The Majestic SEO Podcast - Programmatic SEO - with Kevin Indig, Anna Uss, Anne Berlin, and Mordy Oberstein.

In this episode of the Majestic SEO Podcast, we discussed Programmatic SEO and how you can deliver SEO at scale, in an automated manner.

Joining David Bain were Growth Advisor Kevin Indig, Anna Uss from Snyk, Anne Berlin from Lumar, and Mordy Oberstein from Wix.

Watch On-Demand

Listen to the Podcast

Transcript

David Bain 

Programmatic SEO, how to deliver automated SEO at scale. Welcome to the Majestic SEO podcast and live stream. I’m your host, David Bain, and today we’re talking programmatic SEO, specifically how to deliver automated SEO at scale. So let’s get straight into introducing today’s panel. So starting off with Anna.

Anna Uss 

Hi, everyone, my name is Anna, and I’m the SEO lead at Snyk – the company that delivers security for every developer. I started my SEO career at Wix in Israel, and now I’m based in London, and Snyk is my current employer.

David Bain 

I love the tagline there as well, the short, descriptive word they’re describing the company. Let’s see if we can do the same piece of succinct and descriptive at the same time. So taking us up to our next panelist, Anne.

Anne Berlin 

Hey, David, thanks for having me, my name is Anne Berlin. I am a Technical SEO consultant at Lumar, formerly DeepCraw, and I spend my days crawling websites with hundreds of millions or billions of URLs.

David Bain 

Thanks so much for joining us and and next up Mordy.

Mordy Oberstein 

I’m Mordy Oberstain, and I am the head of SEO branding at Wix, host a podcast, and I spend most of my time when I’m not BS-ing in a podcast, talking about how what Wix does for SEO and how to improve your Wix websites from an SEO point of view to all sorts of users for all sorts of reasons.

David Bain 

Thanks so much for joining us, Morty. And last but not least, is Kevin.

Kevin Indig 

I’m Kevin, a growth advisor to companies like Nextdoor, Snapchat, Ramp and others, have been in the space for last 12/13 years and still love every day, before going out as an independent advisor, I lead SEO and growth at Shopify.

David Bain 

Super panel there, as you can see, as you can hear, let’s stick with Kevin for the first question. So Kevin, how would you define programmatic SEO?

Kevin Indig 

My definition is the systematic creation of templated pages. There are a couple of things that distinguish programmatic SEO from other types of SEO in my mind. The first and foremost is that most pages follow the same structure, and it makes them incredibly attractive from an optimization and testing perspective, because they’re so similar to each other compared to blog articles, which tend to vary in length and structure and topic and so on. There’s also a differentiation that I want to make between programmatic content for sites that have user generated content, and sites that don’t have user generated content. So for example, I would not call an eCommerce online store, I would not call that programmatic SEO, even though there are online stores with lots of brands that have lots of pages. At the same time, a UGC site, you know, that say something like a G2, which has a lot of software reviews or TripAdvisor, which has travel reviews and local reviews. I would also not call that programmatic SEO. And that’s where some people might disagree with me, even though these pages are programmatically created. To me programmatic SEO mostly applies to sites that don’t have inventory, like product inventory, or user generated content, but still create sites at scale. That would be my definition.

David Bain 

So the systematic creation of templated pages, please someone disagree with Kevin.

Anne Berlin 

I think what’s interesting about programmatic SEO defined as you have, Kevin, is that it becomes almost inaccessible to most SEOs practicing today. So I think the argument could be made to expand the definition, so that more practicing SEOs feel enfranchised. So what you’ve described, eCommerce sites that programmatically generate inventory pages, or CMS is that allow for meta descriptions and meta titles to be published without the hand of a human in one way, if we define that also as programmatic SEO, then the techniques we might discuss today for how that can go wrong, and how to review them and optimize them at scale might also apply.

Mordy Oberstein 

That’s a really good point because well, I mean, I might disagree and say that may not be programmatic SEO in the in the in the classic sense of it. But there are definitely programmatic elements that you as an SEO will inevitably incorporate like meta descriptions, title tags, etc. I want to automatically pull in my business name and the the blog post title to automatically be the title tag. So that is a programmatic element. So those are things and I’m sure we’re gonna probably get into a little bit of a debate about the values and the positioning of programmatic SEO. But to say that programmatic SEO doesn’t apply, or the elements of it or you know, some kind of version of it doesn’t apply to SEO is I think is definitely not the case. You’re almost always trying to programmatically do something as an SEO, so you don’t have to do it manually. It sounds like is it like SEO automation? Or is that programmatic? I don’t wanna get into that.

