Zero volume keywords are searches for which the tools (Google Keyword Planner, Semrush, Ahrefs, Moz) show no monthly search figure: it doesn't mean nobody makes them, it means they sit below the threshold the tool can measure. In our test on 10 October 2026 on 20 questions Google shows in the UK in the "People also ask" box, the Keyword Planner returns no volume at all for 15 (75%) and estimates 10 a month for another 3.
Let's start with a slightly comic detail.
In the UK, the search "zero volume keywords" is itself very nearly a zero volume keyword: the Keyword Planner gives it 10 searches a month, the lowest figure it shows, and has no figure at all for the singular "zero volume keyword".
Search it on Google from a phone and page one is mostly agency and tool blogs (The HOTH, LowFruits, Hashmeta), a 2023 YouTube talk by Mark Williams-Cook and a Reddit thread asking whether a 0 on Semrush really means zero. People clearly ask the question. The tools struggle to put a number on it.
So nobody searches for it?
Some people do. You, for instance.
With these searches we see three mistakes, always the same ones:
- dropping them because the tool says zero;
- expecting real traffic from a single zero volume query;
- building a page for every variant (one per town, one per region) in the hope that the total adds up to volume.
Two projects we work on take the first one apart.
For Macropix, a Milan-based LED wall manufacturer, "monitor pubblicitario" (advertising display) gets 390 searches a month in Italy and went from position 88 in 2020 to position 2 in 2025. A small, technical query, made by people who have to buy: the full method is in the case study on SEO for manufacturing.
Sunscape develops solar farms and looks for land all over Italy. "Affitto terreno per fotovoltaico" (land lease for solar) gets 880 searches a month. The same search with the name of a region doesn't exist, as far as the Keyword Planner is concerned. Yet the region-by-region structure we started in March 2026 took the number of page-one keywords from 1 to 38, out of 143 tracked, in six months.
And what has that got to do with zero volume keywords?
It's the point of the whole article: volume measures how big a question is, not how much it's worth. Below you'll find:
- what "zero" really means in Google's data;
- the test on 20 questions, with the full table;
- how many searches they actually get, according to Ahrefs' US data;
- four checks to decide whether a keyword deserves a page;
- the town-page mistake, which Google calls doorways;
- where to find them and how to check them at 28, 60 and 90 days.
Let's begin.
What "zero volume" means in Google's data (three reasons, none of them means zero searches)
"Zero volume" is a limit of measurement, not a measurement. The tool doesn't have enough data to give you a number, and there are three reasons, all of them written in Google's own documentation.
First: the Keyword Planner rounds and averages over a year. The Google Ads help page says it in two lines: "Your search volume statistics are rounded" and the number of searches "is averaged over a 12-month period". A question searched three times in January and never again that year has no way of showing up.
And if the Google Ads account isn't spending, the Planner only shows wide ranges, starting from "0-10": we explain how to read them in the guide on how to use Google Keyword Planner.
Second: Search Console hides rare queries too. The help page on the Performance report explains that, to protect privacy, the report "omits some queries that are searched a very small number of times". How much do they weigh? Patrick Stox of Ahrefs analysed 22 billion clicks across 887,534 properties: in April 2025 anonymised queries accounted for 46.77% of clicks. It's a worldwide sample, not broken down by country, but the order of magnitude is enough: you can't see nearly half of the long tail even on your own site.
Third: searches nobody has made before arrive every day. Google wrote it in 2017 ("15 percent of searches we see every day are new"), repeated it in 2019 when it announced BERT, and John Mueller confirmed it in March 2025 at Search Central Live in New York, according to Search Engine Journal's report: "it's always around 15%". No database built on the past can contain what nobody has searched for yet.
So are the tools wrong?
No. They estimate recurring demand, which is their job. A long, rare query is outside their field of view by design, and that's why in every keyword research tool long tail keywords so often end up at zero.
The test: 20 questions Google shows and its own Keyword Planner doesn't count (full table)
To see how big the gap is in the UK we ran a simple check on 10 October 2026.
