Faragiri

Learn The long tail Keyword research

A bird food seller counted its year: the head query brought more searches than any other and the two hundred small ones brought eleven times the orders.

4Sections
4Min read
4Depth

Bardsey Bird Food sells wild bird seed and feeders by post and counted its search year to June 2026. One query at 2,400 a month, twelve at about 140, and 210 at about 22, which comes out at 4,620 searches in the tail against 2,400 in the head. The tail is larger than the head once you add it up, and that is before anything is said about who is asking.

The head query is bird food. The twelve in the middle are things like sunflower hearts and no mess bird seed, about 1,680 searches a month between them. The 210 in the tail are queries like what to feed goldfinches in winter and which feeder stops squirrels 32mm, and Bardsey had never deliberately written for any of them.

The arithmetic that decides it

Searches are not the reason to prefer the tail. Orders are. The head converts at about one order in every 250 visits and the tail converts at about one in 17, which is the whole argument in two figures, and it holds for a reason that has nothing to do with search engines.

The tail

Three bands of query, and the one at the bottom is the widest

The band with the fewest searches each has the most searches in total and by far the most orders, which is the whole of the argument.

  1. One head query, about 2,400 a monthbird food. Contested by national retailers and worth about one order in every 250 visits.
  2. Twelve mid queries, about 140 eachsunflower hearts, no mess bird seed. Around 1,680 a month between them.
  3. 210 tail queries, about 22 each4,620 a month, about one order in 17, and most of them in no keyword tool.

The tail is worth chasing for direct orders, not as a stepping stone that accumulates into ranking for the head query.

A longer query has already narrowed itself. Somebody typing bird food may be a person with a garden, a child doing homework, or somebody looking for a shop three counties away. Somebody typing which feeder stops squirrels 32mm has a squirrel, a measurement and a problem, and the only question left is whose feeder.

The tail is not a cheaper way to win the head, and that misreading costs the most. Writing 210 small pages does not accumulate into a ranking for bird food, because the head query is answered by a different kind of page and contested by national retailers. The tail is worth having on its own terms, as orders, and a firm that treats it as a staging post is disappointed twice.

Where the tail is actually found

Tail queries are found in your own impressions, not in a tool. Keyword tools drop queries below a threshold, so most of Bardsey's 210 appear in no tool at all; they appear in Search Console, which reports anything the site was shown for, including a query asked eleven times in a year.

Discovery

Four steps to find tail queries in Search Console

Keyword tools drop low-volume queries below their threshold, so Bardsey pulls them from the bottom of its own impression log.

  1. 01Open Search Console queriesCaptures low-volume phrases asked eleven times a year that paid tools omit.
  2. 02Sort impressions ascendingRead the bottom two hundred rows instead of starting from the top.
  3. 03Filter for positions 20 to 60Finds 32 queries the site is already nearly answering by accident.
  4. 04Write one page per shared intentNine pages covered the batch; one squirrel feeder page now draws 61 queries.

One thorough page with a table of gauge sizes can answer dozens of tail variations without listing every phrase.

The method is one filter. Sort the query report by impressions ascending rather than descending, read the bottom two hundred rows, and look for the ones where the shop is at position 20 to 60. Those are queries the site is nearly answering by accident, and a page written on purpose moves them.

Bardsey found 32 such rows and wrote nine pages, because one page can answer forty tail queries if they share an intent. Everything about squirrels went on one page with a table of gauge sizes, and that page now collects impressions for 61 queries the shop never listed anywhere.

Where this stalls: two hundred thin pages

The stall is the other reading of the same advice. If the tail is where the orders are, and there are 210 queries, the obvious plan is 210 pages, and a firm that follows it produces 210 pages of eighty words that answer nothing and are dropped as a set before they rank for anything.

The stall

Grouping 210 queries into nine durable pages

Publishing one thin eighty-word page per query gets dropped as a set; grouping by searcher intent turns 210 queries into nine strong pages.

  1. 210 raw tail queries
  2. Minor phrasing variants
  3. Temptation to make eighty-word pages
Group by what the searcher wantsSort by intent rather than exact wording
  1. Nine substantial pages
  2. Each page answers many variants
  3. Enough depth to earn and hold rank

The grouping is the real work; thin single-query pages fail quality thresholds before they can rank.

The grouping is the work. Nine pages was the right number for 210 queries at this shop, arrived at by sorting the queries by what the person wanted rather than by what they typed, which is what the searcher wants, which the words hide. Whether a page is substantial enough to be kept at all is what separates a page that ranks from one that exists.

What the numbers here are and are not

Google Search Console produced every figure above, free, and it replaced the keyword data Google removed from analytics in 2011. Its query report is capped, so the 210 is what the report showed rather than what exists, and the real tail is longer than the figure here by an amount nobody can measure.

Evidence limits

The exact numbers behind the conversion claim

Every figure came from Search Console and shop orders, which means the report is capped and the conversion sample is small.

  • 210Queries shown in capped reportSearch Console caps query rows, so the real unmeasured tail is longer.
  • 340Visits across the nine pagesReal store traffic used directly rather than any external benchmark.
  • 20Orders from those visitsOrders divided by visits gives the one in 17 conversion rate.

Search Console replaced analytics keyword data in 2011 and remains free, though capped rows undercount the full tail.

The conversion figures are the shop's own orders divided by its own visits, which is the honest version and a small sample on some rows: the one in 17 rests on 340 visits and 20 orders across the nine pages, not on any published study.

Where the 210 came from before any of this is sources, and building the first list, and the reason no tool showed most of them is two numbers, and how badly both mislead. Both belong to finding what people type, and what they mean by it, and the head query is still the one everybody in the trade talks about.

Alongside this

Above this