Marconi Media Blog

Understanding User Intent in Google’s eyes: The Google Intention Engine

After the leaking of the Google Quality Rater General Guidelines for 2011, a lot has been going on on the discussion boards, but it is hard to pinpoint the most valuable information in the 125 pages worth of Google quality rater material. It is a very interesting document, but without a good understanding of the way google algorithms work one cannot but guess what it’s all about. One very interesting tidbit is that Google is actually considering a person’t intention (Intent) when it looks at a search query. And the searcher Intention is the subject of Chapter 2.1 of the guidelines pdf, Understanding User Intent. The chapter explains how a rater should look at the searcher’s intention when he typed for example “tetris”, rather than just thinking abstractly about “tetris”. This would mean that by the same logic, the search results should not contain scientific or explanatory references about what tetris means, but websites where the user can actually play the game called tetris. So Google is trying to guess what your intention is with the words you’ve searched for, this is very interesting. Raters are asked to identify the actual intent of the search, which means the algorithm is doing the same thing, and how it can do that, I can’t tell. The first thing that comes to mind is that webmasters have to create signs to make the algo detect the intent, or what you can actually achieve on a website: play a game, read some lyrics, read a poem, buy a product or ask for a price. The only signal that could facilitate this – that I cann think of – is the usual call to action: “Buy x product now”, or “Order your insurance online now”. If anyone else has a better idea for how wembasters could include signals in their websites please let me know, but right now we’re going to have to presume this is one of them.

So my theory is that the algo has to detect some sort of action that can be done or achieved on the specific website, and it has to look for standard ways to signal the possibility of doing that action. An intent is followed by an action, because that is just the way things work, so in the web design world, we add a call to action and I assume it should be succeeded by some sort of form or add to shopping cart possibility. Today’s websites aren’t about what they once were, you don’t just find out that Apple has launched their new IPhone 4, but rather you are offered 2-3 ways to order the phone, directly, and you can also pay with PayPal or any other electronic payment company. So this is the intention engine. Or at least my theory about it. This would account for some of the sites that have no conversion or possible action, being hit. Like maybe eHow. Maybe the value isn’t simply in the broadness of information, but also in offering the next step to your visitor: like the product or service you’re reading about? You can order it, buy it here and now.

On another level, this seems almost like a paradigm change. Google wants to know what you want to do or where you want to get before you do. You search for tetris and it takes you to the game. Maybe it even takes you to a game with a totally different name that’s actually tetris. The question is: is this all good? And how does this affect SEOs? Do we have to optimise in a different way to make Google understand what our site is offering as the next step, as an action?

Google Panda, Google raters, EWOQ and the death of keyword spamming/stuffing for SEO

A couple of days ago, Barry Schwartz over at www.seroundtable.com wrote about a very interesting manual that has been leaked from Google. Supposedly, this manual deals with the general guidelines of website rating and had been written for the Google search engine quality raters. So this is a lot of new terms to take in and a lot of debatable info has been recently flowing around the internet about these so-called raters, who are supposedly simple people like you and I, who Google had hired to independently rate the search results, or a part of the search results. Very interesting info indeed. I won’t post the link to the manual because I’m not a fan of leaking or divulging information that is for Google eyes only, but let me just tell you that all the data that’s in the Quality Rating Guidelines just brings me to one conclusion: black hat SEO techniques are doomed and I can even safely say that keyword tweaking is also doomed.

Before all of you SEOs out there jump on me saying keyword tweaks still work, let me tell you that there are two sides to that story. There are these nice chaps over at EWOQ that are actually contracted search result raters for Google Quality Eval or Rating. We’re not talking about whitelists here, but rather a human factor to check upon the algorithm generated results. The people who work for Google and are checking the search results are all gathered in this group called EWOQ, and they use proprietary software to screen black hat seo, check site relevance and give a simple quality report, actually just a rating that goes from spam to “vital”, “useful”, “relevant”, “slightly relevant” and “off-topic/useless”. That’s right guys, each and every link (site) is rated from vital to spam, according to how relevant and useful the information is. So before you go tweaking your keywords until you hit the algorithm sweetspot, just keep in mind that EWOQ has it’s eyes on you. Your website might get on the top spot for a moment and fall way back because you have obviously tried to tweak the Google algorithm. So hands off tweaking and of course cloaking/hiding keywords with javascript or css techniques!