Method. Five searches from sectors we work in, in their UK wording: "led wall", "cnc turning", "heat pump", "seo consultant", "solar panels with battery". For each one we downloaded the Google UK mobile results page through the DataForSEO API and took the first level of the "People also ask" box: 20 questions. We ran all 20 through the Keyword Planner (United Kingdom, English).
The result:
- 15 questions out of 20 (75%) with no volume data at all;
- 3 at 10 searches a month, the lowest figure the Planner shows;
- 1 at 40 ("is my house suitable for a heat pump") and 1 at 70 ("what is cnc turning");
- 18 out of 20 (90%) with no data or stuck at 10.
| Starting search | Question shown by Google | Keyword Planner volume (UK) |
|---|---|---|
| led wall | How to build your own LED wall? | no data |
| led wall | What are the different types of LED panels? | no data |
| led wall | Do LED wall lights use a lot of electricity? | no data |
| led wall | How much would a LED wall cost? | no data |
| cnc turning | What is CNC turning? | 70 |
| cnc turning | What are the downsides of CNC turning? | no data |
| cnc turning | What is the difference between CNC milling and CNC turning? | no data |
| cnc turning | Is CNC turning the same as a lathe? | no data |
| heat pump | What is the downside to a heat pump? | 10 |
| heat pump | What does Martin Lewis say about heat pumps? | no data |
| heat pump | Is it worth getting a heat pump in the UK? | no data |
| heat pump | Is my house suitable for a heat pump? | 40 |
| seo consultant | What is a SEO contractor? | no data |
| seo consultant | Is SEO still worth it in 2026? | no data |
| seo consultant | Can ChatGPT do SEO? | 10 |
| seo consultant | Is SEO paid or free? | no data |
| solar panels with battery | Is it worth getting solar panels with a battery? | no data |
| solar panels with battery | Will a 400W solar panel run a fridge? | no data |
| solar panels with battery | Is a battery a good idea for solar panels? | no data |
| solar panels with battery | Is it worth buying a battery for solar panels? | 10 |
Google puts these questions on the page because it considers them related to the search. Its own tool for advertisers has no number to give for three quarters of them. When we ran the same test on Google Italy, the share without data was 57% (16 questions out of 28).
The more technical the product, the bigger the gap: on LED walls all 4 questions have no data, on CNC turning 3 out of 4. We'd already seen the same thing on Google Italy when we measured eight industrial machining processes, where the subcontracting phrasing totals 180 searches a month: the data are in the article on keywords for manufacturing companies.
(Two limits, to be fair: it's a one-day snapshot on a small sample, and the Planner groups very similar variants together, so a "no data" question may have searches counted under another wording.)
How many searches they really get (roughly what's estimated, says Ahrefs)
A zero volume keyword usually gets few real searches: roughly the number the tool can't show you. The cleanest check we know of comes from the US.
Joshua Hardwick, then head of content at Ahrefs, took the keywords for which the Ahrefs blog ranked in the top five with an estimated volume of 10 or less in the US, and compared them with 28 days of impressions in Search Console. Average: 11.3 impressions per keyword. Less than 1% went over 100.
Even the best case stays small: "hreflang tag seo", estimated at 10 searches, showed 20 impressions in Search Console. Double, which is to say twenty.
(A small US sample, as the author himself points out. We haven't found an equivalent test published for the UK.)
So where do the stories of "zero volume" pages with thousands of visits come from? From the tail. In the same article, the Ahrefs guide on "submit domain to search engines" has an estimated volume of 10, but gets about 3,800 estimated organic visits a month because it ranks for dozens of variants.
On its own, a zero volume keyword brings few visits. The traffic comes from the page that answers all its variants.
Then why should I write about it?
For two reasons that have nothing to do with volume: someone asking such a specific question is often close to a decision, and competition is low because hardly anyone invests in it. It's the same reasoning as easy-to-rank keywords, with one more filter: the value of the person searching.
