Monthly Archives: May 2016

Keyword Research

All SEO begins with keyword research and for the most part centers around Google’s keyword research tool, the Google Keyword Planner. In this tool Google will tell us the average number of searches any keyword receives each month, in our country of choice. This is amazing data to have access to!

Step one of our keyword research process is to write down a list of all of the keywords you think might be relevant to your site. Each keyword is a phrase that someone might enter into Google when searching for something for which you want your site to rank.

When you’re making this list, be sure to include every variant you can think of: include singular and plural versions, include common features, etc. For example, if I wanted to do keyword research on headboards, my starting list might look something like this:

  • headboards
  • headboard
  • head boards
  • head board
  • modern headboards
  • modern headboard
  • traditional headboards
  • traditional headboard
  • wooden headboards
  • wooden headboard
  • wood headboards
  • wood headboard
  • bed headboards
  • etc.

There are several tools out there that can help you expand your list (which we’ll talk about in Other Keyword Research Tools), but for now let’s just begin by using your brain.

You can make your list as long as you want — by including all the variations you’ll be learning how people typically search for the product, service, or information that your site offers. You’d be amazed at how often the language customers use is different from what business owners think they should be using.

A couple of examples: I worked on a large site that insisted on using the term “mobile phone” throughout their site. About 10 seconds of keyword research showed that in the US, “cell phone” was massively more popular. In fact, I calculated that the site would make about an extra $5 million per year just by switching from “mobile phone” to “cell phone” throughout their site. Another site I worked on had a great selection of accent pillows for couches. Some quick research showed that the phrase “decorative pillows” had six times the search volume. The point here is that the language used by the majority of potential customers is not always what you, or your bosses, think it is.

Remember, there’s no cost to anything in the keyword research stage, so feel free to include the most obscure variants you want. Cover all of the bases and save yourself regrets down the road.

Do You Deserve to Rank? The Fundamental SEO Question

Before we get into the SEO tactics of what to do and how to do it, let’s have a quick reality check. Set your emotional attachment to your site and business aside and prepare to be dispassionately honest with yourself — or better yet, get someone without any skin in the game to answer this question.

Does your site really deserve to rank?

Everyone answers yes to this – “of course my business deserves to rank!”  But step out of your own shoes for a moment and look at it from Google’s perspective. Is your site really the best result on the entire internet? Because that’s what Google is trying to find.

Sure, the user is searching for artisan snowflakes and you sell artisan snowflakes, but that’s not enough to put you ahead of everyone else. Is your site better than other sites offering the same products/information? Does it look better? Is it easier to navigate? Are your prices/services/selection/ease of use/customer service/online information at least as good as your competition?

SEO is really a combination of technical tactics and marketing. And it doesn’t matter how much you market shit, it’s still shit.

As a SEO, this can be a very difficult question to wrestle with, because everyone thinks they deserve to rank.

Example: Best Buy Reviews

I remember when I was at BestBuy.com, I was asked why we didn’t rank for top reviews searches — many of which had substantial search volume. BestBuy.com has user reviews after all, and my colleagues wanted to know what we should do to get those user reviews to rank for “computer reviews” or “appliance reviews.”

But the problem is we didn’t deserve to rank for those searches because a page of user reviews isn’t what the searcher wants to see. Someone searching for “best laptop reviews” wants to see an expert editorial opinion giving them informed advice. Not a list of 500 random people on the internet giving their personal opinions.

Ultimately, I argued, if we did start ranking number 1 for all these searches, Google would have to change their algorithm, because they would be ranking a bad result. Cnet and Tom’s Hardware and other experts who are really doing the research and comparing the products deserve those spots. I suggested that if we wanted to have the Geek Squad go out and start reviewing and recommending hardware, we could work to make that rank, but spending time to rank user reviews was a poor use of our time.

Example: Lumber Liquidators vs Home Depot

Both Lumber Liquidators and Home Depot sell hardwood flooring (and vinyl and laminate and all the rest). Both are huge, expansive sites. So which is the best result? Here are some things to consider:

  • Their prices are more or less comparable
  • Lumber Liquidators has a larger selection
  • Lumber Liquidators is only about flooring, and isn’t cluttered with patio furniture and bathroom sinks, which is a plus to flooring shoppers
  • On the other hand Lumber Liquidators has a magnificently ugly site, that looks like someone wiped their backside with a used car newspaper ad
  • Home Depot also sells every tool and accessory you would need to install your flooring, all at better prices than just about anywhere else. It’s a one-stop shop, making it a superior result in that regard.
  • Home Depot also has superior in-depth how-to articles and videos on everything you’d need to know to install that flooring. You don’t need to go to a separate site to learn what you need to know.
Lumber Liquitators vs Home Depot site appearance

Which site would you want to shop at?

