Zero Search Volume Keywords: When the Zero Hides a Buyer

A reported zero next to a keyword is a statement about a database, not about the world. It usually means the phrase sits below the threshold at which a tool is willing to publish a figure. The practical question is never whether the zero is accurate. It is whether a buyer is hiding behind it.

This is for anyone who has closed a keyword export because the volume column read nothing. Below: where the zero actually comes from, a five-question filter for deciding which ones deserve a page, and the reason your own analytics has exactly the same blind spot.

Zero search volume, defined

A zero search volume keyword is a query that a research tool such as Ahrefs or Semrush reports with no measurable monthly searches. The figure describes the tool's reporting floor rather than the absence of demand. Some of those phrases have a real buyer behind them. Most have nobody at all. Separating the two is the entire skill.

Why this label does more damage now than it used to

Ahrefs' guide to the long tail publishes figures from its own US index: phrases drawing under ten searches a month account for almost 93 per cent of the database, roughly 2.3 billion entries. The same guide reports, second-hand from Google and from an era before assistants were mainstream, that around 15 per cent of daily searches had never been seen before, and that over 95 per cent of conversational long-tail phrases carry no measurable volume.

Those last two matter more each year. People now put whole situations into a prompt box rather than three words into a search box. The resulting phrasings are precisely the ones that land in the unreported bucket. We cannot tell you how much of that demand returns to a website as a visit. We can say that a library of specific, honest answers is a reasonable asset either way.

Where the zero comes from

A sampling floor, not a census

Every volume figure in every tool is an estimate built from sampled clickstream and ad data. Below a certain frequency, the estimate stops being publishable and gets printed as nothing. Tim Soulo of Ahrefs defines the parent concept carefully, calling long-tail keywords "search queries that get a small number of searches". Note what the definition does not say: nothing about length. Low frequency is the property; extra words are a side effect.

The cut practitioners add on top

Then there is a second filter, applied by the analyst rather than the tool. Despina Gavoyannis, writing in Ahrefs' guide to content gap analysis, suggests trimming a large export down with a floor of roughly twenty searches and a difficulty ceiling near thirty. That is sound triage for a list of forty thousand rows. It is not evidence that row forty thousand and one is worthless, and it gets remembered as if it were.

One narrow phrase against one broad one

Say you sell content software to small B2B teams. The phrase "how to approve blog posts with a two-person team" reports nothing. The person typing it has a defined role, a defined irritation and, plausibly, a card. Set that against "content marketing", which reports a large number and delivers students, job applicants and competitors doing research.

The narrow phrase might bring five visits a month and two genuine prospects. Whether that is the better trade depends on what a prospect is worth to you and what the page costs to produce. For a small team the arithmetic usually favours the narrow phrase, because a new section on an existing page is nearly free and the fit is exact. That is reasoning from unit economics, not a measured outcome, and we present it as such.

Reported zero against ordinary long tail

	Ordinary long-tail term	Reported zero
What the tool prints	A small but published figure	Nothing, or no data
Evidence you are working from	The vendor's estimate	Your own call recordings, tickets, site search and support inbox
Where it belongs first	Its own page	A section inside a page that already ranks
Downside if wrong	Modest	Larger, because a share of these earn nothing at all

None of the figures above demonstrate that an unreported phrase will bring traffic. What they establish is narrower and still useful: the instruments go blind at the bottom of the range, so the bottom of the range has to be judged on other evidence.

Five questions before you commit a page

1. Is there a person in the phrase? "Content workflow for a four-person B2B team" contains a role and a job. "Content tips" contains neither.
2. Can you answer it from something you have done? A real process, a real screenshot or a real number beats a page that paraphrases other pages.
3. What does covering it cost? If a single new subsection on a page you already own answers it, that is the version to build. Our guide to writing a tight brief keeps these small enough to be worth it.
4. Does it sit next to what you sell? A reported zero beside your core use case is worth more than a four-figure term three steps away from it.
5. Does any other signal confirm demand? A question raised twice in sales calls last quarter outranks any estimate a tool can produce.

Four or five passes: write the section. Two or fewer: drop it and move on. Then log the date, because the wait is long; our piece on how long search results actually take sets the expectation. We have run no controlled test of our own on unreported phrases, so we publish no hit rate. Practitioner write-ups describe low-volume pages picking up surprising impressions, and those are anecdotes. Expect a share of your bets to return nothing, which is the argument for keeping each one cheap.

Your Search Console has the same blind spot as the tool

Here is where we part company with the standard advice, including the advice we have given before. "The tool is blind, so check your own data instead" is only half true, and the half that is wrong matters. Google's own reporting applies a near-identical threshold, for a different reason, and it hides a comparable share of the picture.

Google documents this directly. In a Search Central post from October 2022, Daniel Waisberg explains that anonymised queries are those not issued by more than a few dozen users over a two-to-three month period, that they never appear as a row in the table or the API, and that applying any query filter drops them from the totals as well. Patrick Stox measured the scale of it for Ahrefs in February 2026, across 22 billion clicks and 887,534 properties, reporting that anonymised queries accounted for 46.77 per cent of traffic in April 2025, against 46.08 per cent when the study was first run in 2022. He expects the share to keep climbing as queries get longer.

So the fallback has the same floor as the thing it was meant to correct. A phrase too small for Ahrefs to price is usually too small for Google to print, and you cannot confirm it by searching your own query report for it. The filter will quietly exclude the very rows you are looking for.

What to do instead is specific. Read the gap rather than the rows: compare the chart total for clicks against the sum of the rows beneath it, as Google's illustration spells out, and that difference is your unreported demand expressed as a quantity. It will not name a single phrase, but a large gap tells you the long tail is paying you already, which is the strongest argument for writing more of it. Then get your phrasing from sources with no reporting floor at all: call transcripts, support tickets, your own site search, the questions in your onboarding sessions. Those capture a query that one person asked once.

We would retract this if Google loosened the privacy threshold or shipped a bucketed row for small queries, which would make the query report a usable instrument at the bottom of the range overnight. There is no sign of it, and Stox's trend points the other way. The strategic judgement also stays with a human, which is roughly the argument in our piece on what a strategist still does.

A reported zero is a question, not a verdict

Do not discard a topic because of its volume column, and do not commit to one because of it either. A volume figure describes a crowd. Your inbox and your call recordings describe a person. Write for the person, keep each bet small enough that being wrong costs an afternoon, and use the unreported share of your own traffic as the evidence that the small bets add up.

Questions about reported zeros