Trying to understand Google’s algorithm

[DannyS]

Sorry if this is a little long-winded but I’m trying to figure out what I see using Daniel Brandt’s Scroogle. At the end of the last year using the original Scroogle I found that with some popular two keyword phrases in our field, some of sites rated very high in Scroogle but not in Google.

I can’t complain that Florida kicked us out of the top page of Google on these keywords because I don’t think we were ever there – they were fairly new web sites and we were (are) still working on them. With the next Google change we saw the same thing happening with some less popular keyword phrases and some three word phrases but this time we did lose position in Google.

Using Scroogle’s new method, adding +www, we can see similar results. Working from Scroogle results being Google’s before the hilltop algorithm is applied I’m in two minds as to whether the algorithm applies a punishment for doing something wrong, a bonus for doing something right, or both. I’m hoping that there is no punishment in this algorithm so it is just a bonus applied after the first search has been done.

Using the ALLINTEXT search modifier gives results closer to Scroogle’s than Google’s but they are not the same. It does seem that most of the off-page part of Google’s algorithm has been moved into the hilltop algorithm applied in the second stage of search filtering.

So a little test, done more by accident or work commitments, than by design. We had two sites that ranked very high, 1 & 2, on a two word phrase before the second Google application of hilltop, (was that Austin?). Both were out of the top 1000 afterwards. On one of the sites we added some more link building while on the other we pretty well left it alone. The one we worked gradually moved back into the rankings, all be it at a lowly 147, while the other stayed out.

I am of course beginning to suspect the obvious on link building but the one thing that puzzles me is why sites drop out completely, like the top 1000 results. Are there really 1000 other sites with better off-page results or does Google lose interest after a few hundred? If I had an answer to that last question it would stop me waking up in the middle of the night wondering if there was a punishment component in hilltop.

Danny

[cbp]

Ignore Scroogle.

If you really want to understand Google algo, check this thread:
http://www.webproworld.com/viewtopic.php?t=10802

CBP

[DannyS]

Hello CDP

I'm not trying to propagate any conspiracy theories but just trying to understand how the new algorithm works. What we’ve seen does seem to reinforce Gupta’s idea on how it’s working. What interests me is what we are actually seeing using the added +www. Is it bypassing part of the algorithm. If it is, it looks like most of the off page factors are built into a second part of the algorithm. If nothing else we can use it to test the on page optimization is good;-)

Danny

[DannyS]

Hello CBP (sorry for the CDP last time)

I forgot to add the big question again. Why would the application of the new algorithm cause a page to drop from No. 1 to not even in the top 1000? Could it be an error in the algorithm or are there really 1000 sites which such better off-page factors.

When we have done a little work we can get the page to jump from nowhere to 147. I think that results in the deeper regions look very suspect.

Danny

[mtbot]

This is the one-million dollar question. I can't wait to hear answers for this.

[albatross2147]

Have the changes made by Google gone too far?

Previously, a client of ours, a home improvement business based in and serving this city was listed number one for the search: home extensions sydney

The other top results were relevant too.

Now the results (on both .com and .com.au) are either for wierd internet directories and totally irrelevant results.

It seems that our client is being penalised for having the words "extensions", "Sydney" and "home" incorporated in their site text etc.

If people can't get relevant results from their searches (even the Ad Words on the resluts page are a bit wonky) they will vote with their feet errr mouses

[DannyS]

My apologies as I gave some bad info in the first post. I'm using the wonderful keyword tracker at www.digitalpoint.com and it only looks at the first 200 results so my "out of the top 1000" probably means "out of the top 200".

Still my main question is, what are we looking at when we add +www to the keyword phrase? Is it the pre-hiltop algorithm? Is the new algorithm applied after the original algorithm has already been used?

The latter doesn't seem to make much sense as I would think this would be slower and use more resources at Google. Running two algorithms must be a bit of a pain as they would have to put the search phrase through some sort of look-up process first. How long before every search goes through the new algorithm?

I guess it is just what most are saying - to do well with the new algorithm you have to improve the quality and quantity of incoming links. So, with what I've seen, this is how I'm going to move forward:

On Google use the following:

"ALLINTEXT:" with search phrase to check out the on-page factors are optimized.

Search phrase with +www just to see if we are moving in the right direction with off-page factors.

Search phrase alone to see if we made it;-)

If the second and the third are the same then you either have a very good site or the phrase isn't one that the new algorithm has been applied to yet. Does that sound sensible?

Danny

[mtbot]

Well, as you indicated earlier, I think you need to focus entirely on off-page factors (link building).

In my limited experience, "allinanchor:" (anchor text search) is the best indicator for relatively non-competitive keywords.

For competitive keywords, "allinanchor:" is only somewhat reflective of the current SERPS because of the unknown (e.g. Hilltop) algorithm factors. For example, I come up as number one using "allinanchor:" and "allintext:" for my keyword "medical transcriptionist" but I am not in the top 1000 results at GoogleRankings.com. My hope is that after the next internal PR/backlinks update, my site will appear in the top ten.

(In my experience, "allintext:" usually shows the same results as "allinanchor:" unless the keywords are unpopular for anchor text.)

[DannyS]

Thanks Daniel

I was thinking I was getting on top of it, but once I try and apply a little logic I see I'm further away;-)

The problem is whether it's allintext, allinanchor, Scroogle's +www or a less popular keyword the new algorithm isn't being used. I have to think it's the old one still in use.

The more I look at it the more it seems that the new algorithm is not being applied on results given by the original one, so out goes the idea of a filter. It's just a new and totally different algorithm which will eventually take over all the searches.

Here's an example I looked at today with a popular keyword but probably not the major one for the sites, although the allinanchor is probably loaded towards site #2:

Site #1, Google pos. 17, allintext pos. 140, allinanchor pos. 124, scroogle pos. 9

Site #2, Google pos. 241, allintext pos. 4, allinanchor pos. 5, scroogle pos. 2

The difference is of course links:

Site #1, All the web 2000+, Google Link: 22

Site #2, All the web 50+, Google Link: 14

No getting away from it - I had better get back to link building;-)

Danny