5-25 MSN Search Update Continued

[msndude]

Continued from [webmasterworld.com...]

At the expense of repeating myself, I'm going to repeat myself. :-) Here are the guidelines I'm trying to follow when I post here:

I won't discuss the problems of particular sites in any public forum. As long as I can handle the volume, I'm happy to get "sticky" messages, though.

I also won't discuss competitors here -- they have their own forums, and I figure they already have enough people talking about them. :-)

I will not preannounce anything specific here; I might say "we're working extra hard over the summer to fight Spam," but I'm not going to post something like "look for a giant spam-killer release on April first!" After a release goes out and someone notices it, THEN I'll comment on it. If no one notices it, it probably wasn't worth talking about anyway.

Finally, I'm not going to reveal any secret Intellectual Property that would help someone build their own search engine or hack ours. Most of it's hard to explain anyway. :-)

[edited by: engine at 1:54 pm (utc) on June 5, 2006]

[stinkfoot]

Sounds like a good set of guidelines to me. Only thing I cant think of asking is ... what does MSN's vision of the internet consist of.

What I have seen of it over the last year is it has fast indexing of new information, making it a useful tool especially for research.

Seems to have got a major portion of the SPAM out of the index and when I say spam I mean keyword stuffed documents and black hat seo tricks.

Seems to be making good information easy to find.

I have a feeling these are part of the vision but .. is there anything that I have missed that will play a major part in it?

[SmartMoney]

msndude..a question. search.msn.com is only returning 25 pages of results, the 25th page is repeated for every page link after that. Is the limit on results an 'update' or a bug?

it's always been like that.

[Garya]

MS Dude just to show you how worthless msn search is.
I can show 1000 of bad results.
Here is one (auctions) compared to yahoo and google.
MSN
1. www.fcc.gov/wtb/auctions
2. www.treas.gov/auctions

Yahoo
1. www.ebay.com
2. auctions.yahoo.com

Google
1. auctions.yahoo.com
2. www.ebay.com

I guess if you want unpopular sites that nobody cares about then MSN rules.

[asiaseo]

We were more than happy with arabia results but we no longer see that and the results have changed and local content has gone down, back to seeing very little local content.

[asiaseo]

I have just tried ninemsn ( Australia ) and results for ANZ / Asia are good! That is without clicking Australia only, just the normal, the results seem fair and good.

msndude : If that is going to stick for this part of the world then certainly over here I think people will be happy, I always try searches outside of our own area of interest for some general search and those ninemsn results come up with lots of local content and direct websites as opposed to just sites with a few words or a link.

[followgreg]

Until now I 've always found that MSN was heading towards real good SERP's in a matter that Google could worry.

But seriously the more I query MSN since this update the more I feel that these guys won't make it in my lifetime, and I am still quite young and healthy.

One word queries, even some of the most obvious ones are, across the board just so wrong.

Is it that complicated to find real authority sites?

Is it THAT complicated to filter subdomains, hyphenated domains, blogs which florished on top of MSN during this update?

I don't pretend to know better than MSN engineers, I would love to see a more equitabl market share in the search world, but what I am sure of is that if my being a CEO at MS search there would be a few that could worry about the next paycheck. :)

This being said, it is hopefully a moment of transition - MSN dude knows better I guess.

[Rufus_dog]

A sophisticated search engine should not filter sites out based on their urls. There are legitimate blog sites, forums, subdomains and hyphenated domains which offer content. These poor people can't change their domain names just because some others use unethical techniques to get high ranking. What MSN needs to do is find a formula to analyse the content of the site (filter out keyword spam and those using black hat SEO) and the quality of the backlinks. A site with 1000 pages or one with several thousand backlinks is not necessarily a good site(some software can generate keyword stuff webpages and link farms can easily get thousands of links in several weeks).

On the other hand, MSN should not put too much emphasis on keywords in url either, which seems to be a major problem in the latest update. Many people use their own company names as domain names, they can't insert keywords into their domain names just to get high ranking.