When a zero volume keyword is worth a page (four checks in ten minutes)
A zero volume keyword deserves a page when it passes four checks: intent, independence, value and SERP. If it fails one, at most it deserves a paragraph in a page that already exists.
1. The intent matches what you sell
The table has two questions with no data that look like twins. "How to build your own LED wall?" comes from someone planning a DIY project. "How much would a LED wall cost?" comes from someone putting a budget together. For an LED wall manufacturer the second is worth a page; the first is worth a paragraph at most. We explain how to recognise intent in the guide to search intent.
2. It's a topic in its own right, not a variant
Search the keyword and compare the results page with the one for the main search. If the top ten results are almost the same, Google treats it as a variant: a section in the page you already have is enough. If they're different, it's a topic in its own right. The method for measuring the overlap is in the guide to keyword clustering.
3. One enquiry pays for the page
You do the maths with the cost of a lead, not with the volume. For Valore Solare, which installs solar systems in Italy, a lead from Google Ads costs €70.78 on average. A page that brings in two enquiries a month from "zero" searches is worth, at click prices, almost €1,700 a year.
4. The SERP is weak
Forums, off-topic pages, guides from six years ago: if that's what page one looks like, there's room. If there are already three good, recent pages, zero volume won't hand you anything.
Here are the four checks applied to a real case. The searches made by people offering land to Sunscape in Italy, with Keyword Planner volumes on 10 October 2026:
| Search (English meaning) | Monthly volume (Italy) | Average CPC |
|---|---|---|
| affitto terreno per fotovoltaico (land lease for solar) | 880 | £2.09 |
| vendo terreno per fotovoltaico (selling land for solar) | 320 | £3.09 |
| affitto terreno agricolo per fotovoltaico (farmland lease for solar) | 140 | £1.78 |
| agrivoltaico puglia (agrivoltaics Puglia) | 40 | £1.11 |
| affitto terreno fotovoltaico piemonte (solar land lease Piedmont) | no data | - |
| affitto terreno fotovoltaico puglia (solar land lease Puglia) | no data | - |
| affitto terreno fotovoltaico sicilia (solar land lease Sicily) | no data | - |
| affitto terreno fotovoltaico lombardia (solar land lease Lombardy) | no data | - |
| agrivoltaico veneto (agrivoltaics Veneto) | no data | - |
| agrivoltaico piemonte (agrivoltaics Piedmont) | no data | - |
Six searches out of ten are at zero. But someone typing "affitto terreno fotovoltaico puglia" (solar land lease Puglia) is a landowner in Puglia: the exact customer, in the exact place. Right intent, high value, a SERP with little competition.
On the project, between March and August 2026, organic impressions went from 329 to 5,809 a month and clicks from 46 to 195. How those six months played out is in the article on how long SEO takes to show results.
The expensive mistake: a page for every town (Google calls them doorways)
A page per region is a good idea as long as each page says something that's true only for that region. When it becomes the same text with the name swapped, it's spam.
That isn't an opinion. Among its examples of doorways, Google's spam policies list exactly this: "Having multiple domain names or pages targeted at specific regions or cities that funnel users to one page". And the section on scaled content abuse targets cases where "many pages are generated for the primary purpose of manipulating search rankings and not helping users".
With AI, a hundred "service + town" pages take an afternoon to write. It's the fastest way to turn a zero volume opportunity into a penalty.
The test we use is blunt: replace the name of the region with another one. If the page is still true word for word, it's a doorway. A regional page holds up when it contains things that would be false anywhere else: local constraints, examples from that area, figures for that territory.
About to write dozens of pages on searches the tools put at zero?
You've just seen the line between a regional page that brings in customers and a doorway Google treats as spam. Before you produce them, let's look together at the real queries in your Search Console and the results page for each one: we'll tell you which deserve a page, which a section and which neither.
Book a strategy consultationWhere to find them (five sources, none of them a keyword tool)
You won't find zero volume keywords in volume tools, by definition. You find them where people write down their questions.
