On search engine rankings and optimization (SEO)

I regularly hear from my clients that they’ve gotten email advising them that their websites ‘have dropped in Google’s rankings’ and that ‘they need to take immediate action.’

  1. That email is spam. Delete it with the respect it deserves. Identical messages are sent by the million, preying on fear and uncertainty in the hopes of making a quick buck.
  2. Search engine rankings are not simple. Anyone who says otherwise is selling something (see point #1, above).
  3. Rankings change. For years, Tantramar Interactive Inc. ranked at the top of results for the word “Tantramar” but, as a geographical place-name, that was never going to last. Having registered tantramar.com early (1997!) helped for a while, but many others use the word, too, including the high school. Besides, there’s nothing about “tantramar” that inherently means “web design”, so of course other sites have edged it out. As of this writing, though, Tantramar Interactive Inc. holds the top Google results for “tantramar web design”, and that’s fine.

Let’s dig into this.

How search engine rankings used to work

Even before Google launched in 1998, the idea of there being a single, monolithic index of the internet had taken hold. People think it’s like a phone book:

  1. You need to know something
  2. You Google it.
  3. You get the same result as everybody else.

Right? Wrong. Things have changed considerably since then, and they’re not done changing.

Naturally, Google doesn’t share its algorithm — the secret recipe that decides how websites get indexed, ranked, and suggested. The quality of their recommendations is a competitive advantage and is constantly fine-tuned. If it didn’t change, it would soon be reverse-engineered, allowing competitors to copy it and publishers to game it. That’s what so-called search-engine optimization (SEO) aims to do, through various techniques.

How search engine rankings work

We know that Google’s search algorithm takes many factors into consideration. The results anyone gets for any particular keyword search depends on all of these factors.

Some factors pertain to the person doing the search; others are related to the sites being ranked.

Factors related to the user

  • Account history (including search history) — based on what Google knows about someone, it can decide to show different results:
    • Google account activity informs search results. In the same way that you might see ads for golf resorts in Gmail alongside messages about an upcoming vacation to St. Andrews, search results can reflect anything Google knows about a person and their activities or interests. What have they watched, favourited, or shared on YouTube? Which Blogger-powerd websites have they visited? What sites using Google Analytics have been visited? (Services using Google accounts include YouTube; Gmail; Android devices using any Google services, Google Calendar, Google Contacts, Chrome on macOS, iOS, or Windows; Google Photos, Google Docs; Blogger; Google Drive; Hangouts; Google Maps; Google+, etc.)
    • Local browser cookies (data about past searches) can still identify past Google search activity or visits to websites running Google-powered ads — even if a person has never logged into a Google account.
    • Browser extensions, privacy settings, and the browser itself can all affect search results.
  • IP address and physical location — a device’s IP address can be used to merge data from multiple browsing sessions, helping them build a profile, even if users are not logged-in to any Google services. This IP address, which is assigned by internet providers, can also typically pinpoint geographical location, often as accurately as a postal code. Google can then use location to prioritize or filter results. A user in Oxford, Nova Scotia, searching for “pizza oxford” wouldn’t be shown results related to Oxford, England. This can also provide a detailed demographic profile, based on what they know about people who live nearby (will they recommend take-out or 5-star Italian restaurant?).
  • time of day & day of week — A search for hardware stores may prioritize stores that are open on the day or time-of-day the user is searching.
  • type of device — people searching from a mobile device may get different results than people searching on desktop or laptop computers. Phones may also provide detailed GPS data, which can reveal whether the user is stationary, walking, driving, or even riding a train — or may be heading to an appointment (with address) that’s listed in their Google Calendar — all of which can inform the algorithm as it decides what results to send.
  • current/recent trends — what’s going on in the world? This is part of the context that helps determine what people are looking for.

You get the idea. Everyone can get different results.

You are not your customers

Setting aside for a moment the issue of how Google itself decides what to show each user of its search engine, there is a bigger issue.

When people worry that their site has slipped in the search engine rankings, the first thing they do is type something into Google to see for themselves.

The problem is that they don’t typically type the same things into search engines as their customers do. Typing something different is not a valid test. The priorities, motivations, and context are all different. It’s not even that they’re going to get artificially low results: if they type their business name in (would their customers do that?), they may get artificially inflated results.

What this sort of test can reliably confirm is that the site is in Google’s index.

Factors related to the site

The website itself is where its owners/publishers have some control. There are many factors, and how they’re weighted is unknown, but these factors include:

  • Website content that’s relevant.
  • Website content that’s regularly updated.
  • Site speed — all other things being equal, a slow site is deemed less user-friendly and will rank lower. This could be network speed (sometimes half-jokingly called internet weather), large file sizes (think inadequately-compressed images or PDFs), too many dependencies on 3rd-party scripts (e.g. a site makes too many external calls to resources from Twitter, Facebook, ad-servers, etc., and has to wait for them), overloaded servers, etc.
  • Mobile-friendliness — as both an advertising platform and the publisher of the Android mobile operating system, Google has, since 2015, explicitly down-ranked sites that they consider mobile-hostile (See Think your site’s mobile-friendly? Does Google? for more on this), at least for people searching from their phones — and that’s most traffic these days.
  • Domain age — an older, established, and therefore more-credible domain is a signal of quality.
  • Incoming 3rd-party links — if others are willing to link to a site, that’s taken as a vote for it. This was the key insight that led to Google’s early success in the late 1990s. (Link-farms, though, backfire.)

Bottom line: getting a higher ranking

There is no magic bullet

No one can guarantee rankings. Even if they could, they wouldn’t last.

Google itself takes the position that a well-crafted website that follows industry-standard code practices, pays attention to accessibility guidelines (Google itself can be thought of as a user with common accessibility requirements), and is written to be useful to its human visitors will rank just fine.

Review any analytics you have to see what keywords people have used to find your site, but understand that such reports are always open to broad interpretation and have a huge inherent blind spot: they don’t know anything about keywords people have searched for that didn’t bring them to your site. (How could they?) Those might turn out to be crucial.

Understand what your customers are looking for and the language they use; prioritize what matters to them.

Thoughtfully-prepared content, written for real people, and produced by someone who takes the time to format things properly and accessibly is key.

That’s how Tantramar Interactive Inc. produces websites.