Anna Uss 

I really love all the points, and I hear the reasoning like from both sides, but I think I feel closer to Kevin’s position here. Because I think that programmatic SEO is a growth strategy, like eCcommerce is an essence, it’s the core of their business to have those millions of pages. And yes, there is a programmatic element there. And yes, technical SEO is at the core there. But I see programmatic SEO is finding an angle for your business, where you can find something closely tied to your product or to your audience, and step into that area to leverage that high potential growth and high scale growth, growth at scale, basically. And just to add to the initial definition, I think that it’s not necessarily ongoing creation of the pages, it can be created at once and then run through the years just with programmatically presets, updates, like they just come and go, like if something new is added to the database, it adds to the programmatic assets. If something does not exist, it gets removed automatically, and the other thing I wanted to add is definitely the scale that we’re talking about, 1,000s and millions of pages.

David Bain 

Anna, you actually shared before we started recording that you’ve seen a website taken from a few 100,000 pages, indexed to several million pages. So obviously, there are benefits to programmatic. I think one of the challenges is actually deciding on what to automate, if we can use that term as well, and what not to do to automate. So shall we have a little discussion about that? So shall we stick with you for a second Anna, what are a couple of examples of elements that you should automate, that you should utilize programmatic SEO for? And what’s perhaps one thing that you absolutely shouldn’t use programmatic for?

Anna Uss 

Well, in my experience, when we talk about programmatic SEO, everything there is automated, you preset the patterns, you preset the structure, you preset even the sitemap generation, but obviously, it requires the monitoring and the SEO audit. I was thinking that your question will lean more into examples of programmatic SEO assets and like what can be created? I had two examples of that.

One was actually came up on brightonSEO last September, and it was funny that the person who was presenting it, he didn’t announce it as programmatic SEO because like, that was not the terminology or the angle, but what they did was actually pure programmatic SEO, they are a ticketing site, and they have created singer profiles, they bought databases and they have created singer profiles. And similarly as we saw, they were ranking with those profiles, mostly for local singers or for someone least popular, but still amazing and amazing concept, you know, like someone wants to read about this person and then you show the tickets and the events of the of this particular person in that particular region. So I found that really interesting.

The second example is what we did, we are the cybersecurity company and we have created a programmatic asset to compare open-source packages for developers, where we use databases to take all the relevant information, plus, we use our engine to look at security, maintenance and vulnerabilities, and then we use our intel formula to calculate the security score for the package. So I just wanted to highlight the logic there, that it can be an extension to your business, you just need to find where your audience is, and tap into that opportunity and create it, maybe it will be even topper level, maybe it will be awareness. But nonetheless, you just need to know your audience. And then you will find the opportunity.

David Bain 

I’ve certainly seen many comparison websites using programmatic to compare different software packages. I would almost think that that’s a highly competitive there and actually quite hard to achieve a lot of success in that, but perhaps there’s a lot of search volume for that, and a lot of opportunity there. Kevin, I know that you’ve had some background in that area, what are your thoughts on that?

Kevin Indig 

A quick segue into your previous question about automation, I think there’s a really important differentiation to make. So I think that a lot of times when we talk about automation, we actually mean creating at scale, or, you know, multiplying efforts, if you will, where you once defined the parameters of a page, like the title description that the content, the headings, images, and you roll that out over 1000s of different permutations and iterations. And that is somewhat different from automating certain elements. So one example at G2, which you could argue is a marketplace, has programmatic SEO elements to it. One of the most impactful tactics was to insert a variable in the meta titles for category pages that automatically updates. So that is what I would call true automation, right? It’s you once you define that in the backend, so to say, and it will automatically update itself every day or every week. And that can be a powerful element of SEO, that is automated and programmatic at the same time.

Kevin Indig 

And so the answer your last question, David, again, I think it’s it’s probably the most scalable way to do SEO is programmatic SEO. And there are lots of different common recurring elements that we can influence when one is title, we already spoke about, another one are rich snippets, or snippets in general, internal linking, content on the page, and then often, there’s some elements of listing strategy, too, we have with software products which are the listings, you can optimize those as well. For example, the information that’s being displayed, the ratings, maybe some quotes from reviews, maybe categories, tags, and so on, and so on. So part of programmatic SEO that is really important is finding all the elements that you can influence and you can change, and then thinking about, Okay, which of those elements do we touch maybe once and then leave them as is and which other elements have to continuously change? And then what are the automation workflows in the backend that we can define for them?

