A Guide to SEO in the Age of AI, Plus Modern Best Practices

• 

  Justin DunhamPartner
  Ércule
• 

  Ashley McClellandPartner
  Ércule

How Will Generative AI Change SEO as We Know It?

Search engine optimization, the art of optimizing your startup’s website and content to “be found” by search engines such as Google and Bing, has traditionally been one of the most efficient, cost-effective ways to help startups grow their businesses. Research has suggested that more than 50% of all website traffic comes from search engines, with an estimated 90%+ of global traffic sourced from Google Search, maps, and images. Or at least, that’s how it used to be. Search giant Google is applying its research in machine learning to what could be the next great leap forward for internet search: “Search Generative Experience” (SGE). SGE will use the company’s generative AI models to “enhance” search engine results pages (SERP)–in some cases, completely replacing the first page of SERP with AI-generated text, rather than the familiar list of links to other websites.

What do the coming changes mean for startups, and what role should AI tools such as large-language models (LLMs) play in content generation (if any)? We consulted with Ashley McClelland and Justin Dunham of content performance agency ércule, who shared topline insights into standard SEO best practices, as well as a practical, forward-looking guide to the future of SEO and content distribution in the age of GenAI.

Common Mistakes: “Pet Projects,” Intentionality, Quantity vs. Quality

• Avoid Focusing on Technical “Pet Projects”: Projects that are interesting to you may not be of interest or seen as valuable by prospects.
• Intentionally Plan Content for SEO Performance: Rather than retroactively adding a few SEO nuances post-writing, a best practice is to plan your SEO strategy and anticipate your readers’ intent before writing your next article.
• Consider Fewer, Higher-Quality Posts to Start: Rather than spinning up many content pieces that may or may not be relevant, consider a small handful of highly polished articles that address your ideal customer profile (ICP), and whose topics you can repurpose into other channels, such as social media, podcasts, and others.

The ércule team discusses some of the more-common mistakes early-stage founders make. “Number one, they’ll sometimes assume that things they write will be interesting to other people because they're interesting to them. And so you end up with a lot of very specific posts about fun side projects–which, don’t get me wrong, can be great things to write. It can be very important to write about other projects that are meaningful to your startup and help differentiate you. But it probably won't perform in search. Search is about user intent—which Google has always interpreted as questions, which people type in as their search queries, for which they seek answers. So if you don't write something that aligns with the search intent of your potential website visitors–that is, the actual question that real human beings out there are hoping to answer, then it's going to be a lot harder for it to rank.”

The ércule team also suggests that a common misconception is that content marketing and writing are the same thing. “Content marketing and SEO are completely different from just plain old writing. When you’re handling things correctly, there's a ton of other activities that should be happening, of which the writing of the post is actually the very last thing that you do. For example, we’ll hear founders say, ‘Oh, we'll write some stuff, and we'll SEO it up, and put it up on the blog, and get some traffic.’ But that doesn’t really work. You can’t just apply SEO like a coat of paint at the end. You have to design your content for performance.”

A third common mistake is founders rushing into content before first doing the important work of determining their ICP, their positioning, and overall messaging to understand which personas are a relevant part of their audiences, and which aren’t. The team suggests that some startups may put too much emphasis on tactical stuff like keyword research, and not enough on strategic stuff like researching relevant content that genuinely matters to potential customers.

Taking an intentional approach is much better than just writing a bunch of articles and seeing what sticks.” - Justin Dunham, Partner / ércule

In other words, having a data-driven approach to content isn’t a bad thing—but it isn’t everything. “Even if you manage to get organic traffic, there's no guarantee whatsoever that you can convert any of those people into customers. So our advice for founders or early-stage folks is to think about producing a very small number of pieces of content. High-quality stuff, posted with some consistency–maybe one post a month or something like that. And think about things that you can repurpose. So we'd much rather see startups write one good piece that they can, without a lot of extra effort, spin into two or three other pieces on other channels. Everything you produce should be repurposable, it should be high quality, it should answer the implicit question in the search queries you’re targeting, and it should also be useful on your other channels.”

