Lots to Learn from List of SEO Missteps

Lots to Learn from List of SEO Missteps

My father-in-law has a saying when sampling a new brand of scotch: “It’s single malt—how bad could it be?”

The answer? For every fine Macallan or Laphroaig or Clynelish we savor, there’s at least one Sheep Dip to lend danger to his saying. Some single malts, like some SEO strategies, can be pretty flawed.

I’ve audited enough websites to appreciate the goofy—and even tragic—outliers. Like GRIN’s Worst Website in the World project, they leave a memorable impression of what not to do.

For example, don’t code your full body copy text into each page meta description. Don’t publish 50-second teaser interview videos when no full-length interviews exist. And don’t embed a hidden link to X-rated content in the template for every H1 header on your website.

Goofy, yes. Plenty tragic. But these are the outliers. I bet plenty of customers at your local liquor store buy Sheep Dip. What are the SEO missteps most clients—maybe you, too—are making?

For that, we go to the data. TransPerfect recommends conducting an SEO audit on every source-language website before we translate and adapt it for other markets. Our audit weighs 64 factors across content, technical, and linking SEO to determine where we need to tighten up the source site and build in optimization for the new language site(s). I’ve averaged our results from audits we’ve conducted so far in 2015 to illustrate areas you should keep a close eye on.

Content SEO
Titles: On average, 58% of client page titles were duplicated on at least one other page. We want them unique and keyword-focused. 3% were missing titles entirely.
URLs: 17% included param characters – you know, that garbled brew of random numbers and symbols, often used for tracking visitors. Whenever possible, search engines like to see actual words to decipher a URL’s content. And people like that, too.
H1 headers: For the average client, 56% of headers were duplicates, 25% of pages were missing headers, and 21% of pages had more than one. Like titles, we want them unique and keyword-rich.
Meta descriptions: 36% of pages didn’t have them. Although Google isn’t looking for keyword value here, the meta description is what your audience sees in the search result. It’s two sentences of free advertising that can lead to clicks. And you can bet search engines measure those.
Meta keywords: Missing from 65% of pages. That’s fine for Google markets, where they’re not a factor. But in China, for Baidu,—include them.

Technical SEO
Mobile responsive: 44% of clients aren’t yet where Google insists they should be.
XML sitemap: 52% of sites audited don’t have one. We think it’s a great way to speed up indexing in search engines, particularly when you’re launching a new language site.
Page loading: Average homepage load time was 3.8 seconds; average page size was 1,900 kb—a bit heavy when Google advises less than 1.4 seconds and less than 500 kb. It may be partially due to 17% of images weighing in at more than 100 kb.

Linking SEO
Followed Links to Domain: Median values were 13,283 external followed links from 560 unique external sites, and a link diversity of 11%. That’s above our quality threshold of 10%—but always room to improve.
Page Errors: On average, 7% of internal client pages reported 302 (temporary) redirects, and 1% resulted in 400-level page-not-found errors. Of links to external pages, 22% ended in 300-level redirects and 2% ended in page-not-found errors.
Social Signals: Of the social media we analyze for most Western sites, client home pages, on average, gained 626 mentions from Facebook, 371 from Google+, 155 from Twitter and 94 from LinkedIn. Still, 59% of all clients we audited were lacking mentions from at least one—and sometimes all—of those outlets. Definitely time to get more social.

Colt Foutz is director of Digital Content Optimization at TransPerfect, the world’s largest privately held provider of language services and technology solutions. He is an ex-journalist, an ex-teacher, an ex-adman, and a very current father of three.