At first I thought this was a major gamechanger for SEOs around the world, but after thinking about it, you can only come to one conclusion: if you go black hat, you have to live with the fear that you might fall off the top, getting hit by penalties from the human factor. If you go white hat, there is not much left in terms of tweaking after Google Panda 2.5.2 has ravaged the SEO scene: keyword tweaking has been strongly penalised, you can actually see how your site climbs back a good number of positions after you have deleted a couple of instances of your main keyphrase. This is something I’ve witnessed myself, actually.

So there is this question bugging me, because noone has asked it yet: What do these chaps rate? Do they rate our websites directly too, or are they just some quality raters that rate a small portion of the search results called, henceforth, the Google sandbox. Google takes a small part of it’s results and puts it ina  so-called Sandbox so that algorithm changes can be tested in the sandbox and would not affect the websites all around the world. We all wish it was the latter, of course, because google testing it’s own algo would be good news, but how can we be sure?

Anyways I will tell you that there are certain signs on a website that do tell a story of a website that has something relevant to tell. A website that serves up lyrics, recipes, quotes, proverbs, poems is usually not rated spam. However, if there is a high number of ppc ads or other types of ads on it, it could be reconsidered and badged spam. On the otehr hand, if your thinking about a local business, your business, you might want to verify the authenticity of your business by adding phone numbers, a real address and a working contact form to your pages, because apparently, that means you are legit. I can see the point in this, because if people from one geographical location were to invade another geolocation just because the market there is better, the term local business would become obsolete, everyone would be targeting the best geolocations. Also beware of copying paragraphs worth of text even from sources such as Wikipedia! This kind of information is likely to be found on hundreds of sites, because if you’ve thought about it, others have too.

I won’t start on how I can substantiate this information and I won’t tell you about the links to link quality reporting interfaces I’ve found in the manual. I will also not resort to trying to prove that what I’m telling you is true and you have to change your SEO methods because of it. I will simply change mine and hope that other SEOs give up the remainder of their SEO tweaking techniques in favor of creating very informative, specifics-rich content for the visitors of the site. It’s not about the number of occurrances of the keyphrase anymore, Google knows what you’re talking about from your title. Leave the neccessary signs for it to know what your site is about and write the rest for your audience.

P.S.: If you run into a link like this one: https://www.google.com/evaluation/search/rating/task-edit?task=#*$!#*$! in your web acces log or crawler log, apparently, it is a sign of having been rated by a quality rater from the EWOQ team.

Elements CMS 6.0 Explained Part 2

Elements CMS is our proprietary, integrated small, but very efficient Web Content Management System. This is a follow up on my last article, “Elements CMS 6.0 Explained Part 1″. So how does such a small editing panel fulfill all editing purposes, and how does it compete against a full-fledged website editing backend? Well, first of all, you won’t have to switch around between the frontend and the backend to see what the effect of your editing is. Second, this approach to building a website out of parts is rather modular, compared to simple template-based pages, because you can use any amound and configuration of small templates called (henceforth) elements. They canm be rearranged in second by simple drag & drop and there is also an intuitive hovering div that appears when you hover your mouse pointer over one of these elements, allowing you to: edit the element, move up, move down, move to the top, move to the bottom and delete. Plain and simple, yet effective, because you can rearrange a whole page consisiting of titles, texts, photos, galleries and other elements in a couple of seconds. Compared to having to copy and paste texts in a Rich Text Editor I think this is a step forward. Each element is actually a very small template which has xHTML and CSS behind it, so every element you use will be styled according to your design, unlike the texts in a RTE.