Mordy Oberstein 

It’s dynamic, which is really cool. Because you could save yourself a heck of a lot of time. So alot of things that we end up focusing on, because we understand the websites, we created the website fundamentally, so we can do things like if you add your address to the homepage, we can pick it up and add local business structures in the markup for you automatically. It’s like you end up with all these dynamic variables that you can sort of pick up on, and then automate. And that can save you a tremendous amount of time. And if it’s in a controlled sort of environment, you usually don’t end up with any, or was any, but you don’t usually end up with a lot of like accuracy problems. And I think, where you kind of get messed up or I where I’ve seen people kind of get messed up is where almost like with AI, and I don’t want to I don’t think of AI now, but I guess I’ll do it anyway. If you’re using AI in a controlled environment, you’re telling it Okay, take this paragraph and write a header for it. So you’re in a controlled setting, if it’s an open setting where you’re saying, Here’s a topic, write me a blog post about it. That’s where I think things kind of get messy. And if you’re using if you’re trying to create dynamic content, and you’re doing it in a way where it’s very open ended, you’re not giving a very strict parameter, then I think you end up with all sorts of inaccuracies, you end up with things that aren’t targeted, you end up with all sorts of problems. So I think it’s important when you are doing things that the dynamic level to sort of create the environment and the borders and the context for it to operate it.

Anne Berlin 

I am also interested in double clicking on your your original question, When should we maybe not use automation, you know? Mordy, I am in agreement with you before we started broadcasting, you outed yourself as a programmatic curmudgeon, and I think, in this respect, particularly sort of the the unlimited, or the unrestrained application of automation and content generation does have risks. Kevin, you published a piece in January about the BankRate and CNET examples. You know, I continue to think of where automation should not be used under the rubric of YMYL, if we’re talking about financial advising or health advising, I think that programmatic could easily and inadvertently become a misinformation engine.

Kevin Indig 

I would agree that I think just blindly publishing automated content, or generative AI content is very risky. To me, the killer combination is AI plus humans. So having human fact checkers, editors, you know, people who make sure it’s factually correct, and also who publish the output is often not, you know, written in a great way or, you know, it comes with flaws. And so, to me, the question is, you know, how can you how can we leverage AI or automation to quickly create a draft that humans then kind of bring over the finish line? And again, make sure that factually correct. The BankRates and creditcards.com case is really interesting, because, and I don’t know this for sure, right, but on on the site, I think it was CNET that published 75 articles, and least half of them if not all of them had factually incorrect information, sometimes down to the easiest, simplest level. I think one of them was a compound interest calculation, that was just blatantly wrong. And so I think that I don’t think there was a lot of fact checking that went into this. And again, I don’t know this, for sure, I don’t want to blame anyone, or, you know, throw shade at someone, but every human fact checker would have corrected that would have caught that. So that, to me, is the important differentiation between generated content that works, and it doesn’t.

Anna Uss 

I want to say that if we’re talking about millions of pages, I would definitely not go for dynamic content, because it would be impossible to keep track of what is being produced there, and the cost of the error can be really high. So the combination I found that works is using modifiers. It is definitely not providing uniqueness, as of like, the grants uniqueness, but relying on data, if we’re using data, and we are showing data. That is always the best, because no one wants to consume programmatic content doesn’t content, usually, it’s some kind of like data, or most recent information, something reliant on numbers. But then since Google has a hard time interpreting those graphs, data, then providing a text description with modifiers helps with the SEO side of things. So I would go for that kind of combination. But I would definitely not use dynamic content for programmatic, that’s too big of a risk.

David Bain 

Yeah, I’m gonna, I think I think you’re all saying a similar kind of thing and actually leading on from each other, or but by the way, hi to Tevfik Mert Azizoğlu and Olga Zarr, you’re both part of the SEO in 2023 series. You’re watching us live. So thanks for joining us live. If you’re listening to this on Spotify, Apple podcast, come and join us live for next time and ask some questions if you can. What I was gonna say is actually it’s probably worthwhile exploring, tying what you’re saying together. So Mordy, you were talking about things get messy when you do things like, come up with a title and maybe give the AI the opportunity just to write an article. Kevin, you were talking about perhaps actually using that as a first draft, and then bringing humans in there as well. Anna, you were talking about the importance of data. I’m finding that incorporating things like tables and well structured information into content will probably give you a better opportunity to rank. So I guess, what are your thoughts on that? And also, in relation to that? What is Google looking for at the moment for its SERP to best utilize that information and display it within the SERP?

