|Allintitle, Allinanchor, AllinURL broken? |
Inconsistencies when using quotes and searching for 2 or more KW phrases
| 5:08 pm on Oct 2, 2007 (gmt 0)|
I was doing some KW research today in Google UK and I have noticed something rather weird in Google. Basically, the allintitle, allinanchor and allinurl commands are returning strange/very inconsistent results when searching for 2 KW search phrases.
Example, I was looking for <widget rental>. Running allintitle without quotes would return 689 results, allinanchor would return 450, allinurl would return 780. Something rather weird as we are talking about fairly common terms.
However, running exactly the same search WITH quotes would would return 25,000 results for allintitle, 16,800 for allinanchor and 8,900 for allinurl.
Looking at the bolded KWs you can tell that the searches WITHOUT quotes are broader because there are more terms highlighted. For instance, [bold]Widget Rental [/bold] at [bold]Widgets [/bold] Direct UK (no quotes)
[bold]Widget Rental [/bold] at Widgets Direct UK (with quotes)
If I understand how these commands work (I hope I do after all these years working in the industry) not using quotes should mean a broader query and higher number of results. Something supported by the fact that there are more terms highlighted. However, that does not seem to be the case.
The funniest thing is that I have then tried searches for 25 sets of keywords. Funnily enough sometimes using quoutes will return LESS results (that makes sense) but more often than not it will return MORE results.
Have I missed something lately and there's been a change on the way this works? Is Google trying to drive SEOs crazy?
| 10:45 pm on Oct 2, 2007 (gmt 0)|
Yes, I see the same thing - although the actual results are all relevant. So I'd say these special operators are buggy at present, but not truly broken.
There seems to be much more churn in Google's total data-set than usual right now. Add in the fact that these numbers are not gathered in a straightforward way, but rather by a complex process across the various "shards" of data. I've seen similar periods in the past where numbers also made very little sense.
| 9:07 am on Oct 3, 2007 (gmt 0)|
Thanks for confirming Tedster. We'll have to keep an eye on it and see when (if) it gets back to normal.
| 4:08 pm on Oct 3, 2007 (gmt 0)|
I see no discrepancies.
You mention one using quotes and the other not, can you please provide examples of how you are formatting your search?
Here's an excerpt of a post I made a few weeks back on an allintitle topic
3 search queries
This is how I read the 3 formats.
inanchor:blue widget - Searches for "blue" with the serp results only containing results that have the word "widget" in their anchor text.
inanchor:"blue widget" - Searches for pages that have the exact anchor text of "blue widget". "blue widget companies" would not qualify as a matching anchor text.
inanchor:blue inanchor:widget - Searches for pages that have anchor text that include both words "blue" & "widget". "blue widget company" & "blue modifier widget" would all qualify to be returned as a result.
| 5:37 pm on Oct 4, 2007 (gmt 0)|
Hi Jeremy, we are not talking about relevancy but total number of results. A search using inverted commas is obviously more restrictive and should ALWAYS return less results that a query without them. This was not the case when I posted the message.
I have just run a test with 15 terms and it seems to make sense again...
By the way, if you want to get the number of sites that have both blue and widgets in the anchor text of links pointing to them I don't think that you have to do:
|inanchor:blue inanchor:widget |
You should do:
inanchor is meant to be used in conjuction with other terms whereas allinanchor is meant to be used standing alone.
If you start your query with allinanchor:, Google restricts results to pages containing all query terms you specify in the anchor text on links to the page. For example, [ allinanchor: best museums sydney ] will return only pages in which the anchor text on links to the pages contain the words "best","museums" and "sydney".
Anchor text is the text on a page that is linked to another web page or a different place on the current page. When you click on anchor text, you will be taken to the page or place on the page to which it is linked. When using allinanchor: in your query, do not include any other search operators. The functionality of allinanchor: is also available through the Advanced Web Search page, under Occurrences.
If you include inanchor: in your query, Google will restrict the results to pages containing the query terms you specify in the anchor text or links to the page. For example, [ restaurants inanchor:gourmet ] will return pages in which the anchor text on links to the pages contain the word "gourmet" and the page contains the word "restaurants".
| 6:05 pm on Oct 4, 2007 (gmt 0)|
OK now I see what you mean.
I just check a few searches and this is what I found. For example, doing the allintitle for my full name shows less results without quotes which is counterintuitive. That's just what the label says. But if I click to the last page of the results, both queries actually return the exact same number of results.
Sounds like the results counter up top is kinda randomized.
| 8:21 pm on Oct 4, 2007 (gmt 0)|
Didn't I just read something in a different thread about double-quotes searches bringing in supplemental pages by default? Anyway if I do a simple search for dinosaurs and for "dinosaurs" I get a much larger number of total results for the second search, so it's not just allintitle-type operators that are affected.
| 3:23 pm on Oct 5, 2007 (gmt 0)|
Not in the UK. 1 KW searches WITH quotes return the same results as WITHOUT quotes (as you would expect as well). I have checked this in google.com and it seems to work fine as well. Which is that thread you are referring to?
Btw, I have ran another test today and I have seen the abnormal results back. :(
| 5:01 pm on Oct 5, 2007 (gmt 0)|
I'm still seeing a difference, but it's much smaller today -- only about 1% (6,220,000 vs. 6,280,000). It's possible that this particular search is common enough to be pre-cached, which could explain the difference.
As for the other thread, I don't recall and in fact only skimmed over the comment briefly. In might have been in one of those endless "SERP changes" threads.