[1984bb]

I never show such a big thread about MSN in my life ,hmmmm there must be some reasons
....all the Google contras (lost my rank on Google now I love MSN ech..)posting here to make a statement...:)
As about the update I was not affected I hold my top positions but the traffic MSN sends is just breadcrumms.

[zeus]

Hmm now i have the msn.com search again, but I dont like those results, the arabia was better.

[dclick]

msndude, i defenietly appreciate your input and listening to our feedback, like econman stated.
i dont expect you to switch back, etc..

But going forward i really recommend 2 basic fundamental relevancy tests included with your testing procedure.

1. Is my own website, search.msn.com found anywhere in the results for common phrases?
msn search needs to be found somewhere for its most basic phrases/keywords: internet search engine, online search,internet search, search engine (this phrase is found, but msn beta domain appears?)

2. The place/webmnasterworld i am telling everyone about the new results/or news about our search engine, is WebmasterWorld found for common keywords/basic terms?

I think these 2 basic fundamental relevancy tests have to be in there to call an update a true success (in my opinion)... as far as relevancy goes.

Think of it like the phonebook. a search engine is basically a search of a phonebook of websites in a similar basic fashion. Verizon hands out a yellow pagse phonebook, but verizon itself isnt listed under the Telephone Company section. Likewise verizon announces their new updated phonebook in the newspaper, but the newspaper itself isnt listed anywhere in the phonebook where you would expect. At that point, something is off when the telephone company itself isnt found and the place it is announcing/discussing new features of the phonebook isnt there either.

actually speaking of telephone here is another good test. Search msn for: telephone company
no at&t, qwest, bellsouth, verizon, etc..
i expect at least 1 or 2 to show up somewhere for a basic query.

2 basic tests, Is my own domain search.msn.com found in the serps for typical / common keyword phrases and is the site I am announcing updates/news on WebmasterWorld, found for somewhere reasonable for its typical phrases.

i dont expect you to revert back to the old results either.. i thought they lost some relevancy/took a dive back in the jan/feb update.

but hopefully in your experimental sets you can play around with it and these 2 basic tests will appear somewhere within the first couple of pages. Keep at it and i defenitely appreciate your response and listening to our feedback and suggestions.

[SmartMoney]

...no disrespect intended msndude, like others on here I find it amazing that you're here taking the heat, but the more I read this thread I keep thinking:

Why don't we start a collective [open source?] search engine project...with the goal of being taken over by one of the three majors wanting to improve their algo?

Anyone?

[msndude]

Oh we have thick skins here (scales, actually), and we're used to heat too. :-)

Seriously, there has been a lot of great feedback here -- whatever the tone was -- and we appreciate it.

[SmartMoney]

Oh we have thick skins here (scales, actually)...

that's a symptom of working weekends...I know the stock options are great over there in Redmond, but even developers need a little time outdoors once and while :)

[Shurik]

Msndude, thank you for listening.
Even after yesterday index refresh the results are still lacking quality. I believe the main reason is not keywords in domain as many keep pointing out. The main reason is how you now treat IBLs. An average web site may have a large portion of IBLs with the domain name as an anchor text. Keywords stuffed domains have natural advantage. Because of the increased emphasis on the exact anchor text the keyword stuffed domains (blogspot is just a small part of it) have surfaced rather nicely. Combined with the mysterious "site decapitation penalty" (lose of homepage for quality sites) and de-emphasis of the on page factors you have a rather weak index.

Please do some analysis of the current index.
1) Pick 10,000 most competitive keywords (you should not have any trouble in this department)
2) Collect the first 10 results for each
3) Query the results for URLs with keywords in the domain name.
4) Examine the URLs with keywords in them, and the overall density of such URLs
5) Run the same against old index and major competitors. Compare the results.

I’m sure with the use of wonderful Microsoft SQL Server Reporting Services it won't take you long to put together a report like this. Once you visit a few URLs from the list you may become a bit less excited about the new "net".

My overall impression is that you did remove lots of spam but introduced different kind of spam at the expense of relevancy.