“Taking an intentional approach is much better than just writing a bunch of articles and seeing what sticks. Here’s the thing: You might write something you think could work for organic search, but if it’s of high enough quality and relevant to your audiences, it could also be something that you can hand off to a prospect. So in addition to potentially ranking with your insightful blog on CI/CD, you can also approach a prospect with: ‘Hey, we just wrote this guide for best practices in CI/CD and thought you might like to take a look’ as a follow-up asset. It’s something you can also give to a sales rep or something you can put on LinkedIn or other social feeds.”

More Resources: Learn More About How to Identify Relevant Personas for Content:

• Article - Dear Founders: Start with Messaging by Dana Oshiro
• Video - Positioning: Winning the Battle for the Developer’s Mind with Geva Perry

Website Optimization is Table Stakes, Quality Beats Algorithms

• Quality Content Matters: Google’s primary focus is surfacing content that shows experience, expertise, authoritativeness, and trustworthiness.
• Website Optimization Is Now Table Stakes: Website redirects, load speed, and architecture are basic concerns to be managed to a functional level–then put on the back burner in favor of better content.
• Algorithms Take a Back Seat to Content: Similarly, Google’s wonderful zoo full of pandas and penguins and other algorithms are less pressing than high-quality content.

The ércule team suggests that the biggest change in the SEO industry may simply be its focus on content quality–including whether what you produce reflects Google’s stated goal to prioritize content that demonstrates experience, expertise, authoritativeness, and trustworthiness (EEAT). “We continue to see that SEO is more and more about content quality and really reaching your audience–and knowing what they want to read. It's also much more about distribution of your content on other channels.”

The team suggests that technical website work–such as implementing 301 redirects when updating URLs or optimizing page speed by minimizing image sizes and in-page assets–are no longer the significant differentiators they might have been. “Getting your site’s various web pages in a great technical place has become table stakes at this point.”

“All the changes that Google is making are really just driving higher-quality content to the top. And because higher-quality content is really hard to produce at scale, it means that smaller companies need to be more and more focused on what their audience specifically wants to read. Which means that if we were founders right now, we would spend zero minutes thinking about Google’s various algorithm updates. We’d spend zero minutes doing technical hacking of our startup’s website as long as it was in good shape. We would be thinking about what our prospects want to read–what are the questions we can answer that align with some search intent they have.”

More Resources: Learn More About Content Strategy for Startups:

• Article - How Successful DevTool Companies Manage Content Strategies ft. Atlassian, CircleCI, Retool, Orbit, LaunchDarkly
• Article - Thought Leadership for Technical Founders by Karl Hanson

What to Know About Google Analytics 4

• GA3 Is Off. Migrate to GA4 Immediately: GA3 is no longer collecting data–if your startup has a website, and you’re not currently running a GA4 instance, sign up immediately.
• GA4 Covers the Basics, But May Become More Useful as Your Business Grows: GA4 should still track basic website data (sessions, bounce rates, etc.) but can also be useful for activities such as website integrations and Google Ads.
• Google Tag Manager Gives You More Control to Track Custom Activities: Though optional, Tag Manager gives website owners the ability to track a variety of custom activities across their site without having to make code changes.

The search giant has finally pulled the plug on Google Analytics 3, the web analytics tool that everyone used to use for their website. Officially, GA3 was to stop collecting traffic data as of July 1, 2023, to “encourage” users to migrate to Google Analytics 4 and begin collecting new website datasets within the new tool. GA4 is admittedly a relatively new product with different nuances to be aware of, but it should offer some of the same key measurements as its predecessor.

“Despite the announced July 1 date, we noticed that many of our clients had their universal analytics properties continue to collect data for a while–but this is at an end. So anyone reading this absolutely needs to migrate immediately. But overall, when using GA4, we’ll still want to look at conventional metrics, such as web sessions, and maybe bounce rates, and what channels users are coming from. Those fundamentals are the same–a little harder to find in GA4–but they're all still there.”

“Another thing to consider is that GA4 is a much more powerful tool in GA3 in a lot of important ways: Much more like a BI tool for your web analytics. Over time, it has a lot of power that you can use to measure different events and other things happening on your website, as well as things within your product.”

