How to reverse engineer Google algorithms

Please excuse the trick title - but the answer is - it is both impossible and a waste of time and money to try to reverse engineer Google algorithms-or any search engine algorithms, for that matter.

When, like me, you have distilled the instructions of more search engine optimisation (SEO) consultants than I dare to count, they typically suggest in one way or another that you try to do just that – reverse engineer algorithms. Following such SEO instructions may appear to help 50 per cent of web sites, and leave the other half at the bottom of the heap, or more accurately, the curve.

It's impossible to reverse engineer search algorithms mainly because the worldwide web is a constantly changing entity. Every time one goes into the search bar seeking out something and landing somewhere, staying there, and going somewhere else, etc, that activity adds to the growing database that 'web crawlers' mine for insights on how people think and search.

These web crawlers are trying to get smarter at natural language processing, a sibling of computational linguistics, and they do get smarter, every day - and not just when Google or another search engine company makes major announcements on things like the Panda algorithm.

Perhaps in the real world these web crawler ''smarts'' are increasingly dumb and blind to global business-to-business (B2B) companies' online marketing efforts, but that's another story.

If your company is presented with a static list of target ''phrase depth'' or ''title counts'' or other on-page or off-page factors as a to-do list for search engine optimisation that are purported to affect every site on the web the same way, you need to know that this list is pure fiction.

First, it is fiction because the web is never static. This means that if there was some magic formula for website optimisation yesterday it won't be making the same fit with the web today because today's web is different from yesterday's and tomorrow's web will be different again.

Second, a static one-size-fits-all SEO to-do list is fiction because every page and every site is in a unique competitive landscape on the web. What works for Tata Steel in terms of website optimisation is very different from what will work for Bharti Airtel because they reside in a unique landscape on the web.

Third, even in the very small corner of the web that a 'giant' like Tata Steel occupies, the competitive landscape of the web is in a constant flux. And because so many factors are at play - eg a recent report by Google said they considered 200 factors - that affect how one or another competitor for a certain keyword shows up on any particular person's search, the competitor rankings are not linear.

Consider for example the competitive landscape for the keyphrase ''data logger'' shown in Figure 1. Currently Wikipedia owns the top spot. (Note: this graph has figured in personalised search algorithms and in this case has limited the geography to the United States to give a picture of ''ranking'', albeit ''site ranking'' per se is no longer quite an accurate term due to how personalised search algorithms and localised search algorithms shuffle the deck for each person's search every time - one of the least understood and recognized impacts affecting most online marketing efforts today.)

Examining this bar graph, you can see that the site that is currently in second position today is not the one best poised to overtake Wikipedia for this top spot. Rather, a site that today is in the 22nd spot is very close to moving to the number one slot.

This is because if you x-ray the factors comprising the competitive landscape and the particular weighting that each factor has on search engine optimisation in that unique competitive landscape for the keyphrase ''data logger'' there is some major factor or few factors that the No 22 web page has that make it nearly head-to-head with the leader.

If No 22 is really lucky - and assuming they have the math tools to dissect this competitive landscape - they may find that all they need to do is beef up one or another on-page factor, rarely more than an afternoon's work for an experienced SEO copywriter.

This is so even though today that No 22 web page wouldn't show up on the first pages of the average ''Googler'' for that term. In reality, today that No 22 web page would largely be ignored and could just as well be No 92, because we web searchers are an impatient lot who rarely make it to page three.

Fourth (and perhaps most germane to why the static lists of SEO to-do tasks seem to work about half of the time), on the web and in the corner of the web where your web page/s compete you are always graded on a curve. How ''good'' does your site need to be? It only needs to be better than the other pages and sites you are competing against.

Do inbound links matter? In reality, to some web pages they matter not a whit. To others, all other SEO factors pale in comparison to the number of inbound links, the ''authority'' of these inbound links, the title of these links, etc.

Competitive Web Intelligence Report on Keyphrase 'data logger' courtesy of:  http://www.globalb2bcommunications.com/seo-tools/seo-recon/

So why do SEO consultants give you the same set of instructions that they give to all their other clients? This rarely is because the person giving you this list is trying to trick you. Rather, it's just a reflection of how out-of-date and out-of-reality many ubiquitous SEO notions are.

Take a look at the websites of some of the biggest companies in India (or any country) and what you find may surprise you, if you are up-to-date on how search engines really work.

As an example, consider ''keyword metatags''. Pick say 10 companies for a start. Using the Mozilla Firefox menu ''view'' option and selecting ''page source'' you can take a quick look at the code for the web page you are visiting. Take a look at the homepages of the 10 companies you selected. Then, see what they list for ''keyword metatags''.

Then, read up on the announcement that Google made in 2009 indicating that keyword metatags did not affect search - http://www.mattcutts.com/blog/keywords-meta-tag-in-web-search/.

Since keyword metatags do not count towards making your site visible on search engines but they do supply competitors with quick information about your website optimisation efforts, perhaps the wiser course may be to just take keyword metatags off your site.

In fact, your page may actually be downgraded if / when you include lists of keywords that do not have anything to do with the verbiage on that page.  In my experience today, the 10 pages that I pick at random are likely to have a majority that make this keyword metatag mistake.  This is a good example of how out-of-date SEO practices are.

Competitive intelligence on the site's unique competitive landscape on the web is what matters. You need to have the math tools that are capable of weighing the relative importance of one or another factor affecting your unique competitive position in the unique portion of the web where you compete. Today's No 1 site can be tomorrow's No 22, or vice versa.

The bottom line is: don't try to be smarter than Google, or Bing, or Baidu, or Yahoo. If you waste your time and money trying to reverse engineer search engine algorithms, the bigger cost is that you are not paying attention to competitive web intelligence - the maker and breaker of internet marketing success.

Amy Munice is president of global PR firm Global B2B Communications /ALM Communications Inc, that guarantees the numbers of published feature, columns, blogs, and technical white papers and fully integrates state-of-the-art math-based search engine optimisation tools into development and execution of all lead generation strategies for its clientele