Kevin Indig 

Yeah, I love how you differentiate between In structured and unstructured content, right structured content, so much easier to automate then unstructured content. And I would even, I would even challenge you a little bit that unstructured content gets easier to create, I’ve seen some amazing stuff, I’ve done some tests with clients, and it gets better. Again, not perfect, not, not somebody would blindly published but much, much better. And so, to me, it all comes back to user intent and understanding what people are really looking for, when, for example, I mean, let’s go back to maybe two people looking for software reviews, but they’re also looking to compare software. And essentially, they want to, they want to have very specific criteria in mind, that they want to figure out to make a decision, price is a good example. Right? They might compare, let’s say, MailChimp, and ConvertKit. In the one understand for their specific audience, right, so they have 5,000 subscribers, what’s the difference in price, it’s a very specific user intent. But users might not search for that exact phrase, right? They might look for, you know, MailChimp versus ConvertKit price or something along these lines.

So the the art is to identify these patterns, what are the key pieces of information that people are looking for? And then see how you can programmatically add that that content or that information to pages? So do you do for example, we mess it a lot of time and our comparison and alternative pages? And seriously, like with the designer through all the pieces of information, like what can we display here? What can we draw from our reviews? How can we optimize reprocess to generate the data? And how can we how can we maybe you know, combine different pieces of data to come to new information. So for example, what’s an average price per subscriber for certain for email automation software, that is, in my mind, the key. So user user intents, breaking that down to key piece of information, and then figuring out how to bring the key piece of information to all of your programmatic pages.

Anna Uss 

I wanted to add on top of on top of the intent, that it is also very important to understand what is your added value there. Because when we talk about programmatic, it is so easy to create something a combination of information. But in the end, you as a portal as a company as a service, you don’t add anything new to that. And you could find the same information in a blog, for example. And if that happens, then obviously you have no chance to outrank human generated content, right. So, for example, in my case, the added element is the security expertise. So you can find all that information. On other websites, it is scattered, right, you can find maintenance in one place, you can find downloads in the different place. But the added element there is the security expertise that we add that element and we also calculate that score that advises you or discourages you from using this specific open-source package and also offers you alternatives, the packages that perform the same function, but have a better security score. So I think like whenever someone even considers that strategy, you always need to think, first, what is your audience? What can be interesting for them, then looking at the intent? And then thinking, can you add unique value through programmatic or somehow else, so that you outrank your competitors? Like, what’s your specific angle there?

Mordy Oberstein 

One of the things that I think gets really interesting with that, and talking to your point and to Kevin’s point is where the expertise demands a certain type of language structure. So for example, Google talked about the product review updates, and it’s one of the in the last one to February 2023 update, I’ve kind of seen this come out, in a couple of cases where they want first hand experience, and the language structure and when you actually have first hand experience is very, very different than when you programmatically build something, or you have no actual experiencing you’re just lying. And I think Google’s able to pick up on those language structures. That’s literally what machine learning is for. And now that becomes, okay, great, I might be able to offer unique value. All things being equal, I can offer unique value through programmatic content. But you still can’t mimic the language structure with programmatic content, and now what. So I think it’s really good to be really, really careful, even though you might be able to offer unique value, but being able to understand the ecosystem and where that is and where that isn’t applicable. By the way, one thing you could do is you could build programmatic content to try it out and see what sticks. Colt Sliva did that with is it Python and Wix Velo, which is a code stack. And he basically built up a website in order to see what would stick and it’s how you were going to search console and you see, okay I thought I was going to rank for these keywords and get input wasn’t for these keywords and I but I instead I have impressions for these keywords. Because Google understood me differently than I thought it would. So he used programmatic content to spin up a website in order to see how to Google interprets the website. And now what should I do with this thing? Did I did I get it right? Did I not get it right? Should I pivot? So I think you can experiment with programmatic content in that way, but I still think at the end of the day, you have to be really, really careful how what Google is trying to do in the ecosystem, that programmatic content, even though you might be able to have unique value is never going to be able to do.