For people who aren’t fans of GA4 so far, we have some bad news...it’s very likely you’ll find yourself having to come back to using GA4 for some other integration or use case, such as Google Ads.” - Justin Dunham, Partner / ércule

“Honestly, for people who aren’t fans of GA4 so far, we have some bad news–even if you try to use completely different products now, it’s very likely you’ll find yourself having to come back to using GA4 for some other integration or use case, such as Google Ads–which works a lot better when there's a GA4 instance. And if you haven’t already been running GA4, you’re going to have to spin up a new instance that will have no data in it. You really do want to run GA4, even if you never touch it, because at some point, as your company grows and you need your website to do more work for you, you're going to need GA4 for integrations, Google ads, digital marketing...it's just going to happen.”

“So that's the bad news. However, there are a lot of really good tools that you can use alongside GA4 to build more-complex funnels and tracking. One example is the open-source product analytics tool PostHog, or more-conventional tools such as MixPanel.

“Thankfully, installing GA4 is certainly easier than other technical processes. Essentially, these are the steps: First, you're going to go to analytics.google.com site and sign up for GA4. The website will give you a JavaScript snippet to install on your website–basically adding a JavaScript Pixel. While you’re not required to use it, our team really likes Google Tag Manager, which gives you a lot more control over tracking conversions and other activities on your website. It lets you manage analytics tags without touching the code on your site.”

“So if you want to instrument an event, like a signup, or if you decide to experiment with Google Ads, or if you decide to experiment with a heat-mapping tool or something else, you can go into the Tag Manager UI, add what you need, and then get it deployed. If something goes wrong, you can also roll it back really easily. Installing Tag Manager is admittedly an extra step, and may not be a hard requirement when you’re still setting up your business, but it’s very, very helpful. From there, you log into GA4, and it should tell you that it's collecting data.”

“From there, where it gets more complicated is if you decide that you want to add custom data somewhere, or add custom dimensions around how old this blog post was that a prospect was looking at, what category it was in, or instrumenting conversions, or generating custom reports on how many signups your website is getting. And the ércule team is always happy to help with that sort of thing.”

More Resources: Learn About the Technical Side of SEO, Google, and Website Optimization

• Article - Content and SEO in the Early Days: A Founder’s Guide to Blogging by Erik Dietrich
• Article - Doubling Down on Content: Q&A with Mitch Wainer
• Article - When Will I Start to See Traffic? A Field Guide for Managing Expectations in Content Marketing

How to Support Category Creation with SEO

• When Creating a Variation on a Category, Don’t Shy Away From the Original: An important part of winning a new category variant is becoming an established name in the original category (and then explaining why your variation is better).
• Creating a Net-New Category Will Cost You: Sometimes, creating a totally new category is absolutely the right play–but it will require significant investment into your content, brand, and amplifying what you do.

Startups frequently face the challenge of how to define their products relative to existing product categories. Often, startups either:

1. Provide a product in a new category
2. Provide a product within an established category and look to differentiate themselves, or
3. Provide a variation on an established category

The ércule team notes that many early-stage startups may find themselves in the third situation, and shares its recommendations. “One thing we commonly advise startups to do is not be afraid of the category that they’re trying to differentiate from. Instead, you can bring clarity to what you’re trying to do by looking at the research around that category and pinning yourself to its most valuable aspects, but also positioning against it in a meaningful way.”

“To create a category, you need something to differentiate from, but you also want to be known in that category from which you're differentiating yourself. If you want to build search traffic and you want to do a new category, the best way to do that is to build search traffic in the old category and talk about why your new category is a better version of that old category.”

You can bring clarity to what you’re trying to do by looking at the research around that category and pinning yourself to its most valuable aspects, but also positioning against it in a meaningful way.” - Ashley McClelland, Partner / ércule

“Another thing to mention is that while people frequently talk about net-new ‘category creation,’ they should be advised that it’s really, really hard and really expensive. You have to really have to roll that out across everything you do. Your content has to change. You probably need to run an event around the new category. You need to have things to say to analysts around the new category. You have to spend time on PR, you have to educate people on what your new category is.

That’s not to say that no one should ever try to create a new category. You might have a great idea, especially for companies that are at the right time and place, but just be aware that investing in category creation is a pretty significant bet. It’s definitely not the sort of thing you can consistently accomplish with small, inexpensive experiments here and there.”

