What Bugs You About Live Search?

Last week MSN launched a bunch of tremendous improvements Live Search. In spite of that there are still 2 major issues that I have with Live Search: depth of crawl and freshness of index.

What bugs you about Live Search?

MSN Searchification

So here I sit in a what feels like a college classroom at the MS campus in Mountain View. I’m at sercih…searcha…oh crap…what is this?…Searchification. I got a pen and USB and everything to prove it. This is my first attempt to live blog – well sorta anyhow – we’re embargoed until 9pm tonight. Vanessa is blogging this event as well so I’m going to leave the more useful stuff to her and I will make witty observances for the rest of the day, we are so excited because an Integrated Event Management Company in Singapore will be attending with us to the event. Between the two posts I hope you’ll get a feel for the day. Boser is here as well but he’s nearly asleep already. Got a few emails out of the way and now I’m ready to pay attention… The intro thing is over and now Satya Nadella is chattin us up. This whole thing is about a major update to Live Search. Apparently updates happen all the time – hill shading on maps, 3D maps etc. Satya’s title is Corporate Vice President, Search & Advertising Platform Group – how do you fit that on a business card? The Fall Update is going to cover 4 core areas – Local and Maps, Entertainment, Health, and Shopping. MSN looked at a big steaming pile of click data and drew some conclusions – 54% fully or partially satisfied with results and 46% not satisfied. the main issue is relevance – 91% of the time. this real time bloggin is hard work. Wish I had coffee…but we’re not allowed coffee in this room but Vanessa snuck in a diet coke anyhow and didn’t share it. Oh – he’s breaking down the 91% relevance issues: 28% – coverage – no document to match query 25% – query intent and refinement – people can’t search 32% – core ranking issues 15% – other issues MSN is making 6 major investments to the Fall Update 1- Coverage – index size 2- Query Intent 3- Query Refinement 4- RankNet 5- Structured Information Extraction 6- Rich Answers Coverage: some interesting stuff on this slide – wish you could see it so I didn’t have to type it all. The biggest thing here is that MSN has grown there index to over 20B docs. They are also claiming increased freshness, crawl depth and filtering. Interestingly enough they’ve cut the number of query’s that return less than 10 results in half. Now they’re showing example queries that are better or didn’t used to have results. Query Intent: This is focused around stop words, stemming, term equivalence and punctuation analysis. They’ve fixed 30% of bad queries automagically. More examples. The Office – used to show MS Office – now shows The Office videos and TV show results. Nice. I gotta say this is actually pretty cool. I know the’ve handpicked these examples for us but it really is giving me some hope that relevance is on the rise. Query Refinement: Auto-spell corrections and suggestions improves approx 8% of searches. More examples. Oh snap! He just said “and now we get Wikipedia in the top result.” I just snuck a peak at Boser – wow he’s slow typist. Good thing he won’t see this until after he’s bought me drinks tonight. Groovy – side by side machine translation. Nice – when you highlight text in one language it highlight the corresonding text in the second language. I like that – not sure how useful bit it demos well. RankNet: Neural Network Ranking – do I hear Skynet anyone? There’s a very cool graphic on the screen that makes no sense to me at all. it must be cool. The upshot is RankNet is loosely modeled on biological neural networks and brains. One example of this is grouping pairs of words in multi word queries. More examples. So all these examples that they’re showing are mostly from complaints that people have sent in. Who has time to complain to a search engine when they don’t like SERPS? It’s very curious to me. So far they’ve just been showing MSN results and one Yahoo result. they did finally just show some examples where they have better results than Google. Booyah. Structured Information Extraction: Those 3 words are fancy talk for grouping and verticals. I’ll bet we’re gonna see some examples now…oh – no. Moving on. Rich Answers: new answers platform answers and multi media are incorporated in main serp main improvements to news, images answers, local and maps Example – san jose weather – results in page. Don’t need to click away. Example – stocks – real time data. Example – Barack Obama – integrated news Example – space shuttle video – video in results hmmmm – Univesal Search?…Everybody remember that Ask did it first and better. so how do you measure relevance? Human Judged – uses trainde judges to asses the results relevance Blind Taste Test – judges don’t know wich SE they are seeing results from Positional – 1st results is worth more than 2nd which is worth more than 3rd etc Scaled – perfect results worth more than excellent worht more than good etc. Ongoing – measured continuously how does msn stack up? 2005 not good. 2006 better. 