Anne Berlin 

I’m really intrigued by the example you gave Kevin, of proactively restructuring their review forms on G2 to solicit data that’s proprietary that can be used to service these programmatic plays. Anna, when we talked about the examples that you’ve executed at Snyk, you know, this is we’re traditionally thinking about using API’s to plug into external data sources for our programmatic place. I am intrigued, however, by thinking about, well, how can we generate our own data, and then be the only source of these types of information? I work on a national jobs directory website, and we’ve been having conversations about what data points can we extrapolate from the information that we have access to that no other site can replicate? Particularly, this website has an issue with job category pages for very high in demand, but infrequently offered job types, where they’re thin and become de-indexed. Talking about things like can we calculate the rate of applicants per position in a given field? Can we calculate the number of positions in this field at this particular job title that are typically on offer in any given year? So I think it was, it’s a new angle for programmatic, that I’d love to see more examples published about where we didn’t plug into external databases, but instead farmed our own data.

Kevin Indig 

To that point, unique data is always is always king, right. And I think, especially in these hyper competitive verticals, like job search are incredibly competitive. The more unique insights you can provide, the more of a competitive advantage you get, I think it’s that simple. Of course, insights need to make sense. And it’ll be interesting for people. But I very much agree that a lot of times there is people just use the data that they have available, but that they don’t think about how does one and one equal three, how can we generate more insights out of the data that we have by combining it?

David Bain 

So what do you do when you’ve published a couple of million web pages, and they’re doing okay, but you just can’t actually maybe hit the top three in the SERP and you’re not actually bringing in that traffic, that volume traffic? So how do you how do you actually identify the opportunities? How do you actually see which pages probably you should be tweaking? So do use for example, programmatic to look at the existing SERP for all your target keyword phrases, and to identify what other sites are doing better than you and how you can improve?

Anna Uss 

I wanted to answer that with one of the biggest advantages of programmatic SEO is that you can actually do SEO A/B testing, and a lot of A/B testing at scale, as well as conversion testing. So that’s how I don’t think you need to focus on specific pages and where they rank, you need to focus on the large scale, is everything indexed is everything technically correct is everything being crawled and visited by Google. But then if those bases are covered, you have a huge playground to experiment with just a small change or tweak of the SEO title that can improve CTR by 1 or 2% can amount to a significant increase in organic traffic as a whole. The same goes with conversion like literally we tweaked the CTA and we saw 150% increase in conversion. Of course, the conversion was very low before, because there was no CTA. But just adding some elements or like for example, there was a button CTA and we added rich-text CTA, we add a text that explains what happens to you after you sign up. Tremendous results. So that’s one of the things that I think you do after you’ve launched it and after it’s working.

David Bain 

So sticking with your titles example there, obviously, you wouldn’t necessarily want to tweak all of your page titles and millions of pages at the same time. So do you just pick a certain number and then do a split test based upon that? And also, where do you take your ideas from? For example, do you use PPC as a way to actually build ideas for new page titles?

Anna Uss 

Fantastic question. So first of all, in order for it to be statistically correct, there is a certain amount of traffic and certain amount of pages that you need for that group. From the top of my mind, I think it was starting at 40/50k clicks. And for us, it would amount to a lot of pages because some of them get below 10 clicks a month, but it’s like the volume that drives that. And I would be inspired by the SERP. I would generally be inspired by PPC, but also by the SERP just Googling the keyword that the page ranks for seeing whether the titles in there trying to understand what is the user intent? And for example, should it be just the package name? Like, learn more about this package? Or should it be this package name health analysis, or this package name health analysis and maintenance, like trying to understand if people are searching for just the name? Or they’re searching for something specific around it? What are the results that Google shows and try to experiment there? And then it could be split testing? And I don’t I don’t think there is anything wrong in large volumes and like implementing it on one half of your site and the other half of the site. It’s fine. It’s just the learnings.

David Bain 

Obviously, the big win opportunities are potentially the page elements that are actually taken by Google and displayed in the SERP. So what trends are we seeing in terms of what is displayed in the SERP? And what are good areas within a page to be testing from a programmatic perspective?