[asiaseo]

As Rufus_dog said there are many genuine wesbites wityh a hyphernated url, in cases where we did the domain that way was purely because to put the words together was weird, this applies in particular to some local small business names and also due to rather complex spellings people might get the name confused, just because some people have used it unfairly it would be wrong to say that any hyphernated domain name was any more or less worthy than a goups of words joined together.
My rant is the idea that having thousands of links to a site somehow makes it of more value to the searcher. We all know that out there are sites ranking high for a term simply because way down on pages and pages of links they have that search term or name appearing in a link, and they get that position because of those reciprocal links, simple.
There has also been a long discussion elsewhere on this forum about using underscores, hyphen or '.' in file names, and some of the comments led people to believe that using a hyphen was in fact better than an underscore ( can this get any more complex! )

[Mbwto]

I see lots and lots of pages of wikipedia getting top rank. It was simillar to what ebay use to be in google 2 years back. For most of the keywors you will see wikipedia in top 10. If I want to search wikipedia, why will I come to MSN.

[Rx Recruiters]

I agree with the rest of the board. I actually use the msn search via A9, and after this last update, I've switched back to another engine. The results are terrible - it is so obvious for the terms I search - sites that used to rank in the hundreds but had key words in the domain name are now ranking in the top ten. Most of them are not even very relevent to the search, other than a key word being in the domain name. This update brings back thoughts of how Yahoo used to reward sites that had the keyword rich domain names, which led to the hyphenated domain spam in their directory that we still see some of now.

I loved the old A9 search results (after they went with MSN), but definitely won't be using the new ones. One search I did brought up an entire top 7 consisting of subpages of one site, and the next 3 in the search results were pages from another single site. Both of these sites had keyword rich domain names, and both had always ranked below #50 before the update (How do I know? - because I fanatically check search results across several search engines for a couple of terms daily)

[wariental]

MSN dude you mentioned your website can be considered spam and be deindexed(fully). Well that hasn't happened to me even though I've lost several pages in the index but I can't seem to get any new pages indexed. It used to be possible within a day or two but now I cant get anything new indexed.

I can assure you these new pages iam putting up are not spam or black hat. The only thing I can think of is msn doesn't want to index anything new comming from my domain or there has been a huge delay in getting pages into the index.

Let me know if you can.

[RichTC]

msndude

DEEP CRAWL OUR SITES, DEEP CRAWL OUR SITES, DEEP CRAWL OUR SITES!

Sorry but until your search bot is effective as Googles or Yahoos at deep crawling and indexing total site content you simply wont be able to get enough data to even start to put together half respectable SERPS.

The only sites you currently index well are bloggs and sites with less than 500 pages. Anything 10,000 upwards and you can count msn out.

Sorry but your search bot technology cant currently match that of gigablast!

[Garya]

It obvious MSN is getting worse.
So let stop all this complaining and just stop using it and don't pay for there pay per clicks untill they can be called a search engine.
Oh well I used to only use Google, but they sacaficed on relavancy
to combat spam. So many sites are not in there results.
Was using yahoo and msn til now , back to only using yahoo.
They seem to have found a balance of the 2.
1. Yahoo
2. google
3. all other search engines
4 MSN

[BillyS]

>>So let stop all this complaining and just stop using it and don't pay for there pay per clicks untill they can be called a search engine.

Garya - First off, this is a completely idle threat and ultimatums are never a good idea. Personally, when poor results are showing this is perhaps the best time to advertise on MSN.

Give the guys a chance. They didn’t buy a search engine, they are creating their own. They’ve been doing this for what - a little over a year now? Yes, I understand this is MSN and they’ve got deep pockets. But all the money in the world cannot replace experience and Google and Yahoo have a lot more experience tweaking their algos.

And I don’t think MSN cares about the complaints, they are looking for suggestions.

[CainIV]

Well put, Billy

[RichTC]

Billy... i think you should change your name to Polly Anna!