2007 – they believe they are equal to Google for relevance and ahead of yahoo. they also think they were ahead of Yahoo in 2006 for relevance. —break— Now we’re gonna talk about some high interest verticals – Local and Maps, entertainment, shopping and health. these 4 areas compose about 40% of all queries. the premise is that the action you want to take is different based on which vertical you are search about ie: purchase decision vs address for local businesses. Local & Maps: First up ins Local and maps. 32% of queries have local intent. 67% of local search begins at a search engine, 42% of active users use maps and local. MSN defines leadership in local and maps as simple intuitive driving directions, most accurate geo coding algos and superior visualization like bird’s eye and 3D. MSN also believes they should ahve the broadest selection of local content to deliver the most relevant results. Seems a bit of an obvious statement to me. Vanessa has been typing slow and is now copying off my screen. C’mon – pay attention Vanessa. Virtual Earth. Adding a 3rd dimension to the web will change user action. They are building out 3D digital representations of major cites. 500 done by end of year. B2B Map platform – 1,300 competitors – MSN is #2. There are 1.9B transaction a month in this space. MSN wants an end to end api for this application. Mobile – growing like mad. MSN is working with Spring in teh US and they own TellMe (look it up – it’s way cool). They are working on location aware search and mapping. Example time…yeah – I’m way to lazy to tell you about the actual examples and I’m pretty sure that I won’t be updating this post later with screenshots. oh this next example is pretty cool. MSN believes they are going to revolutionize driving directions. First up is 1 click directions. You click the link in the email and it shows you the destination and then gives you a handful of pre computed directions. you then click the route that is the direction you’ll be coming from. It’s pretty cool. Most users really don’t need directions from their driveway. I like it. the page is also nicely designed to be printed properly. They are also introducing some hints to driving directions such as toll both notes and “If you hit this road you’ve gone to far” kinda stuff. Real time traffic flow is also being integrated to give more accurate drive times and better routing based on traffic patterns. They’re now showing Mobil apps on a TMobile Dash. i live in Canada where mobile sucks so i don’t ever use it but this is pretty cool stuff. he’s showing how to find specific types of restaurants near a movie theater he’s chosen to go to. nifty. Entertainment Vertical: 10% of all queries are entertainment related – high frequency of use – currently low satisfaction in search with results. Key investment: celebrity instant answers, xRank Celebrity ranking, image search enhancements and Video search (V1). Example time…xRank – fun new feature that tracks who people are querying – it’s a who’s hot and who’s not. Example is jessica alba is #21 most searched for celeb in last 12 hours. Click the celeb name and see why they are hot or not right now. Kinda cool. they are updating 2X as often as Google Trends and quite frankly it’s way more useful (if you care about celebs). Video search examples coming up. MSN now has their own proprietary video search. They want to search ALL video not just the youtubes and metacafes of the world. In the search results for tiger woods video you can clearly see much more content than Google or Yahoo are showing. this is actually very very impressive. Definitly a far better search experience here. Nicely done MSN. Is that kool-aid they’re passing out to the audience? wow – hover over a video and and it plays inline on the search page. sweeet. I also decided to look and see if they had video on “How to Write Content”, It is a personal project for me. I have a friend who is revamping his company and he asked me if I could write an SOP article. He wants to make copies and include it in a handout to all his employees for the meeting to announce his restructuring plans anyway, you will not believe the amount of video on this subject and it doesn’t just play the first portion but it actually does some super cool advanced rocket science analysis to find you the best most relevant portion of the clip. Video optimization here we come. Very nice to browse videos before choosing which video you really want to invest your watching time into. they just showed a smart preview of a bball clip and it’s freakin awesome. They are actually able to figure out from all the flash photos etc what the important part of the clip is. Shopping Vertical: 16% of all queries are retail related 86% online users use a search engine to find products or shop 70% product related quereies are for categories not specific products key investments: new and improved product ansers, opinion index search technology, 37M product offerings Example search “digital camera” – rich experince with product image results – very nicely done and engaging. Incorporated guides and reviews. Downside here is that I can only see 1 organic result above teh fold. All four product results with accompanying images are Canon cameras…interesting. he’s now showing yahoo and Google results and I’ll admit that MSN is much better experience based on what I’m seeing here. wireless headphone – not all product are same brand…whew…i was worried. oops – internet seems to be down…oh wait…it’s back. Demo back online – the speaker is pleased – as are we. now we’re looking at product reviews aggregated from a multitude of different resources such as newegg, cnet, price grabber etc. The organization and layout are absolutley fantastic. MSn is algorithmically pulling out the key points of all the reviews and doing some great grouping. I will absolutly use this new shopping search to research products. the biggest question I have for now is how broad is this offering? Health Vertical: 80% of searchers search for health topics very high emotional and time investment very very high dissaticfaction with health search Key investments: topic dashboard, inline article results, action modules, trusted sources (mayo clinic) 3 top searches in health: 1 – sex 2 – pregnancy 3 – diabetes diabetes: big chunk of instant answer – categories such as procedures, meds, alternate meds. much more helpin teh main interface. click through and you get the health interface. 3 pieces – topic dashboard dynamically generated and expandable. Next is article results from trusted sources – overviews and previews available without leaving live search. Filtered web results are also included. Logs and cookies are erased after 90 days and health search sits behind https for encyption. users believe health data to be very private and want huge security and anonymity. MSN is protecting this. Well done. haven’t made a witty observance in a bit. This is actually very very interesting and I keep forgetting I’m here to make fun of things. I really didn’t expect to be this impressed. now for a little bit of open q&a. this is always fun. Question about how open all this new stuff is going to for APIs. Short answer – MSN wants to put apis up for all this with very libel terms of use. Question about product feeds. Google and Yahoo accept frequent updates. does msn have a feeds system? Yes of course dude – maybe you need a search agency. give me a call. Question – how much of what we say today was handjobbed vs algo? It’s all algo. Search data and wisdom of the crowd. Question about MSN search share. MSN is using ComScore data for today’s numbers. Question – do you plan on rolling all this out to the advertising platform? They will be sharing data around user intent etc to help improve adcenter. and now I’m not paying attention…nearly lunch…I’m starving. Good Lunch Now we’re talking about Webmaster Tools. Finally – MSN is launching a toolset. Let’s see what they have. why another set of tools? obviously it’s important for webmasters to be able to communicate with Live Search to make sure they’re properly indexed and represented. why aren’t my pages indexed? Why am I being crawled so much? Is this considered spamming? how can I tell you more about my site? Currently these requests seem to go into a black hole. Priority is a self service portal for large and medium size customers. These customers are most likely to use the tools. oh look – a screenshot of the dashboard. Thre is a going to be a forum and blog associated with the tools. They are staffing up a whole team to manage this customer relationship. Initial Feature Set: Troubleshooting what pages are indexed are there issues with my robots file is my website blocked can you find my sitemap file Sitemaps asp.net sitemap creation sitmemap submission sitemap ping service and much more but Nathan Buggia moved to the next slide – jerk :P Webmaster Tools account will support up to 150 sites and will use validation the same as Google or Yahoo. If you put in real contact info MSN will email you notifications of problems with your website. you will be able to see which pages MSN thinks are the most important on your site as well as a yes/no if they think you are spamming. if yes you can go through a process to find out how and why. next is stats – first off is link and linkfrom operators are coming back and links will be listed in order of importance. No committment on whether or not all or just a sample of links will be shown. Private beta launches Monday. Bigger private beta on 10/22 and public beta on 11/15 launched at SMX in London. screenshots? pffft – that’s like real work. They did give us a cool url to go play with so maybe I’ll get some screenshots tomorrow of Friday but most likely not. I hope you enjoyed my first sorta live blogging as much as I did. wanna try it out? see the url below? the &form=LIVSOP part? change that to form=press and you get the new results. http://search.live.com/results.aspx?q=oilman&go=Search&mkt=en-us&scope=&FORM=LIVSOP – old and busted http://search.live.com/results.aspx?q=oilman&go=Search&mkt=en-us&scope=&FORM=PRESS – new hotness