Kevin Indig 

I think another one to add there is internal linking, that I think is absolutely critical, especially if you were talking about millions of pages, Google needs to find all of these pages and only XML sitemaps might not be enough, maybe initially for an initial crawl. But if you want Google to come back and re-crawl pages over and over, you need a very, like a very robust internal linking structure. And that is not super trivial. And then content is another one, right? If you have some some programmatic content that is not super valuable, sometimes need a bit more information for Google to differentiate all the different pages, trying to talk about a client without giving them away too much. But there’s a company that I’ve worked with that basically crawls the web through RSS feeds, and then provides price of a piece of content. And they can create those feeds for all sorts of topics. You can imagine anything from, you know, politics of finance, startups, SEO, marketing, you name it. And for them, it’s super easy to spin up millions of pages that Google can index with all of these different feeds of content. But then the key question is, how can you add more content insights, whether it’s structured or unstructured, to differentiate all these pages a bit more for Google to not just have this display and feed of content? So those are some of the things interlinking and the actual on page content that are found to be most impactful when we speak about millions of programmatic pages?

David Bain 

Mordy, what what are your thoughts?

Mordy Oberstein 

Well, I was gonna just go back on Kevin’s point really quick, it’s a really big problem, because indexing is getting much more difficult. Google is getting much pickier. And in the this, you know, the alignment of all these programmatic pages increases the chance of Google’s not going to crawl all of these pages. So doing things like differentiating, and even differentiating and headers, if you’re making some sort of clear differentiation. And making the page a bit clearer for Google to understand, could have enormous payoffs. Like, yeah, you’re there are things that you can do to add your really unique content to Google to see that this page is unique, and then crawl an index page. But sometimes it’s doing small little tweaks to the page can really be a big difference. And you’re working with millions of pages. It’s probably much easier to do that and see. Okay, great. Let’s see what happens now, do Google now index pages, great we’re done, but actually going through all those pages by hand and actually finding some really unique content to add. So as Anna mentioned before, you can really experiment, does this work? Try one thing, next thing does this work? Next thing does that work and take it step by step. But it’s going to be something that if you’re doing programmatic content at scale, I would imagine that your as time goes on is going to get harder and harder and harder for Google to decide I’m going to index all of that.

Kevin Indig 

I’d be curious to hear ends point It’s about crawling and indexing.

Anne Berlin 

Yeah, my brain had already gone to bringing up the very controversial subject of zero click search and how the widespread adoption of structured data in the manner we’ve just been discussing, could in depth positioning many of these sites that are our clients or our employers to have their content used without generating a click or transaction.

In terms of crawling and indexing. This, in my experience goes particularly in service of understanding the trends that are affecting coverage. So what kinds of patterns can we identify that are distinguishing between the programmatic pages that are getting indexed and aren’t. And for an effort like this at scale, you’ve got to do that in some structured manner. And then supporting internal linking? Again, you know, you mentioned this, Kevin, as a non-trivial issue. This is something I think that is the hardest struggle, and how to devise and implement and maintain a large scale internal linking strategy, particularly because internal linking is not a sexy topic, and can often be difficult to prioritize. I feel like this is one of the arguments that I bang my head against the wall when talking to clients about trying to do something to generate incremental traffic growth, and I say, let’s go back to fundamentals, how are we doing with internal linking health and that’s like, not fun.

David Bain 

Anne, in terms of the zero click subject, you started your thoughts with, is there anything that programmatic can do to identify which URLs are actually resulting in resulting well being displayed in the SERP but not resulting in any traffic at all and therefore actually doing things to perhaps try and increase the amount of brand presence in these results so that even though you’re not doing actually receive any traffic, you’re improving your brand notoriety as a result of the the content that people see.

Anne Berlin 

Like Anna, I’m an on-SERP-research devotee. So I think it’s really important to keep your eyeballs on what the SERP looks like in on the devices and in the settings in which our data shows our users are searching. I work for a publisher that the majority of their traffic comes from very short tail animal searches. And the SERPs have changed very dramatically with the introduction of VR, in particular. And I was just thinking yesterday looking at being a AI, and how the SERPs and Bing are evolving to display source credit in a way that has not traditionally been as visible in rich results in these types of SERPs. So I have somewhat of a pessimistic view on the viability of a brand visibility play, particularly on Google.

Mordy Oberstein 

That kind of really brings up the whole issue of you know, how you go about targeting longer tail or more, I’ll call them more nuanced kind of queries and nuanced kind of information with programmatic SEO. And is is the Google ecosystem sort of being steered in a way where they don’t really want you to win with longtail or they want the user to be pushed through there’s you go to the SERP everyone talks about all the different SERP features, but if you to me, if you go to the SERP, you’ll find, probably if you go through the page, three or four different ways that Google is trying to filter out the query, whether it be through people also ask or all these related searches are the bubble filters at the top of the server, broaden this search, refine this search, they literally call it refine this search. There’s all sorts of ways that SERP is all about Google trying to say we know you search for a short tail keyword, but you probably meant something longtail by it, let’s move you there.

