Text and search engines

Google_Hummingbird_Logo[1]

After Google introduced their new Hummingbird algorithm in August/September 2013, writing for search-engines has become both much harder and much simpler.


What is Hummingbird?

Where older versions of Google were refined, improved and adjusted from time to time, Hummingbird represents a complete rewrite of the Google search engine, and a drastic change in philosophy. Older versions of Google focused on matching keywords on a site with keywords in a search phrase, so content makers had a lot to gain from trying to match the highest number of keywords possible. Some of the less scrupulous of them resorted to including an artificially big (and often not very relevant) list of keywords on their site, often hidden from the user but not from Google. This practice was risky, as Google would punish such practices, but could also potentially be quite rewarding if not detected by Google. This is, thankfully, no longer the case.

Google Hummingbird, by contrast, does not focus on keywords, but how words and concepts are connected to each other. A bit simplified:

Older versions of Google looked at the dots.
Now, thanks to Hummingbird, Google also looks at the lines connecting them.

This means that Google Hummingbird performs a semantic search, focusing on the user’s intent instead of the letter of their search queries. Based on how the search is worded, Hummingbird will know the difference between a search for the Paris Hilton hotel, and the blonde socialite of the same name.

Google Hummingbird is also capable of understanding synonyms, so users need not use the exact word they are searching for but still receive the correct search results. This also makes writing content more flexible and dynamic, and allows for better language without sacrificing findability.

It also seems likely that Google Hummingbird is based on a limited form of artificial intelligence, ie it can learn. Search phrases and results are continuously analysed and refined, in order to make later searches perform better.

This means that certain types of content will do better than before, and other types will not do as well as before. For instance:

Google rewards content that:

  • Answers a question.
  • Is old and «evergreen», ie doesn’t need updating often but is still of high value over time.
  • Receives a lot of traffic from other sites, ie is linked to and considered valuable by others.
  • Is useful and imparts information.
  • Is written for a genuine reader.
  • Is shared on social platforms.
  • Is long and coherent.
  • Contains in-depth material.

Google penalises content that:

  • Displays «keyword-stuffing», ie has an artificially high number of keywords.
  • Is old and out of date.
  • Is not properly worded or written in complete sentences.
  • Is «empty» of useful information.
  • Is written for a search engine and not for the reader.
  • Is not useful to a genuine reader.

How to write for Hummingbid

The single best thing to do to optimise your content for Google is to keep making quality content.

«The Hummingbird approach should be inspirational to anyone managing and planning content – if you aren’t already thinking like Hummingbird, you should be. In a nutshell, think about why people are looking for something rather than what they are looking for. A content strategy should be designed to answer their needs, not just provide them with facts.» – Steve Masters

As long as you make sure you are writing for the reader, you’re good, as you then also automatically write for Google.

With Hummingbird, Google seeks to create a search engine that mimics the needs of a genuine reader.

There are, however, a few things you should keep in mind:

  • Relate the content to its author. Google may very well use this connection to assess quality.
  • Make sure your content is shared. Getting people to vouch for the quality of the content by sharing it or calling attention to it, is the surest way for Google to assess it as quality content.
  • Know what topics your readers are looking for information on, and provide it, both broadly and in-depth.
  • Keep in mind what your readers will want to know, and how they will ask for it, instead of what you want them to know and how you want to present it.
  • Make content that keeps the reader on the page. It’s fine if the reader goes on to other pages, but make the content attractive to the reader, in order to make him/her stay on the page as long as possible. Google uses these statistics to assess quality.
  • Anticipate the readers’ questions, and answer them clearly in your content.
  • Try to limit yourself to one issue or question answered per page. You will not gain extra cred with Google for answering many questions on the same page, so the traditional FAQ is not a useful means to optimise traffic.
  • Top tips. This format naturally lends itself to Q&A matching. Rather than just saying what to avoid, focus the title on the real reason(s) to adopt or avoid the suggested actions.
  • Don’t be afraid of long texts that go in-depth, as long as the readability is good.
  • Forget keywords – think key phrases instead. Focus on concepts instead of keywords.
  • Write how you would search.
  • Keep your content unique. Don’t repeat the same content on different pages or sites, and if you have to, make sure you rewrite the text and word it differently.
  • Lastly, and most importantly:

«Do not compromise the quality of the text or the user experience for rankings. Ever.» – Lindsay Massie


In the future

Google has announced that they are planning to introduce a reward for sites that contain factual content. This is not yet implemented, but it will certainly be in the not too far future. Google has not gone into detail as to how and how much, but they have announced that what is factual and what is not will be evaluated by a form of calculation of consensus among websites that are deemed trustworthy. Hence, it may pay to have true and useful facts, and content based on this, on your site. Even better, a site like visitnorway could conceivably be positioned as one of the trustworthy sites which all others are compared and evaluated against.