Once finished, your page exists in an inventory of pages, represented by a list in your page editing tab. If you want these pages to be visible to the public, you will have to go over to the Menu editing tab, create some menu items and then assign a page to every menu item. The unlimited number of pages is not a sales pitch, it’s true, because once you have your menu and the assotiated pages, you can create submenus and link to other pages you have created or even add a link element in the page and link to a new page that has no menu item linked to it.

Ok, so that was the basics of Elements CMS. Let’s see a bunch of specification, explained:

- Advanced menu editing – add your menu items, edit the texts that might be associated with them – you know those menu types that have short texts under or near their menu items; add submenu items to the menu item you’ve just created
- Automatic Google xml sitemap generator – Elements will generate a sitemap automatically, including the pages you have checked for this purpose (there might be unfinihed pages you mgiht not want to include in your xml sitemap), and also including the page priority and update frequency you’ve set in the page editing tab
- Newsletter sending – this will send your newsletter to all your subscribers, using a cron job to avoid getting listed as mail a spammer – with Elements CMS you can edit your Newsletter just like any other page of your site. This means you’re not restricted to a simpe RTE, but can use graphical elements to create the Newsletter.
- Language management – you can add your own languages and uplaod small language flags, that is if your site uses small flags to display languages. Every new language means a new, separately treated part of the site that you can edit separately.
- IP Logs and IP ban – you’ve guessed it, this is a system to log any unauthorized intrusion. We have added brute force protection and IP log at the same time, because let’s say somebody had found out your password and wants to do some real damage. How woudl you know who hacked into your site’s administration? Well, Elements CMS logs every login and operation. This is how you will be able to detect any unwanted acces to your website. Ne that a hacker or just an angry co-worker who knows your password, you’ll know soon enough if there had been an intrusion. You will want to ban the intruder’s IP address of course, so you have the chance to do just that, because we have a field with a list of all the unwanted IP addresses.

[to be continued]

Elements CMS 6.0 Explained Part 1

Among the tens of thousands of websites created upon a free CMS like WordPress, Joomla or Drupal, have you ever asked yourself if one could really get differentiated as much as it’s owner would like? And would that difference really ‘make a difference’? Because to my knowledge, WordPress is the best in terms of SEO, but still, it could use some professional tuning to reach the degree of optimisation a custom-built, well optimised site has. My small but productive team of coders and myself have come up with a new approach to a CMS that’s almost a equal to a switch of paradigms and has a plethora of white hat SEO features.

I personally really think that layout-template driven CMSes are a thing of the past. Sure, anyone can fill out a simple template, because it is so easy to see where the title goes, where the text goes and so on, but how about a free hand in the way the template looks, and in the place where you put your title? Some would argue that a RTE (rich text editor) gives you that free hand, but will you also press the tiny html button to add proper SEO tags to your titles and subtitles to help Google understand what it is that you are talking about in your page? No, at least not in most cases. So don’t think in terms of page = template, but rather page = any number of elements, in any order, and in any layout you want. That is the heart of our CMS paradigm, free hand in layout, content, SEO and a helping hand or rather a guide in doing your SEO. Our Elements CMS 6.0 lets you add any kind of element to any of your pages: main title (h1), subtitles (h2), images, texts, lists, quotes, galleries, banners, menus, submenus, video players, audio players, Youtube and Google maps elements. Virtually any kind of visual or functionality-driven element can be added to a website, including complex booking elements or any custom-built tailor-made functionality the customer needs. So the main idea is that you’re using dozens of elements on any page you like, instead of just one template per page. This opens the door for new opportunities in terms of SEO, you can put your titles anywhere in the page, not just in the specific place the template designer had designated for this purpose.

