Us SEOs put a lot of effort into increasing rankings. We spend thousands of marketing dollars creating link bait, weeks reorganizing the internal linking structure to best support important pages, and hours complaining that Target’s terrible category page or eHow’s poorly written post is still outranking our perfectly crafted page.
But we rarely spend much time trying to increase the click-through rate (CTR) by appealing to searchers with a more optimized page title and meta description. Talk to any PPCer, and they’ll confirm that their jobs aren’t just about coming up with keywords and paying for the highest ad position. It’s also about increasing the click-through rate of their ad copy.
SEOs usually make sure that page titles and meta descriptions exist, at the very least, and start with the target keyword. We’ll remind our clients or coworkers that search engines only display 70 characters in a page title (512px if you’re very specific about it) and 150 characters of a meta description. We rarely worry about the human aspect of page titles and meta descriptions.
Think about it this way: every time Google ranks your page, it’s giving you free ad space.
Optimizing page titles and meta descriptions for click-through rate is hardly new in the SEO sphere, but it’s not done nearly enough. The trick is, you can’t just read generic recommendations around the web, you need to actually test some variations, albeit in a fake environment, since Google doesn’t have an option for split testing.
SEO Review: Basic Guidelines
Quick note: this section is for people who are not full time SEOs, and can’t rattle off the basic guidelines for page titles and meta descriptions from memory. If you’re an SEO or have spent time in this field, skip this section.
Page titles are limited to 512 pixels, which is roughly 60 – 70 characters based on the width of the letters you happen to use. Google will cut off your page title when the last full word can’t fit within the 512px limit, and show that there’s more missing with an ellipsis (…). Typically, page titles should start with the keyword that you’re targeting, or that keyword should be very close to the beginning. The further right a phrase is placed, the less impact it has on rankings. That’s why most companies place their brand at the end, since they want it to be a part of the title, but they aren’t really competing for it.
Google prefers to use the page title that you write, but if it’s overly long or doesn’t represent the page, Google will ignore your input and write its own version. I’ve seen this most in cases where the page title is so long, the brand name isn’t seen, so Google will cut the page title off early and add “- [Brand Name]” to the end.
Meta descriptions are limited to 150 characters (or 120 on mobile, so err on the shorter side when writing these) and trail off similarly to page titles. Meta descriptions have no direct effect on rankings and are often replaced with an actual snippet from the page if Google deems that bit more relevant to the query. Still, if you write a great, keyword-rich meta description, Google will be more likely to use it.
Rich snippet data, like reviews, authorship, and video clips can also affect click-through rate, but I’m going to skip them for this post to keep things simple.
When you write up potential page titles and meta descriptions, you can watch the length with the word count in Word or conditional formatting/formulas in Excel, but I prefer to use the handy-dandy SERP Snippet Optimization Tool from SEOmofo. It takes your page title, meta description and URL, and shows a clip of how that input would look in Google’s search results.
PPC Insights: What Makes People Click
I worked in PPC for about 4 years before I started working at Distilled, so I’ve had the benefit of constant, almost immediate feedback on every bit of ad copy I’ve written. A lot of that feedback was company-specific, but there were a few rules that I learned almost always make an ad more successful:
Include keywords in the title and description. No surprise, Google bolds them, which really highlights that your page is a very relevant one. When you’re targeting one keyword very specifically (which you will be, since you have to test by a specific keyword), include it in the page title and the meta description.
Start meta descriptions with verbs. It means that you’re telling the searcher how s/he will get to interact with the page, which is easier to imagine and more compelling than simply describing the contents of the page. Consider these two options:
Both offer essentially the same message: lots of shoes, lowest price. But the first option talks about the site and forces the reader to interpret how s/he will use the site. This only takes a split second, but on a busy search results page, sometimes that’s all it takes.
The second option describes the website in terms of how the searcher will use it. In my experience, this is much more compelling, both for ads and organic snippets.
Here’s the trick to make sure you’re explaining how to use a page rather than what the page is: always start the meta description with a verb. It makes the sentence active rather than passive.
End the page title with the brand, unless you’re targeting the brand as a keyword. When people are searching for “brand name shoes,” sticking “My Store” at the beginning of the page title shoves the unbranded keyword to the right, which is out of the line of sight for skimmers. Consider the eye tracking data SearchEngine Journal posted:
People skimming through search engine results pages skimmed about 10 characters in, and back and forth about 20 characters. That means that you really want their search term in those first 20 characters. Don’t let your brand get in the way.
Each brand has its own way of appealing to people. Keep the page title and meta description in the same tone as the text on the page. Use your AdWords account to see which ads are most successful, and keep those brand-specific rules in mind as you write organic text.
All of these rules could be wrong: TEST
Now that you’ve learned a bunch of CTR theory, let’s put your actual pages to the test.
Google doesn’t let us split test page titles and meta descriptions for click-through rate, and it’s dangerous to compare metrics from different time periods. I wouldn’t recommend that you test with paid search either, since you’re limited to fewer characters and visitors expect more aggressive marketing in the top and right columns. So, how does one properly test for organic click-through rate?