So who ranks? Well, Home Depot does, and probably deserves to. However, with enough links Lumber Liquidators could probably overcome their User Metric failures. But if I was running SEO for Lumber Liquidators, top of my priority list would be dealing with that horrible site: surely there must be a better way to present the site while still staying on brand (as far as I can tell from the site, their brand is “in your face cheap cheap cheap buy now and quality isn’t what you’re here for right?”)

In the modern internet world, a site that hideous just doesn’t deserve the #1 ranking in Google.

An Ecommerce Site Test of Deserving to Rank

Here’s another way of looking at it — though this is only really applicable for ecommerce sites: an idea I once had to promote my little ecommerce site AwesomeDice.com was to approach the 10 leading gaming bloggers and ask them to review dice sites: not just mine, but all the competition too. I’d give them the money to buy a cool set of dice from every online dice store out there and ask them to make a post about the experience, rating each site on selection, usability, price, how long it took the dice to arrive and anything else they could think of. It’d be great exposure, and since I had no links getting those 10 links to my site and all the competitors would still land me ahead of the game.

Ultimately I decided not to do it because I was worried that my brand new site wouldn’t hold up well against the others in the eyes of a critical blogger — particularly my product selection at the time. Then it hit me that this was the problem: if you’re afraid to do this for your site, then that’s probably a good sign that your site doesn’t deserve to outrank the competition. It then became clear that my first task wasn’t SEO, but was instead to get my product selection up to par.

Chase What Google’s Chasing: the Best Sites

“I’m sick of following my dreams. I’m just going to ask them where they’re going, and hook up with them later.” – Mitch Hedberg

Part of this notion of deserving to rank is tied closely with something Matt Cutts, Distinguished Engineer at Google once said: “Don’t chase after Google’s algorithm. Chase after your best interpretation of what users want, because that’s what Google’s chasing after.”

The idea here is not to obsess over every minor ranking signal tweak to the algorithm, but instead to build the kind of site that Google is trying to rank at the top of their results. If you do that, every algorithm change Google makes will only help you. I can tell you from experience, it’s a pleasant thing to sit back and let Google algorithm updates boost your rankings year after year.

Alas, building a great site that deserves to rank isn’t enough, as we’ve seen over and over again. You still need SEO (or a large ad budget) to get eyeballs on your site in the first place. After all, no one can recommend your site if they haven’t ever seen it.

SEO From Zero Example

When I first launched AwesomeDice.com I was starting with a brand new domain with no links at all, and naturally the site didn’t rank for any keywords — despite having solid on-page optimization and site structure. I spent a lot of time in that first year working to build links.

Awesome Dice site home page

After the first year I was ranking okay for most of the keywords I was targeting: not great, but I was finally getting organic traffic. But then I pretty much stopped working on building authority — I was busy with other stuff, like shipping orders every night!

Yet over the next year AwesomeDice.com continued to slowly improve in rankings until it was ranking number 1 or 2 for nearly every major keyword I tracked. This happened because people continued linking to Awesome Dice even though I stopped working at it: they linked on forums when they talked about some cool dice they just got, or they linked on blog posts when they wanted to share dice pictures, or they linked to the content I created. Because I had built a site that was better for shopping for dice than the competition, and because I had enough rankings that some people were able to find the site, I steadily earned more links and better rankings.

The worse your site is compared to every other site on the internet that could fill the same user need, the harder it is going to be to get it to rank.

But delightfully the opposite is also true: the better your site is the easier it will be to rank it. Because a great site naturally triggers all those signals that cause Google’s algorithm to rank a site well.

Certainly Google isn’t perfect and doesn’t always rank the best sites and subpar or truly crappy sites make it into the rankings. But if you pay attention, they don’t usually stay there long.

So you could even say that step #1 in SEO is: build a site that deserves to rank #1. Only then do you start on the work to rank where you deserve.

Mythical Ranking Factors

In addition to the factors that influence your ability to rank in search engines, there are some factors that are either old and retired, or just plain myth. Yet these are myths that get reposted again and again in the echo chamber that is the SEO blogosphere, so it’s worth taking a moment to dispel them.

If you are coming into SEO with no preconceptions and no bad habits, it’s probably still worth glancing over these because I promise you, if you start talking SEO with people, you will quickly find someone whose only knowledge of SEO is that you need to put keywords in the Keywords Meta Tag.