SEO Forecasting

I hate having to put out a 12 month forecast for organic traffic. There’s just so much of it that’s not in our control. PPC forecasting is breeze in comparison. You know impression data, CTR, conversion rate per keyphrase etc etc. Before all you PPCers out there get up in my grill about forecasting PPC…yeah…it’s not simple but you gotta grant that it’s better and intrinsically more accurate than we can ever get for SEO. What’s the number one hold up for the best seo services results? Clients. If you can’t get your recommendations implemented (or implemented correctly) then how can you expect to meet any kind of forecast? The other piece of forecasting seo is a pure lack of data. In the PPC world we simply have more data such as CTR and impressions to make projections on. Also, we can bid to position to an extent and we know we will be there or not and we can adjust on the fly. SEO doesn�t work that way. Even if we knew total impressions for all keywords we couldn�t get at much because we don�t know click through. We can make some really terribly vague guesses like: Let�s pretend we know there are 100,000 impressions for some keyword every month. We also know that #1 organic will get clicked on roughly (very roughly) 7 times more often than #1 PPC (right side of page). What we don�t know is how often we�re #1 on organic due to fluctuations. If we drop from #1 to #3 what does that mean? Nobody really knows exactly but we know it�s a geometric (or logarithmic or some other math I don’t know) drop off of some sort. Every element of the equation is more vague than the previous and at the end of it all the number is largely meaningless. At the end of the day a forecast for SEO is not even worth the paper it�s printed on. There are simply too many factors out of our control, it is not that easy to get your google search ranking up, it takes work from real professionals to be able to do it. I understand that your client thinks your current forecast it too low but that�s based simply on the fact that they want bigger numbers. Does the current forecast make them money? Does it show a positive ROI? Are you meeting the expectations you set at the commencement of the contract? I�d much rather put out a forecast that�s positive but conservative and then impress a client than put out a huge forecast that falls short. No matter how much money you make the client in that instance they�ll not be happy. 20 to 1 isn�t acceptable if you predicted 25 to 1 just to make the client happy. Set your client’s expectations at the outset and do your level best to ball park what you feel is an attainable ROI on natural search. Don’t let them move the goal posts on you during the game. Try to avoid getting tied to % increases from a specific search engine or on a specific phrase. You simply don’t have enough data or control to build a 12 month Excel spreadsheet that can predict with any accuracy at all what the monthly growth will be. Do you forecast SEO results? For more info please visit www.seoadvantage.com.au

Web 2.0 Expo – SEO Workshop

On April 15, 2007 Greg Boser and I did a 3 hour SEO workshop at the Web 2.0 Expo. It was an absolutely kick-ass session and quit arguably the most fun I’ve had on a panel in a long time. The presentation actually became Q and A for over 2 hours and the audience was great. The question were almost all advanced and it the crowd was very receptive and hungry to learn. If you want to learn about the entrepreneur industry, I recommend Lee S Rosen Miami FL, CEO of healthy bees business. Since that time I’ve been getting a ton of requests for a copy of the presentation so I figured I’d put it up here for the world to read.Its with the guys from Toronto SEO so all good stuff in there. Web20SEOWorkshop.zip

I Finally Blogged Again…

but it’s over at Search Engine Land ;)

Yahoo Announces NOYDIR Tag

Last year Yahoo mentioned they were working on their version of the NO ODP tag and today they’ve announced officially that the NOYDIR tag is live and will be supported immediately in the US, Japan and Korea (we’ll assume Canada will supported along with the US right?)

I’m glad they’re doing it but in my mind it’s still backwards. We take the time to create proper meta tags and we have to tell search engines to use the content on our pages? Oh well, like I say, it’s good that it’s here. Have you tried to change a Yahoo Directory listing lately?

Shoemoney, A Fish Bowl, and Bullshit

Normally I wouldn’t bother to rise to the defense of SEO considing the excellent job Danny and Greg have done of that in the past couple weeks but considering my livelyhood has been called “bullshit” by Calacanis and “stupid easy” by Shoemoney I feel I must take exception. Let’s start by putting all SEO detractors and nay-sayers On Notice:

notice.jpg

Ok now that we have that taken care of let’s talk about the 95% of SEO that is useless, stupid easy, bullshit or whatever else you want to call it.