Kevin Indig 

Yeah, I love that call out Mordy. I wholeheartedly agree with you that Google tries to get people to a very clear expression of what people want as quickly as possible. I do think though, that there’s, there’s a lot of, I think the most of the potential right now in SEO is probably an in longtail and midtail. And the reason is exactly that the short had Google basically fires all cannons at people when they look for something super short, right? Like all the sort of features, buy something, let’s search for it locally, navigate, here are some brands or you want something else I was like, you know, like a like a desperate drug dealer who just tries to get some stuff. I’m being pedantic here. But the point I want to make is that I think programmatic is a huge opportunity to target longtail searches that express a clear user intent. The challenge is that it’s much more difficult to measure, the more longtail you go, the more Google samples data in Search Console. So getting understanding what queries you rank for becomes a real problem.

Kevin Indig 

And then the other the other side of the coin is just understanding that because that is such a problem, right? It’s understanding what longtail keywords actually to target. So in some cases, you have to run a shotgun approach where you just launch a bunch of pages based on your best understanding of the target audience gut feeling, and then maybe some qualitative information like, you know, calls with sales or customer success or something like that. And then you see if those pages actually drive meaningful traffic and convert and then you scale them. So there’s this problem that more more the call out here is a critical fundamental one, there is an opportunity with programmatic SEO, but isn’t yet this clearly, streamlined, super straightforward approach is a bit of a leap of faith people have to take.

David Bain 

I feel that there’s many other angles and conversations that we can have in relation to programmatic SEO. And perhaps we could try and get the team together again, for part two, if possible, because I’m sure we can keep on going for a long time. But how about if we close up with one final, great tip from everyone about programmatic SEO? And maybe just a couple of ideas in relation to what type of tip you could share? Maybe what’s one thing that you haven’t really used programmatic for, but you intend to use programmatic for at some point over the coming year? And another thought is, how should you be working more closely with other marketing teams? Maybe paid search teams, maybe entirely different departments with programmatic SEO? How should you be educating them about programmatic SEO to better work more effectively, together? There are lots of different angles that you can take on this. But what would you share as a closing tip about programmatic SEO? Anne, shall we start off with you?

Anne Berlin 

So in terms of which other teams you should be working with, I’m thinking about on those websites that are primarily UGC. The issue that I’ve seen is there are not always sufficient controls around the structure and nature in which UGC content can be built. And so that becomes an issue at scale, that the data points you want to draw on for a programmatic play are only intermittently present. So trying to build better relationships with the folks who interact with your users, whether it is a client or whether it is a customer, and helping work with them to think through ways to modify the interface to get higher quality data out of UGC.

David Bain 

Superb, thanks Anne, Anna, what woud be your closing thoughts?

Anna Uss 

I have three closing points. The first one is, know your business and your audience and try to think out of the box. Don’t only think about high competitive keywords in your niche, obviously, you need to target them, but try to think about these other strategies they that can catch your audience somewhere where their interests are. Second thing is have something unique, a unique value you can add with your programmatic assets. And lastly, focus on indexing and focus on optimizing your assets structure, try to think about how will the user journey look and how you can interlink those pages. It can be HTML Sitemaps, which shouldn’t be like sitemap per se, it can be a directory, it can be grouped by a certain factor like popular, not popular, trending, maintained, not maintained, try to be creative about coming up with that factor that groups those pages and create a sitemap on the assets to net for Google to navigate through it. It can be related pages, and that can be also dynamic that pulls them in a certain pattern or manner. But to unlocking that indexing potential and moving from 300,000 pages indexed to 2 million pages indexed and constantly adding.

David Bain 

Mordy, can you beat three closing tips?

Mordy Oberstein 

Sure, three good ones are just like three. Okay. I’m gonna agree and say the team that you really want to work, the closest I think is going to be your your customer success team or customer support team. You know, you’re pulling in all this traffic, and it’s great. But is it really effective? What are the pain points of the user facing afterward? Because the point is not really to bring in the traffic. The point is then to figure out like, what’s working for your customer base? And is what you’re doing programmatically actually working and being effective. So I don’t think I’m saying anything novel on that. But I think it’s very, very important when you’re working at scale, to be able to say, okay, take a step back, and what’s actually working.