1. Search Console. In the Performance report, filter queries with a regular expression: ^(how|what|which|why|where|when|can|is|does)\b for questions, ^(\S+\s){5,} for queries of six words or more. Remember that the table shows at most 1,000 rows. The rest of the procedure is in the guide to Google Search Console.
2. "People also ask" and autocomplete. Our test says three quarters of those questions are at zero: they're the nearest and most overlooked source.
3. Your site's internal search. Terms that return zero results are questions your site doesn't answer yet.
4. The sales inbox. Emails, quote requests, support tickets: the questions that arrive at the end of the journey, written in the customer's words. We cover this among our creative SEO strategies.
5. The questions people ask AI. When ChatGPT searches the web, it rewrites the question into long queries. In the Nectiv study published by Search Engine Land on over 8,500 prompts, the average search is 5.48 words long and 77% have five or more. According to Peec AI, across 20 million searches generated between October 2025 and January 2026, the average length went from about 6 to about 12 words, and the United Kingdom is one of the five countries in its sample. For a Keyword Planner, almost all of these searches are zero volume. The mechanism is explained in the article on query fan-out.
How to tell whether it worked (the 28, 60 and 90-day check)
You judge a zero volume keyword on the impressions the page collects after publication, not on the volume the tool gave it beforehand. The cycle is the one SEOmator suggests in its guide, with the thresholds we use:
- Before: note the query, the date, the results page and the page that has to answer it.
- One change at a time: a new page or an extra section, not both together, or you won't know what worked.
- At 28 days: does the page appear in Search Console for the query or close variants? If there are no impressions, check indexing.
- At 60 days: look at the average position across the variants, not for the single query.
- At 90 days: clicks and enquiries. If they're there, widen the group; if there are impressions but no clicks, rework the title and the answer; if there's nothing, stop.
The Sunscape figures you read above were measured six months after launch, not three. Ninety days is enough to decide whether to carry on, not to judge the result.
Frequently asked questions about zero volume keywords
It depends: on its own, a zero volume keyword brings few visits. In the Ahrefs test on keywords estimated at 10 searches or fewer in the US, the average was 11.3 impressions in 28 days. Traffic arrives when the page also answers all the variants of the same question.
It depends on the results page. If the top ten results are almost the same as for the main search, Google treats it as a variant and a section is enough. If they're different, it's a topic in its own right and deserves a page.
No. It rounds volumes and averages them over 12 months, so rare questions don't show up. In our test on 10 October 2026, 15 out of 20 questions shown by Google in the UK had no volume data at all.
Not if the pages only change the name of the place: Google's spam policies list pages targeted at specific regions or cities among their examples of doorways. A regional page only holds up if it contains information that applies to that area and not to the others.
Yes. When ChatGPT searches the web it rewrites the question into long queries: 5.48 words on average in the Nectiv study of over 8,500 prompts. Searches that specific almost always come out as zero volume, but they're the ones used to choose which sources to cite.
Volume is a measure of the past (and searches are getting longer)
In short:
Zero volume means "below the threshold", not "nobody".
Of 20 questions Google shows in the UK, 15 have no data in the Keyword Planner.
On its own, a keyword like this brings few visits: what counts is the page that answers all the variants.
It deserves a page if it passes four checks: intent, independence, value, SERP.
A page per town with the name swapped is a doorway.
There's something the industry doesn't say much. Search volume is an average of the past twelve months, calculated on queries someone has already typed often enough. Meanwhile searches are getting longer: the ones AI tools generate on users' behalf went from about 6 to about 12 words in four months. Every month, the share of demand a volume tool can see gets smaller. Keyword research that starts from volume and throws out the zeros is looking at a shrinking slice. Research that starts from customers' questions, and uses volume only to rank them, isn't. That's how we approach SEO for the businesses we work with, and it's why the first question we ask isn't "how many searches does it get?" but "who's searching?".
Until next time!
