construct that would force the engine to backtrack.

Translated into practice: 95% of what you need behaves identically, and the 5% that fails is always the same 5%. This table is the whole thing you need to memorise.

Three traps if any of your traffic is in Spanish

The first and most expensive one is accents. Search Console stores the query exactly as the user typed it, and across Latin America most people type from a phone without accents. If your question pattern only lists the accented spelling of "cómo", you lose everyone who typed "como" — which in our data is the overwhelming majority. Every Spanish alternative goes in twice, no exceptions.

The second is that class shorthands such as the digit and word-boundary escapes operate on ASCII. A word boundary wrapped around an accented word behaves oddly, because the accented character is not an ASCII word character. The safe, boring alternative: instead of word boundaries, use explicit spaces and the end-of-string anchor. It always works and it reads better.

The third is the dot. When filtering URLs you will want to escape the dot in the domain, and backslashes survive copy-pasting badly between chat apps, docs and the browser. Trick: put the dot inside square brackets. Shorter to read, identical behaviour, and no backslash to lose along the way.

That is everything you need. What follows are concrete patterns, each with what it returns and the business question it answers.

Ready-made recipes: splitting brand from non-brand queries

This is the first segmentation anyone should run, for a simple reason: brand queries inflate your metrics and hide reality. If 60% of your clicks come from people who already knew you and typed your name into Google, your site-wide CTR is a comfortable fiction and your average position reflects no competitive battle at all. Splitting brand from non-brand is what turns the report into something actionable.

The pattern has to cover the correct spelling, the frequent typos, the hand-typed domain and the spaced-out variants. People spell brand names badly, and every typo you fail to list leaks into your non-brand bucket.

Recipe 1 · Everything that is brand

Filter: Query → matches regex. Swap the alternatives for your brand and its real typos, which you find by sorting the query table by impressions. The domain dot is deliberately left out so the pattern catches both the dotted and the spaced spelling.

Recipe 2 · Everything that is NOT brand

Same pattern, but choosing doesn't match regex. This is your real discovery universe: people who did not know you and saw you anyway. On a young domain it is usually 100% of the report, and that is where all the unfinished work lives.

Recipe 3 · Competitor brand names

Showing up in searches that name a competitor is a strong signal: Google already considers you part of that result set. These queries are usually low frequency and very high commercial intent, and they are almost always underserved because nobody writes content for them.

Recipe 4 · Brand plus a specific doubt

The dot-star in the middle acts as a wildcard: anything in between. The pattern returns queries where someone searched your brand next to a specific doubt. It is the best raw material there is for an FAQ page or a reviews section, because it literally tells you what is blocking the purchase.

Watch the order: this pattern requires the brand to come before the doubt. To catch both orders, duplicate the pattern with the two arrangements separated by a pipe — without lookahead there is no compact way to express "both terms in any order".

Start with what cannot be done: Search Console has no CTR filter and no position filter in the interface. Clicks, impressions, CTR and position are metrics, and filters operate on text dimensions. Any tutorial promising you a regex for "queries with CTR below 1%" is confusing the two.

The real way to do it is four steps, and regex is the first one:

What this looks like on a brand-new domain: our own numbers

Rankiamos.com logged 581 impressions and 1 click over 90 days. This is not a textbook example with comfortable figures: it is a new domain with no authority, starting from zero. That is precisely why it is useful, because it shows what you can extract when the numbers are small. These are the commercial queries with the most impressions after running the four steps:

Search Console data for rankiamos.com, 90-day window closing in July 2026. The whole site recorded 1 click in that period, so every query above is effectively at zero clicks.

The useful read is not "we are doing badly", which we already knew. It is this: the commercial query with the most impressions sits at position 23.8, and another sits at 18.4. Neither is an impossible jump from page one. Meanwhile "seo latam" at position 69 is a head term you do not win with a tweak — you win it with months of content and links. That turns a six-row table into a concrete prioritisation decision.

One aside before moving on: if these filters show several different URLs competing for the same query, do not treat it as a CTR problem. It is a different issue with a different fix, covered in the guide on spotting keyword cannibalization with Search Console.