  • Keywords Meta Tag: this is meta tag that goes in the header of your webpage code. It is invisible to users and is supposed to tell search engines what your page is about. As you might imagine, a ranking signal that you controlled and that users couldn’t see was very quickly abused. So much so that this is not a factor at all, and hasn’t been for a decade. Tests have proved there’s no ranking benefit to the tag, and both Google and Bing have publicly acknowledged that the keywords meta tag has no ranking benefit. You can read the Doctor’s detailed post on the keywords meta tag if you really want more history and detail.
  • Keyword Density: For a long time SEOs thought there was some magic ratio to word count and keyword repetition. Like a 12% keyword density (meaning 12% of the words are your keyword) is ideal: anything more will look spammy, and less is sub-optimal for ranking. This is not true, at all. It is true that using your keyword twice is better than once, but it’s not necessarily true that using it 20 times is better than 10, and using it too much is a spam signal. Just don’t worry about keyword density: it’s not a real thing.
  • Domain Age: While it’s true that at one time the age of a domain had some influence in rankings (particularly for very new domains), this is no longer a real thing. The age of your domain won’t make your rank better or worse; however, an old domain is certainly more likely to have links pointing at it, and those links could help the domain rank better. Of course if that old domain has spammy links pointing at it, it could rank worse as a result.
  • Adwords Advertising: some novice SEOs persist in believing that Google gives a benefit to companies who advertise with them. This is not true in any way. I have worked on sites that spent millions per week in Google ads, and not only did they not get any ranking benefit, but they were 100% incapable of getting any SEO-side assistance to problems no matter how hard they leaned on their Google ad reps. The wall of separation between the search and ad sides of Google is real.
  • Social Media: You don’t have to go far down the twisting corridors of the web before you hear someone telling you that social media is the future of marketing and the future of SEO. Many will talk about social signals in Google rankings. For all practical purposes, this is mythical bunk. Links from social media are nofollow links. Google can’t even see many Facebook links, let alone understand connections and engagement in Facebook. Most importantly, social media signals are far, far easier to game than links.The waters are dirtied by the fact that most correlation research finds that there is a positive correlation between pages that rank well and social media shares — often a very strong correlation. Alas this is where we have to remind people that correlation does not imply causation. Google has even stepped up to confirm that these are not ranking signals. It just turns out that pages that are really good and get lots of traffic by ranking well in Google are also likely to get a lot of social shares — which makes sense, since they’re getting a lot more eyeballs on the page from their search engine traffic!I want to stress here that social media can be a beneficial and influential traffic channel, and that a strong social media presence can also bolster your SEO tactics by spreading the word of your good works and spreading brand awareness (that can translate into branded searches). But social shares and likes themselves are not a ranking signal.

Kinda Sort of Ranking Factors

In addition to the three main categories of ranking factors, there are also a handful of “kinda sort of” ranking factors. These are typically things that can, maybe, a little bit, help your ranking in certain situations. Or sometimes things that Google has stated are ranking factors but there is little to no evidence that there actually is.

It’s worth being aware of these: maybe your site will fall into a category where this actually matters. Almost certainly you will come across posts claiming with confidence that these are ranking factors and you should implement them at once or else you’re a SEO fool — these latter will often offer to help for a fee.

Kinda Sort of Ranking Factors

  • Domain Name: At one point in time, the domain name itself was a meaningful ranking factor. If the domain name exactly matched the keywords you were trying to rank for, you’d rank better just because of that: in the SEO world we call that an exact match domain (EMD). If we want to rank for Artisan Snowflakes and our site is www.artisansnowflakes.com, then that is an EMD. A PMD, or Partial Match Domain includes the keywords and some other stuff, like www.myartisansnowflakeshome.com.

    It’s worth noting here that there are actually two separate advantages to EMDs and PMDs. One is the algorithmic boost, but another is that when people link to your site, the anchor text of the link usually ends up being the keywords you want to rank for. Thus artisansnowflakes.com has tons of anchor text links for “Artisan Snowflakes.”

    The downside of EMDs is that you tend to lose brand User Metrics signals, and with brand signals growing in importance, it’s far better to have a strong brand name than an EMD for a site that is built to last. The actual ranking boost for a EMD is very small these days, and most of the advantage actually comes from the anchor text of links. In general, I do not advise pursuing EMDs. It’s a poor long-term strategy.

  • Site Speed: Google has publicly said this is a ranking factor. It really isn’t, not at all (technically latency might matter, but not load time). By any practical measure or test, Google will not reward a faster site with better rankings.