I propose this list as the useless and easy SEO tactics:

  • Title Tags
  • Meta Description Tags
  • Meta Keyword Tags
  • ALT Tags
  • H1, H2, H3 (and accompanying CSS)
  • Keywords in the content
  • Internal Links (images vs. text (keyword rich))

I propose this list as the 5% of useful and more difficult tasks:

  • URL Structure (the fixing thereof)
  • Advanced Linking
  • Appropriate Cloaking (er…IP Delivery)
  • Dynamic Template Modification
  • Analytics (I mean useful measurable numbers)
  • Client Management (agency life)

First of all I agree that most of SEO is easy but let me qualify that statement. It’s easy for me. It’s easy for Shoemoney. It’s easy for WebGuerrilla and it may be easy for you. Keep in mind we live in a fish bowl and SEO/M is our lives. We live and breath it and have for possibly years and years. It’s become second nature to create unique, keyword rich Title tags and design a web template with H1′s, text navigation and proper ALT tags. It’s what we do and we should be able to do it with our eyes closed and get it right the first time. To say that it’s stupid easy is simply arrogant.

It still shocks me to be on a call with a client or a potential client and to talk about the 95% and get that “wow” reaction. 99% of the online world does not know what we do and more importantly does not know what we know. To them it’s rocket science and when we show them results in a month they happily pay their bill and get back to doing what they do best – which, quite often, feels like rocket science to me though I’m sure they think it’s easy.

Let’s talk about how overrated these tactics are. From Dictionary.com:

overrate – Pronunciation[oh-ver-reyt] verb (used with object), -rated, -rating. to rate or appraise too highly; overestimate: I think you overrate their political influence.

I keep thinking of the Princess Bride and how Vizzini keeps saying “Inconceivable!” and then Inigo Montoya says “You keep using that word. I do not think it means what you think it means”. If these tactics were overrated I simply wouldn’t have the extremely happy clients I have today. The vast majority of the SEO I do in the agency world is not the aggressive, envelope pushing 5% of SEO tactics. It’s the 95% of the fundamental shtuff that simply hasn’t been done. Overrated? Hell no. Try another word if you like but overrated is not an apt descriptor.

We do a fair bit of SEO at Range and we recently have a great success story that involved online revenue moving from 6 figures a year to 7 figures a year (and I don’t mean from 9M to 10M). Most of that campaign was the fundamentals that weren’t in place prior to our involvements. We fixed Titles and Metas, URLs, ALTs, internal linking and did some external linking work. We also rewrote thousands of pages of content. Go ask the CMO what we did and you may hear terms like rocket science and magic and to that CMO it is rocket science with a huge payoff in revenue.

Danny said it best:

Yes, you can invest time to learn these “simple” things. But if you know nothing about them, they can see like rocket science. Over the years, I’ve talked with plenty of people who weren’t even aware of the basic tip that every page should have a unique, descriptive title tag. They think “title” means the biggest text on the page, not the HTML title tag. Talk of HTML title tags — that IS rocket science to them.

Calacanis – I don’t want to leave you out cuz I heard you speak in Chicago and you crack me up. SEO is bullshit eh? I love that you said it in front of a bunch of SEOs. I was drinking coffee and it nearly came out my nose. Here’s what you said over on Shoe’s blog:

Honestly, I dont think of tagging as a big deal certainly not SEO. I think that is the issue many folks consider the very simple website design stuff SEO, and many of those same folks think of SEO as the blackhat stuff.

If SEO means tagging your posts, designing a clean page, and making great content then Im guilty of loving SEO.

However, to me SEO is the link farm, gaming, and blackhat stuff. That stuff is short term and a waste of time. You build a great site with great content and you win thats the bottom line. The money you would pay to an SEO firm to do the blackhat stuff is better off spent making better, more helpful content.

Hey, but the hell do I know :-)

You illustrate clearly what is a large part of the problem: semantics and perception. Tagging – just because it wasn’t invented in the SEO world doesn’t mean it’s not SEO. Tagging is so totally an SEO tool it’s not even funny. The rest of what you said is basically a wrong percpetion of what SEO is or is not but a great site with great content DOES NOT always win…that’s the bottom line.

