The future of technical SEO is SEO on the edge—Barry Adams

I'm Barry Adams, and this is SEO in 2023.

Arie, what is your number one SEO tip for 2023?

I think in 2023, SEOs need to seriously start looking at doing SEO on the edge. Now, in itself warrants probably some explanation. What is SEO on the edge? I mean, we all know and love CDNs like Cloudflare that help us make our website more performant, that makes sure that our international Users aren't necessarily communicating over the internet to servers located on the other side of the planet, but always connecting to a version of a website that is hosted very close to where they are. That's the beauty of CDNs. That's what they do. SEO on the edge is basically using the power that the CDNs have, the features that they've expanded into and have been building over the last few years, to adjust our website on the CDN rather than on our own servers, so that the CDNs. can be manipulated. The cached copies basically of web pages at the hoster can be manipulated to implement all kinds of SEO enhancements like redirects, canonical tags, hreflang tags, schema markup, etc., etc. And why would we want to do this on the edge rather than on our own web servers? Primarily because it, it's easier to test and try things out rather than trying to make things happen in a very busy developer's queue on our own web servers. It's a very good mechanism for testing changes out on limited sets of web pages. Like for example, a set of category pages in an e-commerce website or a limited set of product pages, rather than rolling it out across an entire website. And it's that sort of ability to do basically A/B split testing for SEO that can really elevate your technical SEO to a higher level and try things out and see if they actually work in your particular scenario, in your particular context, before you make a ticket for your developers and potentially have to go through a very lengthy and expensive development project for what might be an untested and untried implementation. So, I'm really excited about SEO on the Edge and how it's evolved over the last few years, because it's not a new concept. It's been around for at least 5 years, but it's reaching a certain level of maturity now that really makes it ready for the limelight.

So, you mentioned Cloudflare there. Is Cloudflare the best CDN for this?

In my opinion, yes, it is, because Cloudflare Workers are really at the cutting edge of what you can do on the edge on Cloudflare. The features and capabilities are second to none. And to top it all off, the, the free level of Cloudflare that anyone can sign up to has Cloudflare Workers enabled already. So you can basically test this stuff out for free on Cloudflare, which I don't think you can do on any other CDN.

And just firming up the definition of SEO on the edge, is it just about executing JavaScript onto a CDN, or can you execute other technologies? Does it not necessarily have to be JavaScript?

It doesn't have to be JavaScript at all. That's the beauty of Cloudflare Workers and SEO on the edge. I mean, we already know that, for example, with the Google Tag Manager container, you can do a lot and manipulate web pages. The downside of that is that it requires client-side JavaScript code to be executed and which, you know, Google does when it indexes pages, but doesn't do consistently. With Cloudflare Workers, while the code can be— itself runs in JavaScript, it doesn't run client-side. It runs on the CDN, which means the changes that are being made to those web pages are made server-side, and Googlebot, and as well as your end users, see the completely rendered changed page instantly in all HTML and CSS, and doesn't rely on any client-side JavaScript for those changes. changes to be revealed to Googlebot, which means it's much easier and much more thorough and much more reliable way to implement SEO changes rather than try and do it with Google Tag Manager containers.

Are there quite a few websites that use multiple different CDNs depending on where in the world a user happens to be? And if so, is there some kind of layering of the CDN sequence that the page loading has to go through in order to actually ensure that the, the most up-to-date version of a web page is loaded.

Yeah, that is one of the, I would say, downsides, or possibly one of the risk factors of doing or manipulating web pages on the edge, that you need to know exactly where in the technology stack it's happening and keep very close documentation of what you're changing, where you're changing it, and how it fits into the whole flow. Because it can be quite easy to lose track of what you've changed on the edge and what you haven't changed, because it's not on your live server, it is on the edge, in the CDN. So, you need to build a very robust internal process where you use the CDN to do certain things and document it extensively, with the aim eventually of implementing it on your actual technology stack, on your actual live server, which the CDN takes its root copy from. So, that you never run into scenarios where suddenly changes disappear from one day to the next, because, you know, your live web server gets updated with the new set of production code, and you clear all the CDN caches to make way for this production code, but some of the changes on the, on the edge aren't being taken over by the new version. And there's all kinds of other risks that you can get into. So it's not something you just jump in willy-nilly and say, oh, let's just go and try stuff out. And there are providers actually that help you with this. There are some providers like SearchPilot. They have a meta CMS that, that does this on the edge. Semrush have now stepped into this as well. They have a platform to do A/B split testing for SEO on DX. There's other providers out there as well who do this, and they can probably help you to a large degree to automate this process and make sure that it becomes part of a robust process and you don't necessarily create huge gaps between the website that your end users and Googlebot sees and the website as it exists on your own live server. Because that is probably the biggest risk that you have by doing SEO on DX, that you're creating a fork of your website on DX And that fork diverges too much from what your actual website on your live server is like.

So is there a recommended minimum size or genre of website that SEO on Edge is most suitable for?

