With over 50 code releases a day at Wayfair.com, we iterate quickly, and our teams have the ability to be fast and agile. The downside is there are a lot of opportunities for a mischievous bit of code to slip through safeguards and negatively impact SEO. And if a bad thing can happen, it inevitably will. However, some of these accidents reveal insights that can improve our understanding of Google’s algorithm and enhance our effectiveness as SEOs.

On the Wayfair SEO team, we call these accidental SEO tests.

What’s in a title tag?

One of our earliest accidental SEO tests showed that title tags don’t impact rankings at all on a well-optimized page. Our theory is that there is an on-page optimization saturation point, beyond which further on-page optimization no longer improves your ability to rank.

This test started when an unrelated code release wiped out our title tag overrides. Our title tags are automatically written using rules based on categories and product names, but we have the ability to override these title titles with ones we create manually. So the default title for the sofas page might be “Sofas | Wayfair," but we might override that with “Sofas & Couches | Wayfair.”

This accidental test killed the manual overrides, but left the default logic in place, and was live for a couple weeks before we noticed it. The most fascinating part of this test was that although the title change happened to all our sites, it impacted one particular site in a unique way.

Organic traffic on Wayfair was unchanged:

Organic traffic on AllModern, however, fell 17%:

Why did one site take a big hit and another wasn't affected at all—particularly when the default title tags were identical for many of the pages? I believe it’s because of how well the pages were optimized for their targeted keywords.

The Wayfair sofas page has sofa keywords all over the page, in the URL, and is generally very topically relevant. AllModern, on the other hand, is targeting “modern sofas” rather than the generic “sofas.”

But the AllModern page and URLs rarely use the word “modern” at all. Since everything on the site is modern, it’s redundant to call every product modern. This means the title tag is one of the few topicality signals that help Google understand that the page is about a specific style of furniture.

We saw confirmation of this in our rankings: Rankings for modern keywords had fallen across the board with the title tag change, while rankings for more generic terms like “baby bedding” didn’t change at all.

Further tests

By happy accident, this SEO test seemed to support a theory I’ve been testing on and off for years. After reading a correlation study that showed no correlation between title tags and rankings, I set out to prove it wrong. Instead, I found that well-optimized pages held the same rankings regardless of changes to their title tags, while poorly optimized pages' rankings were hugely influenced by them.

Below, I list several examples of the pages I tested. (You can click through to each one to see the quality of its on-page optimization yourself.) These tests were conducted on everything from tiny B2B sites and blogs, to small ecommerce sites and ecommerce giants. Testing dates occur within the range of 2011-2015; so for many examples, rankings have changed in the years since testing concluded.

Keyword phrase: d&d dice
Site: Awesome Dice
Original Title: All Dungeons & Dragons D&D Dice | Awesome Dice
Test Title: Fantasy Gaming Dice | Awesome Dice
Result: Over a three-month period, there was no ranking change at position No. 3.

Keyword: ferrules
Site: Trans-Matic
Original Title: Custom Ferrules & Eyelets – Trans-Matic
Test Title: Custom Eyelets – Trans-Matic
Result: Over a three-month period there was no meaningful change in how this page ranked for the targeted keyword, as it fluctuated between positions 18 and 19 the SERPs.

Keyword phrase: pet abilities mop
Site: Warcraft Hunters Union
Original Title: New Pet Abilities in MoP | Warcraft Hunters Union
Test Title: Quilen Abilities Mists of Pandaria | Warcraft Hunters Union
Result: Rankings fell from No. 1 to No. 3 within a week, and stayed at No. 3 or No. 4 for a month. When the original title was restored, the ranking returned to position No. 1.

These example pages fairly represent all of my testing. Most of the other tests were conducted on different pages on the same sites and with similar results. However, more testing is always better, and I’d love to hear about results from other tests.

Optimization saturation

What appears to be happening is there is a saturation point for on-page optimization or topicality. If you think about it from Google’s point of view, it makes perfect sense. When Google is looking at the code of a page, it’s trying to understand how topically relevant the page is to the search query.

Let’s say someone searches for “sneakers” and Google checks its index for all pages about sneakers. If the word “sneakers” is on the page once, then it’s unlikely that page is very relevant to a “sneakers” search. It could just be a passing reference. Barring other authority or user signals, that page would rank very poorly.

If “sneakers” appears on the page three times and in the title, though, then it’s much more likely that the page is about sneakers; so, from a topicality perspective, it should rank better than the page that only uses the keyword once. If “sneakers” appears in the URL and the headlines, and is also in the body of the page text eight times, it’s pretty safe to assume that the page is all about sneakers—should rank even better.

At some point, a page is 100% topically relevant, in my view. The page is clearly all about sneakers. Adding the word “sneakers” more often or in more places isn’t going to help; you can’t be more than 100% all about sneakers. Google is now going to use authority and user metrics to determine where you rank alongside the other million pages that are completely topically relevant to sneakers.

The interesting part of this is the bar for reaching 100% topicality appears to be fairly low. You don’t need the title tag to get there. You don’t need the URL to get there.

You simply need a page with a decent amount of text that is clearly related to the query topic and intent.

And when you think about it, that makes perfect sense. It’s the way Google should work.

What it means for SEOs

While optimization saturation can be bad news for SEOs who do nothing but re-optimize pages, it definitely frees the rest of us up to concentrate our efforts on things that do matter.

Here are a few takeaways for SEOs:

• Good enough might be good enough. If you have pages with crummy URLs, without your keywords in them, they might perform fine if the rest of your optimization is pretty strong.
• There seems to be no reward for perfect. If your on-page optimization is good enough, investing the time to move it to perfect is likely not worth the effort. (Speaking here strictly in SEO terms, moving to a more perfect user experience is a different and worthwhile endeavor.) Rather than obsessing about keyword counts and image alt text and headlines, you can get good enough to hit optimization saturation, then move on to using your time and other resources in better ways. (That said, if you’re starting from scratch, of course you will want to hit as close to perfect as you can; it just might not be a good use of resources for an established site.)
• Title tag targeting: On a well-optimized page, you can use the title tag to target synonyms or alternate keywords. That said, I would strongly recommend that your title tag be used to optimize for clickthrough rate (CTR). A solid CTR can help improve your rankings in the SERPs and deliver more traffic from your current spot in the SERPs at the same time.
• The exception to the rule: I have seen, anecdotally, one exception to the optimization saturation rule. It seems like even on pages that are optimized to the point that the title tag ceases to matter, a meaningful increase in unique, quality content can still boost rankings. I’ve even seen pages taken from 200 words to 400 words receive a boost in their rankings, just like pages taken from 1,000 to 1,500 words. It appears that, even for pages that seem to be 100% topically relevant, Google still interprets a higher word count as likely better for searchers.