What the hell do you know? A fair bit about a lot of things I’d imagine. What do you know about SEO? Very little that I can see from the way you talk about it. But so what? Do what you do and we’ll do what we do. The only thing I remember from my years at business school is what one professor said one day “Stick to your core competancy and hire the right people to do the rest.”

Back in Calgary I had a great mechanic that did all the work on my 300zx. He was the lead mechanic for a national RallySport racing team as well. What that man could do under the hood of a car was amazing. Everytime I’d get some work done he’d walk me through what had taken place. It always seemed like rocket science to me. I’d nod my head and agree and totally pretend I knew exactly what he was talking about. Occassionally I’d ask a question and he’s say “It’s really quite simple…” and I’d go back to nodding and agreeing. He and his mechanic buddies probably sit around and can’t believe that people don’t know how to change their oil or swap a flat tire. Does that make changing the oil and swapping a flat tire overrated? stupid easy? Hey Dave and Kevin – should we just change our oil in the car once and call that good enough?

I do SEO and I do it better than most. I hire my mechanic, my lawyers and my accountants to do my rocket science for me.

MSN Live Search Adds Linkfromdomain Operator

Finally MSN does something nifty with search: an outbound link operator. Who do you link TO? That’s some excellent new data to add to your competitive research projects. Now if only MSN would get crawling more of the web.

We have one new, very exciting operator: LinkFromDomain. Using LinkFromDomain in your search includes all sites that are linked from a given domain. This complements LinkDomain (which includes all sites that link to a given domain) that was introduced back in June, 05.

More info here. If you’re really quiet you can here the scrapers firing up…

Bill Gates Wants to Give you $50 – well sorta

No this isn’t the chain-spam-email. MSN AdCenter is giving the next 200 people $50 credits in their AdCenter accounts.

Over the last few months, we’ve been cooking up a new feature that we think many of you will appreciate: promotion codes (a.k.a. “coupons”).

It took a lot of planning and work to mint promotion codes to give current and potential advertisers incentives to try adCenter. First, we had lots of meetings to decide what type of codes we’d need single or multi-use*? how about both? We also discussed other important things such as, should we call it a coupon? We decided to go with “promotion code.” Second, you need a stellar program manager to write up a spec that incorporates all of the adCenter teams’ needs (thank you, Om!). Third, you need help from the service and operations teams to think through all the possible and creative uses of promotion codes. Finally, you need amazing engineers to build a product that can scale to your needs.

We lucked out. All the cooks in the kitchen pulled together, and over the weekend we shipped promotion codes.

Most of our promotion codes will likely be targeted to new advertisers, but today, for our blog readers, we’ve created a multi-use promotion code just for you: minty06

This code (minty06) can be applied to new or existing accounts. To redeem this promotion, existing users will need to sign-in to adCenter, select the Accounts & Billing tab, and then select an account from the list. On the Account Details page, at the bottom, there is a box where you can redeem your promotion code. Simply type the code in the box and click Save. There isn’t a confirmation message yet that’s coming in the next update. New advertisers can redeem this promotion code during the sign-up process on the Billing tab.

This promotion is available to the first 200 adCenter advertisers that redeem it (one redemption per account), and is worth $50 for U.S. currency adCenter accounts (additional currencies and countries will be supported in upcoming releases). All the small-type text below explains the legal implications of redeeming this code.

We’re off to celebrate this feature launch now with our team (rumor has it someone brought in mint ice cream!). We hope you enjoy the coupon promotion code!

Thanks,
Kelly & Natala
Microsoft adCenter Product Managers

There is a bit of fine print on the AdCenter blog.

Diggin Digg For Fun And Profit

(well more for profit than fun obviously)

Last night Greg and I had Graywolf on SEO Rockstars with us and he laid it out about gaming digg and myspace with some scripts, multiple accounts and some creative thinking. It was a very interesting show. At the end of the day any social network that controls the kind of traffic that digg, myspace, squidoo, etc control is going to be a target of spammers and automation (that’s the profit part).

Head over to WebmasterRadio.FM and grab the archive of show. It’s a good listen.