For those of you who have been in the field for a while, you may already know the answer: SERP Turkey. SERP Turkey is a tool made by fellow Distiller Tom Anthony that creates a split test for organic search listings in a few easy steps.
To put together a SERP Turkey test:
- Write the keyword you want to test search results for in the homepage of SERP Turkey:
- Search for that keyword on Google, copy and paste the source code into SERP Turkey so the tool can see what natural results look like:
SERP Turkey will process the code and turn it into an editable version of the search engine results for your chosen search phrase.
- Create variations of that search engine results page. You can do that by clicking “duplicate” on the default that SERP Turkey automatically creates:Create a duplicate version for each snippet you want to test, leaving the default as a control. SERP Turkey is set up so you can really test anything about search results pages, but what we want to do is create variations where we only change our search result: leave its position and the competition’s page titles and meta descriptions. You should end up having a Variants page that looks something like this:
- Now you can go to the Dashboard (in the top navigation). You’ll get a test URL, which will randomly generate one of your variations. You can give this URL to testers, who will be randomly shown one of your variations and asked to click through to the result they’d be most likely to click on. SERP Turkey will record which result they clicked on.
- This can be done manually, but is set up to be the optimal way to use SERP Turkey with Amazon’s Mechanical Turk. You can learn about setting up SERP Turkey with mTurk here.
Remember to save the URL to the dashboard of your study! That’s the only way you can access it again.
Once your testers have all chosen their favorite search result, you can compare click-through rates for different variations in the “Results” column of the “Manage Variations” page.
If you find that one of the variations has a statistically significant increase in click-through rate over its competition, try changing your actual page title and meta description, and see how many more clicks you get for that keyword next month with Google Webmaster Tools.
Let’s make this whole process more concrete with an actual example. (Before you wonder: no, Target isn’t a client, that’s why it’s okay for me to do this in a blog post.)
Right now, Target is ranking #2 for “bathing suits:”
Compare Target’s page title and meta description to that of Victoria’s Secret. Victoria’s Secret’s:
- Starts with a verb, to communicate the experience visitors will get at victoriassecret.com
- Uses standard brand language, showing the company’s personality
- Actually uses the keyword
- BUT: the page title should be trimmed to fit the space
Target’s page title and meta description, on the other hand:
- Looks very generic and redundant (“women’s clothing, women”)
- Doesn’t ever include “bathing suit:” that snippet of text is so bad because the meta description doesn’t use “bathing suit,” so Google pulled text from the page
Step 1: Should we be targeting “bathing suits?”
If you want to target a particular keyword but find that all results seem to be targeting a synonym, take the time to see which one is actually more popular.
(From Google Keyword Planner.)
Okay, so, “swimsuits” is a more popular term. I was raised saying “bathing suits” more often, though, so my gut says this might be a regional thing. Google Trends agrees:
Bathing Suits Swimsuits
(The darker the blue, the more often the term is used there.)
Based on my completely unresearched knowledge of Target’s target market, I would say that we should be targeting “bathing suits” rather than “swimsuits.”
If this were an actual client, I would also look into:
- Whether “bathing suits” or “swimsuits” converts better through paid search
- How much traffic Google Webmaster Tools says target.com is getting from “bathing suits” vs “swimsuits”
- Whether my assumption about different target locations is correct, from Target’s internal data
There’s also the potential to steal traffic from victoriassecret.com by targeting “bathing suits” while Victoria’s Secret is primarily targeting “swimsuits.”
Step 2: Create variations
For the first round of testing, I like to come up with different hypotheses about what will draw people to a page:
- Target customers are looking for good deals, so they’ll appreciate promises of good savings
- Target customers like variety, so they’ll be most interested in a wide selection
- Target has been working with designers to create lines of clothing that are more exciting than their standard Mossimo brand: promoting a designer will get attention
To test these, I wrote up a few page title and meta description combos. You can typically get multiple messages in each, but I tried to highlight one selling point per variation.
Note that I am not a writer: I don’t think that any of these are quite as compelling as Victoria’s Secret’s, but then, Target has a less aggressive advertising tone. If you have access to a writer in house, definitely get their feedback here.
Step 3: Run these on SERP Turkey
(See previous section for how to run the test.)
Once you’ve run the test, the results will be shown in each separate variant page on SERP Turkey. It’s easier to read the results if you pull out your snippets only and directly compare their click-through rates:
As Seen on Google Results Page
These are made up numbers, but if the click-through rate did jump from 14% to 20% for one of the variations, it’d be fair to assume that changing your page title and meta description could increase traffic for the keyword targeted.
Remember, though, that these tests have to test by a certain search term, and the changes could lower click-through rate for another search term(s). In this case, I would follow up this test by testing the winning variation against the original for the keyword “swimsuits.” If the changes cause a drop in CTR for “swimsuits,” we’d have to use the clicks by keyword in Google Webmaster Tools to decide if the change would net more total organic visitors.
And that’s it!
Not such a complicated process, once you get your hands dirty. I, personally, love the added verification before we put changes live on the web.
What about you? Have you tested snippet changes, and what results did you experience? Let us know in the comments below.