Can you believe that everything related to editing your site happens in this small gadget? Of course, there are tabs, can’t fit everything into one gadget but we have carefully structured all the features on separate tabs like Settings, Page edit, Element edit, Menu edit, Sitemap, Newsletter, Languages, IP logs & ban and Help. You can have virtually any number of main menu items, submenu items and pages, but this depends of course on the number of items your site’s width can accommodate.

WYSIWYG editing, a.k.a. What You See Is What You Get editing. Well, this is new to the html-driven web, but it’s basically about getting exactly what you see and edit. With Elements, you edit your site on-page, without having to open two browser tabs. In our CMS the backend and the frontend have been fused, al the backend fits in that small gadget up there. In case you were wondering, yes all operations go through AJAX, everything is real-time and fast.

[to be continued]

Google Panda 2.5 Update

Google Panda strikes again! Se pare ca Google continua cu lupta anti continut de slaba calitate: se intrevad modificari clare in rezultatele livrate de gigantul motor de cautare. Mai multe site-uri bine stabilite, create de noi, si-au inceput binecunoscuta fluctuatie de pozitii, denumita Google Dance. Nu putem sti daca este doar un “pre-quake” sau Google lanseaza insusi update-ul Panda 2.5, dar ne asteptam de o vreme la acest update, si vor mai fi multe altele in acest an. De unde stim ca se intampla ceva mai important? Urmarim site-uri de nisa, site-uri foarte bine stabilite, care intotdeauna erau in pozitia 1-2 max 3, pentru ani de zile. Daca ele dispar din prima pagina, ok, poate este vorba de niste penalitati. Insa daca vedem o fluctuatie gen pozitia 1-3-5-3, e clar, ca este vorba de ajustari de finete la motoarele de cautare. De obicei site-urile de nisa, cu vechime revin pe pozitia anterioara, si acesta poate insemna mai multe lucruri: 1. Google foloseste asa-numite whitelisturi, adica rezultatele de top sunt ajustate de o mana umana care readuce fruntasii in locurile fruntase 2. Google si-a dat seama ca acea ajustare testata in direct nu si-a servit bine scopul cum se asteptau baietii. Pot sa va spun ca suntem martorii unor astfel de upgrade-uri inca din martie, si au avut loc vreo 7. Ceea ce ne ingrijoreaza pe noi, firmele si agentiile de web design, este ca din cand in cand apare un site nou-venit, pe prima pagina a cautarilor specifice, si acel site nu contine deloc informatii. Site-uri fara adresa de contact, fara numar de telefon, fara lista de servicii, fara portofoliu, scrise in cateva minute, campeaza cateodata pe paginile 1-2-3 ale rezultatelor pe cautarile cele mai importante. Bineinteles ca acest lucru ne duce cu gandul la efectele unui algoritm, care nu pot fi controlate pe deplin, mai ales fiind vorba de miliarde de rezultate, chiar mai multe rezultate pentru o singura locatie. Candva, la sfarsitul acestei avalanse de update-uri, probabil ca vor “slefui peretele” baietii, si incoerentele vor disparea, speram. Pentru ca daca ne gandim logic, daca un update are un nume, trebuie sa aiba si un sfarsit, cand urmeaza un alt update cu un alt nume, sau poate o perioada de liniste.

Reguli CSS refolosibile

Pe un site web exista de obicei maxim 2-3 tipuri de fonturi, in caz contrar am putea vorbi de un stil incoerent de tipografie. 2-3 stiluri CSS n-ar fi rau, si ar rezulta intr-un fisier CSS ultramic, dar de obicei folosim diferite variante ale aceluiasi stil, ceea ce presupune de obicei reguli CSS diferite pentru fiecare. Putem crea reguli generale CSS pentru aceste tipuri de fonturi, astfel incat de cate ori avem nevoie de cateva proprietati des folosite pentru textul prezentat, facem referire la stilul generic pentru a nu repeta 5-6 randuri de proprietati la infinit.