More Resources: Positioning Yourself Within, or Against, Established Categories

• Video - Category Leader Essentials with Netlify, Slack, DataStax
• Article - Linkerd: A Case Study in Developer Tools Category Creation by Ashley Dotterweich
• Video: Early Stage Positioning for B2B Startups with Justin Warren

Preparing for AI and Google’s Search Generative Experience (SGE)

• Google’s New AI Results Will Fundamentally Change SEO, But Likely Not for Tech Startups: Google’s SGE will likely start replacing website search results with its own AI-generated content primarily for common search queries such as popular cooking recipes or common tasks.
• SGE Is Unlikely to Affect Search Results for Highly Specialized Technical Content for Some Time: Given how challenging it can be to properly train AI models within an abundance of data to “feed” them, it’s unlikely that Google will display its AI search results for any kind of highly specific technical specs–including the kind of specialized solutions that many tech startups focus on.
• For Early-Stage Startups, Content Should Be an Exercise in Focus: It’s probably not the best idea to generate thousands of low-quality blog articles at an early stage. Instead, intentionally focusing on specific topics and categories, and treating content like product launches, may bear the most fruit.

How should startups prepare for Google SGE, the search giant’s AI-powered search experience, which is set to start returning AI-generated results in response to certain broad queries? While SGE made headlines when Google revealed the new tool’s functionality may replace the entire first page of SERP on mobile with generated results, it’s unlikely that the new system will do so for anything but the most generic of queries–such as basic cooking recipes (such as “Best chocolate chip cookie recipe”) or how-to instructions for mundane, day-to-day tasks (“How to change a flat tire”). For highly specialized technical content, Google will likely not rely on LLMs for content creation, and instead continue to defer to specialized websites–including those of ambitious startup founders–for the foreseeable future.

But what does generative AI mean for content creation? With the ability of GPT-style tools to rapidly spin up content, are we in for a new online golden age, where all the content online is of immaculate quality crafted by artificial intelligence? Maybe not, suggests the ércule team.

“The results we’ve seen from people attempting to produce highly technical product content using generative AI tools like ChatGPT, Bard, and other chatbots has generally been of really low quality. We’ve spoken to founders who are concerned about having to compete against GPT-created content in SEO rankings for highly technical terms such as ‘DevOps’ or ‘cloud containerization.’ They ask us, ‘Why should I spend a bunch of time writing an article about this if I'm just going to get swamped by AI-generated content by other companies?’”

As it turns out, using artificial intelligence as a kind of automation solution purely for content creation isn’t really working, or providing a great user experience. “The point is that the quality of AI-generated content for those topics is relatively poor. That's not the case for, let’s say, a recipe website or something else like that where AI-generated content is more of a threat. In other words, for technical founders and marketers at early-stage companies, AI-generated content is probably not something they have to worry about yet. Content that covers highly commoditized topics–like a basic cooking recipe–seems much more likely to be the kind of thing that Google’s SGE is compiling as datasets to train itself on and serves results for. Content that discusses highly nuanced technical details is far less likely to be returned by AI for some time.”

Today’s artificial intelligence tools suffer from numerous well-documented issues, including hallucinations, biases, and other issues that may prevent them from being foolproof content generation engines. “For startups in this space, we’re hesitant to even recommend any specific AI tools for SEO, because for technical content built around technical products, credibility is so, so important. And today’s generative AI doesn't give you credibility on this stuff.”

“Where we’ve seen generative AI be most helpful is for companies that are generating content at scale and have a process in place to use AI for production—for example, e-commerce businesses may find it useful to use GenAI for extremely basic stuff such as product descriptions, . To be clear, we’re talking about less-critical content that somebody still looks at, but that they don't write from scratch. Another place that we're seeing our clients use AI, which might be more interesting, is programmatic campaigns, in which AI generates pages on your website for you. And generative AI can also be OK as a way of generating the very first draft of a blog post, provided that the outline was done by a human. So you need the outline to be done by a human, or the first draft to be done by a human. From there, as a founder, the job of you and the content creators on your team is to go through what has been generated and add specific examples, points from your own experience, statistics that you have, code samples, all the things that make an article credible, relevant, and differentiated.”