I think you can do it on any website of any size and any scope. Having said that, if you have a very simple brochure website that runs on a very simple CMS like WordPress or Wix or whatever it is, doing SEO on the edge is probably a bit overkill because it's easier to do it directly on the website. But the bigger the website is and the more complex the website is, the more advantages you can get from doing SEO on the edge. Because generally for those sorts of websites, making major changes in your live site can be very high risk, very expensive, and very time-consuming. And that's where, you know, making these changes on the edge, on the CDNs, is so advantageous. So you can sort of bypass all those obstacles and get it done fairly quickly and do it also in limited context, limited circumstances, so that you don't necessarily risk the success So if you're working on a site and implementing SEO on the edge for the first time, what are some of the initial elements that you would look at for the potential biggest initial wins for SEO on the edge? I think one of the easiest things to implement and potentially one of the most impactful is automatic redirects. At the moment, you know, managing redirects on large websites can be quite daunting, difficult task as well, not necessarily worth the enormous amounts of effort it can take, especially if it's a large website with a potentially millions of redirects, whereas on the edge you can fairly easily write simple scripts that map redirects almost entirely automatically and make sure that your end users never end up on pages that used to exist but don't exist anymore. And, you know, again, you want to be a little bit careful with that because sometimes you actually want the website to serve a 404 Not Found page for when a page actually has never existed. But, you know, at-scale redirect mapping, especially if you do site migrations, that's probably one of the most time-consuming and difficult processes. But on DX, it probably makes it a bit easier to do. Same with implementing hreflang tags. It's one of those things that can be very difficult to build in, but can be done relatively straightforward on DX. Updating schema markup, you know, if you have additional attributes you want to add in your schema markup, and you're not entirely sure if you should do it across the entire website, on DX you just do it for one section of your website and see what happens. So those sorts of practical implementations are probably the best and simplest way to get started with With SEO on the edge.

You mentioned Search Pilot and Semrush there. I'd like to talk a little bit about testing, using the edge to test various new hypotheses that you come up with that may actually be able to assist with some aspect of your SEO. So how would you get going with using the edge for SEO testing? And which elements would you tend to test initially?

There's almost no limit to what you can test. It's one of the reasons why I like mentioning Search Pilot because they have a newsletter that they regularly write about case studies of their actual clients where they run tests and see what happens. And sometimes the outcomes are quite counterintuitive. One of the tests, for example, that they run was using emojis in titles or listing prices in meta descriptions and seeing whether or not Google picked those up. And the beauty of doing those tests on the edge is that you can just pick one category of your e-commerce website and make that change just for that category and see how it performs compared to the rest of the website. So you have sort of a control running side by side. So you can fairly effectively isolate that particular change that you've made and see what the actual impacts are, which is not something you tend to be able to do if you roll it out across the whole site because websites are complicated. Your website might have different levels of topical authority for different product categories. So you know you can very easily get lost in in what is causation, what is correlation. Whereas with SEO on DX, you probably have a bit of a tighter connection between the change that you make and the impact that it has. And sometimes, you know, we implement these things on websites because they are SEO best practice, but it doesn't always mean they actually improve your your traffic and your rankings, which is why running these tests before you run it out across the entire website is generally a smart thing to do because you're sort of you know. undermines that best practice and say, right, let's actually test this. Let's actually gather some data, do a proper test on a significant number of pages that gather a significant amount of traffic and rankings. Let that run for a couple of weeks, a couple of months, and see what the actual impact is before we decide to make that change wholesale across the whole website.

You mentioned the possibility of including a price and a meta description. The thing that scares me about potentially incorporating a price or an offer somewhere in the SERP is that you're not in full control over getting rid of that if that particular offer finishes. Should that be a concern, do you think?

I think it's a concern anyway, because Google has a mind of its own. You know, the SERP snippet that Google shows doesn't always necessarily reflect what the web page itself says. Google just, you know, increasingly decides for itself what to show there. At best we can do is send strong hints to Google, and this is what you— what we want to show. I think in most cases, it's not too much of a concern, Especially if we back it up with structured data. If we have structured data that has the actual accurate data in there, Google tends to show that, especially in the context of e-commerce, quite accurately in search results. It becomes more of a worry if we have websites that are bound by certain legal restrictions that they have to offer accurate content, in which case, you know, you don't want to start experimenting anyway with putting inaccurate information in your web pages. But even then, Google sometimes has a tendency to overwrite your snippets in search results with its own copy, which may not necessarily be accurate as to what your webpage actually says. But that's an entirely different discussion to have. And let's not open that particular can of worms.

Yeah, Google would also probably say, well, why don't you just use Google Ads instead? You can control that completely. But that's another can of worms as well, possibly. Look, you've shared what SEO should be doing in 2023. So now let's talk about what SEO shouldn't be doing. So what's something that's seductive in terms of time, but ultimately counterproductive, something that SEO shouldn't be doing in 2023?

I think it's— I mean, I work with a lot of websites that produce a lot of content. And it's very seductive to focus on the volume of content. And just, oh, let's put more articles out there and try to cast as broad a net as possible and try to capture all the traffic with what you can honestly consider to be fairly low-quality content. It's very seductive to just have a content calendar and say, oh, we're going to write X amount of articles every week and publish them and tick these keywords off our list and off we go to the next bit. not really consider the value of that content, the intent of the search that that content is supposed to fulfill, the quality of the content that you deliver. And that is very tempting because, you know, it's output, especially if you work in an agency environment, it's output that the client pays for, it's output that you can make visible to a client and say, hey, this is what we've done, but it's almost entirely valueless output. And I really want us as an industry to have different conversations with our clients and with websites in general that go away from, you know, X amount of hours and amount of articles and X amount of links and more towards this is the value add that we've brought to your website. This is the actual quality that we've delivered for your audience, for your users. This is what we've managed to contribute for your business goals as an organization. So less output-based, more value-based, I would say, especially when it comes to generating content for websites. We need to put more effort into individual pieces of content and make sure that we are proud of that content when it is published and we genuinely feel it deserves to be a top-ranked result. on Google search results because it fulfills the need that the searcher has when they type in a specific search on Google.

Barry Adams is an SEO consultant at Polemic Digital, and you can find him over at polemicdigital.com. Barry, thanks so much for being on SEO in 2023.

Thank you very much for having me, David.

Get your copy of SEO in 2023, the book, over at seoin2023.com.