Astfel, putem avea o regula generala:

.museo_generic {
display: inline;
float: left;
font-family: MuseoSans, arial, sans-serif;
font-size: 12px;
color: #FFF;
}

la care putem adauga fiecare sub-stil care are nevoie de aceste proprietati:

.museo_generic, .heading1, .heading2, .heading3, .lettertitle {
display: inline;
float: left;
font-family: MuseoSans, arial, sans-serif;
font-size: 12px;
color: #FFF;
}

Astfel, am evitat sa rescriem de fiecare data aceleasi proprietati. Poate ca pare o chestie banala, dar poate micsora cu mult marimea fisierului CSS, si implicit, reduce load-time-ul, ceea ce conteaza pentru SEO. Diferentele de proprietati se pot apoi defini in reguli CSS separate, dar in care nu repetam proprietatile comune.

Aceste reguli se pot folosi pentru orice grup de proprietati care se vor refolosi in mai multe reguli CSS, pentru a optimiza fisierul de stiluri.

Joomla, Drupal, WordPress vs custom-built CMS

Internetul abunda in site-uri create in cateva ore, pe CMS-uri gen Joomla, Drupal, WordPress sau alte CMS-uri gratuite, customizabile in cateva ore sau zile. Intrebarea este: cat de mult merita sa ne bazam pe ele, daca este vorba de un site al firmei, care nu trebuie sa serveasca doar scopuri de prezentare, ci trebuie sa fie “the one”, tehnologia perfecta care sa si aduca site-ul in topul motoarelor de cautare, prin structura si optimizarea lor. Exista pareri diverse, si contradictorii despre tehnologiile gratuite, de aceea am dori sa ne exprimam o parerea BIAS-ata, personala. CMS-urile gratuite sunt niste tehnologii de obicei foarte bine testate, cu o functionalitate buna, fara durerile de cap date de erorile unui site creat de la 0. Ele mai au si un grad de optimizare ridicat, pentru ca sunt dezvoltate in continuu si programatorii care lucreaza la ele au grija ca ele sa aiba un set minim de optiuni, module si setari SEO. Dar ce se intampla cand este vorba despre “the long run”? Adica doresc sa-mi construiesc un site pentru firma, merg pe varianta mai ieftina si rapida, Jommla, executie 2 zile, site fainut, sau mai bine merg pe mana firmei de web design, creand un site de la 0, spunandu-le exact ce am nevoie, si primind ceva customizat exact pe nevoile mele. Depinde de firma. Personal, ne place sa construim site-uri de la 0, dar si noi avem module care le integram in fiecare site. Doar ca acele module le lucram exact dupa nevoile noului client. Pe deasupra, ne ocupam de unele aspecte care sunt intr-un fel neglijate la CMS-uri gratuite: SEO. Un site bazat pe Joomla, Drupal, WP poate avea un grad inalt de optimizare, intr-adevar, dar gandindu-ne pur logic, putem spune ca “Un site bazat pe CMS gratuit poate avea un SEO maximum atat de bun, ca cel mai bine optimizat site bazat pe acel CMS gratuit”. Am dat deci de limitarile softului. Acestea sunt limitarile de care ne debarasam, creand site-uri gandite cu optimizare “in mind”, adaugand automatisme SEO in cazul in care e vorba de sute sau mii de pagini. Apoi ne gandim si la modalitatea controlarii acestui SEO, pentru ca nu e de ajuns sa permitem sa se poata edita toate meta-urile si setarile SEO pentru fiecare pagina a site-ului: cream formule de generare a meta tag-urilor SEO, astfel ca editand o singura formula, modificam poate 5000 de titluri. Finetea sau automatizarea acestor setari este cea care lipseste din CMS-urile de larga folosinta, si tocmai cu acest lucru incercam sa facem noi diferenta.

