Late April, we came across SEO PowerSuite’s blog about their new TF-IDF feature in Website Auditor. In that blog, Google mentions that it counts the number of times a keyword is on a page in relation to the total number of words, but Google looks to do more than that to understand meaning. We love it when we find statements like that from Google. In this case, we can see if Google is telling us the truth.
We test if, in fact, they look at how close terms are to each other.
Test Downloads
Transcription Below
This was my second favorite test from this group. Does the distance between keywords affect ranking?The idea here came from a Google blog post from 2014.
In that blog, it mentions it counts the number of keywords on a page in relation to the total number of keywords but the idea was, this was prerank brain, Google wants to do more than understand meaning. What do these words actually mean and not just seeing them on the page and counting them but also what the words mean together and what they were queuing in on was the distance between words.
The quote was “Term ratios are a start…” - how many times a keyword was on a page divided by the number of words on a page. “…but we can do better. Search indexing these days is much more involved, using for example the distance between pairs of words on a page to capture their relatedness.”
This I thought was a really cool statement. This puts into my mind that if you have two keywords and if you were to put them in a search, if they are next to each other in a page, would that show that they are closer and there is an actual stronger relationship between these two pages. What I ended up doing here on this set up was 5 pages were indexed in the normal way then an additional keyword was added to all the pages the same number of times.
In this one instance, the keywords were right next to each other. In the next, they were five words apart and in the last instance, on the three other pages, they were eight words apart. The idea here is that if you then search for both keywords together, will the page that has them together rise to the top and will the page that has the next closest together will be on the number two spot – and that is exactly what we got.
The keywords together got number one. We are searching for these keywords as one thing. The page that has them together moves to number one. The page that has 5 words was number two and the 8 words were below that.
What we have seen and what we are showing here is that Google is seeing the relatedness between two words simply by being next to each other or closer to each other. Then when you search for those particular terms, how close they were on the page will affect the ranking of those particular pages.
There are a lot of implications for this. One thing is when you’ve got two words and it sounds weird if it didn’t have a preposition in the middle – in or at e.g. plumbers in phoenix or something like that – you can’t say phoenix plumbers. I guess that is ok, not the best of examples – but you know those words that sound awkward when placed together. You can conceivably have them a bit farther apart and you would still have at least some relatedness between them.
The other thing I think about are variations on a keyword, matches of that keyword – you would probably want to get them close to each other. Another thing I think about is LSI terms.
In order to make a closer relationship say, between target phrase and particular LSI terms, you really want them to get close as you can. Just having them on the page might not be the right way but having them on the page strategically – really close to your target phrase or what this page is about is a real priority.
Is it also the same as stop words like “in” and “the”? I thought those got pooled, anyway.
I am not entirely honestly – like pooled, pooled – I think if you were to search “Phoenix Plumber” and “Plumbers in Phoenix”, I think the results may shift a little bit which would tell me that maybe there is a slightly different importance that Google is still actually reading.
That is actually something worth testing, especially within your niche. Test that out – so I’ve got my plumber and I am in Phoenix, will the same pages win both terms? Look at it that way and I think that will tell you, in your particular case, if Google is pooling or not pooling the stopwords. You can also look to see how Google handles the particular variation.
I think Ted actually said this the other day and it was a really good observation, what kind of variations you could do with your city or state – should you do Phoenix Arizona, just Phoenix, or Phoenix AZ, doesn’t matter, and how you could look for that is to do searches on all those variations and see if the same pages come up in the same words or if they change. Then, look at those changes and see what is best for your particular situation.
You can actually see how Google are treating those terms – if they are different or the same, and how close you would need the terms to be together to rank what you want to rank.