Yes msn have been in the market for a year, but they dont help their own cause by putting out statements saying what was it now?, "we will have the most relevent search results within 6 months" etc etc

Sorry but they are miles away:-

1. They havent learnt how to get the search bot to deep crawl yet

2. They are still messing about with outdated DMOZ descriptions - so much for MSN being original!

3. They still cant work out the differance between an authority site and a blogg

4. The SERPS are still the worst on the internet and imo behind that of Gigablast and ASK Jeeves

I could go on.... They have a lot to do

RichTC

[angiolo]

For me search.arabia.msn.com is one of the cleanest serp MSN has ever had.

In some sectors spammy sites are still ranking fine, but MSN is in the right way!

[futuresky]

I've got to say MSN results seems to be giving good relevance from my observations. They seem to be able to dig better results for what I need on difficult searches that Google and Yahoo give much less relevant results. I'm talking about things I'm looking for by the way, not how MSN treats my own sites. I'm popping over to MSN increasingly frequently now.

What MSN needs to consider is some strategic partnerships with non microsoft browsers (and other things). For example I use opera and it's easy to search with google (just putting "g searchterm") but to search msn I first have to go to msn.com, wait for the page to load (which is a bit slow....needs to be quickened up for faster loading in my view), and then type the search in the search box.

As for crawling, MSN seems to be doing a better job as far as I can see than any other bots...at least on a medium sixed (90 pages so far) site that I'm building where msn bots are the most regular bot visitors.

Good that indexing is fast to, thus allowing users to get up to date info. For proof of this try a relatively new search time, for example "those machines are full".
465 results on yahoo
796 results on google
1,151 results on msn

I think MSN are definitely heading in the right direction. If they can keep scraper spam out (not that I've seen much) they're on to a winner.

One thing that would be nice by the way would be if the link colour would change on the SERPS for sites you've already visited.

[Rufus_dog]

"For me search.arabia.msn.com is one of the cleanest serp MSN has ever had."

But isn't this the SERPs before the update? At least when you type "search" you will find yahoo, google, msn there. But now, when you type "search", what would you get? Hmmm, kind of hard to believe that this update is heading to the right direction. Something is missing or way off in the algo.

[jdMorgan]

I agree with the assessment in the first post of the original thread -- something is affecting the the 'authority' ranking of certain pages/sites. I have a very old (but updated), non-profit, no-ads, no-spam site with at least a dozen original articles on it that are unique on the Web. ODP liked it well enough to include it in several categories on their own. This site, though it competes against scores of sites on the same general topic that have little or no unique or widely-useful content, has been demoted from the first page of MSN results to the middle of the third page for it's definitive keyword. The only hope for this site in MSN is that it seems to be rising at the rate of about one position per day, so maybe it will recover. But if this recent update is 'over' then that probably won't happen.

The purpose of the site is completely non-commercial, so I can't justify spending the time to tweak it to suit MSN -- After all, it's been ranking fine since the days of AltaVista, and its purpose is to serve its visitors directly by providing information only. The result is that MSN users can no longer easily find the resources it offers, while users of other search services can.

The site still does fine for multiple-keyword searches, but it's historically been above the fold on page one for even the single-keyword search. Interestingly, it's the only site after the number one search result with an indented listing and a "Show more results from <this site>" link. All the other sites between the top result and ours on page three are single-page listings.

All sites in this special-interest sector are very small, so it may be a corner case, but after waiting several days to comment, it still looks like an authority issue to me. My overall impression is that results are better in some cases, worse in other cases, but inconsistent. That would be the concern, because an engineerable algorithm should be either better or worse, and not just better in some cases at the expense of other types of searches.

Jim

[msndude]

Just a quick note to say I'm not ignoring you guys, but I'm on vacation this week, so I'm just going to check in from time to time -- mainly just to respond to any kind of serious misunderstanding.

We are tracking the feedback -- especially the specifics. Thanks for sticking with us.

[Liane]

OT

msndude,

For some reason of late, I keep having to reload the MSN home page in order for anything to show up. Sometimes it takes three or four tries. I am using a Mac and Safari.

I know Mr. Gates is a PC/IE kinda guy ... but I think it would behoove MSN to at least try to accommodate those of us who aren't! ;)

(Or, I can stick with Google if you prefer?) :)