    But, there’s a great big but.

    A much faster site will result in a better user experience resulting in more time on site and better conversions. It’s not a linear progression, but at certain plateaus, certain jumps up in load speed, it matters a lot. Heck, Amazon once said that on average every 100ms of load time cost them 1% in sales. And as we know, these User Metrics are watched by Google and thus those can influence your rankings.

  • HTTPS: Google has said that making your site HTTPS secure by default will be rewarded by better rankings. It does not, at all. In fact a Googler later clarified that you would not see any boost to organic traffic for being secure.

    In practice, if you move your non-HTTPS site to HTTPS, you can expect to see a drop of around 10% – 15% of your organic traffic for a period of 4-5 weeks after the change, after which your traffic will come back up as your rankings stabalize. Some sites see sustained ~5% or so loss of organic traffic, while others see a full recovery.

    Only ecommerce sites or other sites that must have a security certificate should even consider HTTPS, and even then there is not, currently, any net gain. If you’re building a brand new ecommerce site, it’s definitely a good idea to start at HTTPS out of the gate. It’s always possible Google will make this an actual ranking signal at some point, and you can take advantage of certain technical advances available only to HTTPS.

  • Mobile Friendly: Google has said that they give a ranking boost to sites that are mobile friendly, and there was even a name for the 2015 cutover to this algorithm change: Mobilegeddon.

    In practice, I saw absolutely zero change in mobile rankings after Mobilegeddon, or the 2016 update. I carefully monitored mobile friendly and unfriendly pages on my own sites, and many competitors. There was no change. That said, some sites did report seeing meaningful ranking changes, particularly in the local SEO space. Mobile friendly sites also benefit from the “Mobile Friendly” tag in mobile searches — though currently that doesn’t seem to be driving any improved CTR. Google let’s you test whether your site is mobile friendly Here.

    Most sites really should have a mobile-friendly website. But SEO is not the driving reason to do so.

  • Freshness: For certain queries Google favors pages that have been updated recently. In extreme cases, like news, Google favors pages that are brand new (but from sites with good authority). Note that this is all query-based: if you’re searching for information on maple trees, the freshness of the content doesn’t matter. Similarly freshness does not appear to be a significant factor for most ecommerce searches.

    Anything about news, current events, celebrities, or politics is likely to enjoy rankings boosts from recent updates. For other queries, which encompasses most sites on the internet, you can safely ignore freshness. it’s not the universal thing that so many SEOs think.

How Ranking Factors Work Together

We’ve now covered the high-level view of a lot of ranking factors, and it’s worth taking a moment to discuss how they work together. Not all ranking factors are equal, by a long shot, and the broad categories of ranking factors — Topicality, Authority, and User Metrics — are not equal either.

Broadly speaking:

  • Topicality is the least important: the weight of topicality as a whole has decreased as User Metrics increased. This makes sense too: topicality is the easiest signal to game (cheat). As discussed in Topicality Ranking Factors, topicality also seems to have a cap: beyond a certain point, additional on-page optimization doesn’t seem to help any further.
  • Authority is the most important: this has always been the case, and is still the case today. However, Google has grown much better at sorting through which authority signals to count, and which to ignore (or devalue). SEOs used to building crappy garbage links sometimes feel like Authority isn’t as strong as it once was, but every good SEO test continues to prove that Authority remains an overwhelmingly strong ranking signal.
  • User Metrics is in the Middle: Once not even a factor at all, User Metrics is now much more important than topicality, but still second behind Authority.

It is important to stress that you cannot ignore any of them. The three categories of ranking factors work together in a multiplicative way. If your site is going great in Topicality and Authority but has shitty User Metrics, then odds are you are not going to rank unless you have a truly overwhelming number of links. A site can be weak in Topicality and still rank through superior Authority and/or User Metrics (Wikipedia stubs are known to do this, for example).

How Google weights each ranking factor isn’t perfectly known, but we do know that those weights change based on the query. For some low-volume queries Google doesn’t have good User Metrics data, and so is ranking almost entirely on Topicality and Authority. For others Google has massive User Metrics info and a site with great User Metrics can dominate another site with hugely more Authority (normally considered the strongest ranking factor). But there are also cases where a demonstrably worst site has so many links that it can outrank vastly more deserving sites.

Basically the more data Google has, and the better the data is, the better Google is going to be at ranking the site that the plurality of internet users agree is the best site. But Google isn’t perfect.

Where to Start?