Google PageRank Update 4 August 2011

Si iata ca revin sa va anunt despre noul PageRank Update din ziua de ieri, 4 August 2011. Am scris intr-un articol anterior ca banuiesc ca Google va scoate update-uri ale liniutei verzi mult mai des de-acum incolo. Mai exact baietii de pe webmastertalk.com au abordat acest subiect, si se pare ca este ceva sambure de adevar in el. 3 update-uri de PageRank in decursul a 60 de zile sau mai putin reprezinta un efort enorm, daca nu s-a schimbat chiar algoritmul de afisare a liniutei verzi. Personal am observat cresteri destul de mari in randul site-urilor pe care le-am creat recent, de exemplu www.cazarebusteni.ro a crescut la PR4 (de la 0). Suntem mandri si de faptul ca acest blog a primit PR 3, dar am observat un “pattern” de distribuire a PR-ului dupa acest update: Pagina principala primeste PageRank x, si toate paginile care au link direct de pe prima pagina primesc PR x-1, deci au acelasi PageRank.

Pe de alta parte, am observat scaderi de PR in randul unor siteuri despre care stiam ca primisera linkuri platite (nu avem tangenta cu ele, nu noi le dezvoltam). Acest lucru l-am semnalat deja in posturi anterioare: Google a declarat razboi continutului de slaba calitate si linkurilor platite. Dar cum recunoaste continutul de calitate si linkurile cumparate? Cel dintai sa zicem ca s-ar realiza prin algoritmi foarte avansati (sau chiar si o mana umana), dar linkurile? Poate ca relevanta domeniu care ofera link —> domeniu care primeste link, si de fapt calculul relevantei tuturor legaturilor dinspre un site determina daca acel site a oferit linkuri cumparate unor site-uri care n-au nimic in legature cu el…

Ramane de vazut, dar ne asteptam la alte PR update-uri, si inca vreo 500 de tweak-uri la algoritmii de cautare.

Am lansat magazinul virtual www.americanbrands.ro – Ed Hardy, Christian Audigier, Affliction, Sinful, Remetee, California Style

Am lansat cel mai recent proiect al nostru, http://www.americanbrands.ro/webshop/edhardy. Este un magazin virtual bazat pe platforma eCommerce IntelliMall dezvoltata de Echipa Marconi Media, care face posibila uploadarea unei intregi baze de date de produse dintr-un fisier Excel. Webshopul preia produsele (teoretic orice numar de produse) dintr-un fisier excel formatat dupa niste reguli concrete, si le incarca in baza de date. American Brands webshop deserveste o categorie speciala de brand-uri foarte cunoscuta in Statele Unite: Ed Hardy, Christian Audigier, Affliction, Sinful, Remetee si California Style, tocmai din acest motiv am creat o identitate vizuala pe background intunecat, cu contraste spre auriu/argintiu. Nu este un magazin online extrem de complex, pentru ca am observat ca aplicatiile complexe produc un Bounce Rate (rata de parasire a site-ului) usor ridicat. Am folosit un cos de cumparaturi minimal, care doar afiseaza numarul de produse din el, astfel facand loc mai mare produselor din webshop. Am renuntat la filtrarile si sortarile ultracomplicate, abordand un listview care pune in prim plan pozele din site. Fiind vorba de un magazin online de haine, cele mai multe fotografii contin doua poze cu fata si spatele hainei, dar aceste poze nu sunt uploadate ca poze separate, ci site-ul face un singur fisier jpeg dintr-un fisier fata,jpg si unul denumit spate.jpg (doar un ecemplu).  Putem spune ca este onoarea noastra ca am creat un magazin virtual care deserveste creatiile unor designeri de vestimentatie recunoscuti in toata lumea.

FaceLift pentru MarconiMedia

Dupa in jur de un an de la lansarea site-ului nostru full xHTML + CSS – inainte era full-flash – a sosit timpul pentru un facelift pentru pagina de start: http://www.marconimedia.ro. Lasam la latitudinea Clientilor sa decida daca este o schimbare in bine sau nu.