In terms of where I think things are going with all this, I think what you’re going to see, to go and kind of pivot back on the longtail conversation. I think where programmatic SEO has a hard time with is, programmatic content in general is where you have longtail keywords and the competition’s at the level of content out there is particularly nuanced, right? So you have a very niche blogs, or very niche websites talking in a very nuanced experience kind of way. And I think what you’re going to end up seeing is a mass adoption of AI. But I think that people who are going to do this effectively are going to use the AI as a foundation for creating the content other words, so I don’t need to hire five new writers, what I can do is I can have aI spin up the foundation of the content, and then I can take it from there. So it’s like almost like a, I hate to put it this way a glorified editor, kind of writer position. In other words, you’re gonna have the content, you could have the AI spin at the foundation of the content, and then be able to compete possibly be able to compete, it’s very nuanced niches, by supporting that foundation of content that you build programmatically with a little bit more of a hands on approach. So it’s sort of a quasi-programmatic application of AI, I think that might have a good chance of being very successful building out content at scale, and building on conduct dynamically and programmatically. It’ll be interesting to see if that really does compete with, you know, with niche specific websites. We’re doing this all quote unquote, by hand. But I think that’s where we’re heading off to. And I don’t know what that actually is going to look like, because I don’t know how the Google is going to handle that in the algorithm, but that’s where we’re gonna end up I think.

David Bain 

Quasi-programmatic. Love it. Kevin, what are your closing thoughts?

Kevin Indig 

Yeah, I want to build on top of what Mordy said, we’re now entering a phase where we see tools pop up that allow us to create programmatic content in minutes, not hours, or days or weeks. Minutes. And most of it at a decent quality were already there, there are tools out there that can do that for you relatively cheap. And the big bottleneck then is less the creation part, it’s much more how to how to keep the quality high. In my mind, it’s all about quality. And that is not something we can get set up with machines or automation. That is something that still has to happen with humans. And so the big question is, how do you when you can create 1,000s, 100,000s, maybe millions of pages relatively quickly and cheaply – how do you quality assure all that stuff? And so there’s some approaches where you sample or where you try to have, you know, mass fact checkers, and all that kind of stuff. But that in my mind is is a is a problem. We’ll solve that problem, I’m pretty confident that in the next probably, I don’t know, 24 months (Don’t quote me on that) but in the foreseeable future, we’ll be able to use some sort of large language model to make sure that information is at least 99%. factually correct. I think that’s a solvable problem. But until then, that’s going to be the major bottleneck. And then after that, the question is what’s what’s how to differentiate yourself? Is it going to be designed? Is it going to be unique insights, as we spoke about before, is it going to be some sort of, you know, scaling human input, right, like how do you attach an author to programmatic content is another problem that’s unsolved yet? So these are the problems that are see the foreseeable future and long term but I think what a lot of people don’t realize yet is that we are at a place where we can create a lot of content at a decent quality for very little money, and that changes the game tremendously in my mind.

David Bain 

I’ve been your horst David Bain. You’ve been listening to the Majestic SEO podcast with Anna. Anna, would you like to share with the listener where people can find out more about you?

Anna Uss 

I have a website called annauss.com. It’s Wix website. to check it out, but also on LinkedIn and on Twitter and Anna Uss, just find me.

David Bain 

Superb. Thanks for joining us.

Anne Berlin 

I run dark. But if you want to connect with me I’m most active in the Women in Tech SEO Slack community.

David Bain 

Superb. Thank you Anne, and Mordy.

Mordy Oberstein 

I’m on Twitter, @MordyOberstein. So I would just go to Barry Schwartz his Twitter feed and you can see me commenting and harassing him most of the day.

David Bain 

Great stuff and Kevin.

Kevin Indig 

kevin-indig.com is where you find everything about me.

David Bain 
Lovely stuff. Well, thanks so much for everyone that panelists, viewers, listener. If you want to join us next time, sign up at majestic.com/webinars and of course check out our other series at SEOin2023.com.

Previous Episodes

Follow our Twitter account @Majestic to hear about more upcoming episodes!

Or if you want to catch up with all of our previous episodes, check out the full list of our Majestic SEO Podcast episodes.

Comments

Comments are closed.