SEOs need to prioritize what signals to work on for maximum return. Often this means starting with Topicality: even though it’s the weakest signal, it’s so incredibly easy to optimize that it’s usually the fastest way to get returns. But if a site has adequate, if sub-optimal, Topicality but no links, then Authority is going to be a better place to start.

User Metrics signals are often the hardest to work on (and involve the most conflict for SEOs, since that job usually belongs to a combination of Product/Storefront and Marketing teams) but also tends to have out-sized returns, since improvements in conversion rate results in more money from every channel.

Unfortunately the Doctor doesn’t have a single prescription that will work for every site, or even a flowchart to determine how to prioritize your SEO. A lot of determining the size of opportunities is based on experience. I can say, however, that you should rarely be working on only one category of ranking signals for an extended amount of time.

 

User Metrics Ranking Factors

Back in the olden times of SEO, Topicality and Authority were basically the only ranking factors and SEO was a simple matter of on-page optimization (often to the point that the page read poorly to actual humans) and link building (often with spammy tactics that would get a site penalized today).

But as Google’s user base grew, Google started using signals from those users to help it determine which sites to rank. For a long time the effect of this was pretty subtle, but Google has been turning the User Metrics knob up and up until now it makes a serious impact in the ability of sites to rank well.

User Metrics ranking factors include:

  • Click Through Rate: the percentage of users who click on a site in the SERPs can increase or decrease a site’s ability to rank. This percentage is based on the position (we expect #1 to get more clicks than #2, of course) and how the click through rate (CTR) compares to other sites in the same position. Google has in the past tried to say they don’t use this for ranking, but tests have repeatedly proven this as a factor.
  • Branded Queries: Searches for a site’s name and the percentage of all searches that are branded is a way that Google determines how sought-after a site is. Of course this only works if your site has a brand name that Google can associate with your site: if a site is buy-sneakers.biz you might have an advantage in anchor text links, but Google is unlikely to associate searches for “buy sneakers” as a brand signal for your site.
  • Compound Brand Queries: Searches that include a keyword along with a brand name (making the query navigational) will improve your site’s ability to rank for that keyword. For example, if lots of people search for “sci-fi books amazon” then Google will start to rank Amazon better for searches for “sci-fi books.” I’ve done some interesting tests in the past that have proven this effect pretty definitively.
  • Time On Site: The longer a user spends on your site, the more likely it seems to Google that your site was a good result. This is, of course, almost certainly compared to time on site for other sites ranking for the same search query.
  • Pogo Sticking: If a user clicks through to your site, stays a very short time (compared to other sites ranking) and goes back to Google and clicks on another site in the search result, this is called pogo sticking and is a strong signal that your site was a bad result, which will eventually hurt your ability to rank.
  • Traffic: While not yet proven, there is some compelling correlation data that suggests traffic to your site can be a ranking signal by itself. This doesn’t seem to apply to paid Google traffic, but I’ve seen many cases where a surge of traffic results in a boost to rankings where no other ranking factor changed. This also explains edge cases where known non-factors (like social links) sometimes appear to help rankings.
  • Task Completion: This is an unproven, but compelling idea in the SEO world. What I  know is that every site I’ve seen that substantially improves the overall user experience (typically resulting in much better conversions) ends up with better rankings. The theory for this is that Google tracks task completion: they know what users search for, and what every page on the internet is about. So Google knows the last site a user visited before changing tasks (searching for a different query, or moving on to websites unrelated to the query topic). That last site visited is likely a good result for the query, even if it wasn’t the result Google served.

While it’s true that many User Metrics ranking factors tend to favor large brands (who have more branded searches, compound brand queries, and often better CTRs), User Metrics also give a great opportunity for the smaller guys of the internet to get ahead. After all, the giant brands are often burdened with slow, lumbering bureaucracies that make it incredibly difficult to stay on top of trends and they typically have sub-optimal user experiences.

User Metrics also make it much more difficult to rank when your site doesn’t deserve to rank. Determining if you really do deserve to rank (meaning you are the best site on the entire internet for a query) is a serious question SEOs much ask themselves before investing a lot of work into a crappy site.

How the Hell Does Google Know What Users Do On My Site?

I get this question a lot when I talk about User Metrics, so now’s probably a good time to get ahead of it. Google collects user data on every site visited via the Chrome browser, via Android devices, via anyone on a Google Fiber network, and it’s even been proven that Google collects at least some data on users of other browsers with a Google extension (such as YouTube plugins). Some of this we know via testing, and others (like Chrome) we know just by reading the User Agreement.

With all of this data, Google is far, far beyond what is needed to get a statistically significant sample of the population of the internet. You cannot hide the shitty user experience of your site from Google.