But shouldn’t founders take advantage of the wondrous AI technology of OpenAI, Meta, Microsoft, and other chatbot vendors to spin up massive quantities of content in confident-sounding natural language? “Where we don’t see GenAI as super useful–and this could come back to bite us eventually–is as an important time investment for early-stage founders to generate thousands of blog posts. We just don't see it as useful for end-to-end writing of this kind of content. There are creative ways to use GenAI that can be beneficial, but to make content that will move the needle, you really have to focus your workflow on creating quality over quantity.”

Will the AI revolution change the fundamentals of SEO? That depends on your definition of what SEO fundamentals are. The ércule team suggests, “If you think the fundamentals of SEO are backlinks and low-quality content and the other, somewhat underhanded things that SEOs used to do 20 years ago, then, yeah, AI will change the fundamentals. But we think the fundamentals are really high-quality content that's distributed in a lot of places–which isn’t all that different from traditional SEO.”

“In terms of distribution–for instance, let’s say we write something on our blog. But if we can also get it in front of people on our social media channels, we can get it in front of people who are hopefully going to talk about it in an interview. From there, we can cover the topic on a podcast, and stuff like that. That's really going to help us. So for today’s SEO, and possibly tomorrow’s, what matters is distribution, understanding your audience, and making sure you’re creating something they really want.”

If you think the fundamentals of SEO are backlinks and low-quality content and the other, somewhat underhanded things that SEOs used to do 20 years ago, then, yeah, AI will change the fundamentals.” - Justin Dunham, Partner / ércule

“SGE and all these generative experiences from various LLMs are going to probably tend to summarize the top results, at least for a while. Those have always been the results that Google ‘thinks’ are the ‘best’ or ‘most relevant.’ So, if we're in those top results, that's where things become even more important...because it's really easy to rank on page five or six of Google search results for a bunch of topics. But it's harder and also much more valuable to rank much higher for a smaller number of topics.”

“So again, we come back to focusing on who the people are who are reading your article, and thinking about being at that early stage for building out your content, and really focusing on a limited number of topics, and ranking for those things. That is where we see things being picked up by AI.”

How will AI change day-to-day SEO in the future? “In some ways, it’s a new toolkit for everybody to use. It certainly makes our jobs easier in terms of analyzing content and making recommendations. It potentially makes writing a bit easier. But we still think that SEO will come down to high-quality content that is well distributed, not stuff we put on our blog, and then forgot to tell people about. Stuff that we're revisiting and tending to and paying attention to. That sort of thing isn’t going to change all that much.”

“Here’s a metaphor we use when we talk to early-stage startups: A good way to think about your content, and your marketing in general is not that you’re writing a blog post to get signups out of it. Instead, you’re building another product, and you’ll need to have the same sort of patience that process requires–but with all the potential return on investment if you do it well.”

Your Content Distribution Strategy

• Early-Stage Startups Should Focus on a Few Channels: One channel is your website’s blog. In the interest of focusing on getting results, it’s a good idea to think about zeroing in on one other channel aside from your blog.
• Find Channels Where Your Customers Live. Learn How They Work: Your best bet is to go where your ideal customers already are and understand whom and what they listen to–and how to successfully drive engagement in that channel.

Once you’ve figured out how AI fits into your SEO strategy, and how you’ll approach creating content, where should you put it? The ércule team recommends focusing on the channels that make sense for your audience.

“You’ll want to choose the right channel for your strategy. For example, businesses targeting enterprise buyers will likely find LinkedIn to be the place to be. For finding developers, there are a lot of places where people we work with are seeing success. For example, private Slack communities could be fruitful–if you’re a founder and you're part of a Slack community that relates to your product, that's a great source for content ideas, not just for promotion. People are asking questions and you can answer them–and from time to time, maybe you can answer those questions with a link back to something that you wrote. There are many communities out there for developers–a few examples are Hashnode and dev.to, and Reddit can be a good place to get traction.”

“Regardless of which specific channel(s) you choose, as a founder, you're probably only going to be able to do one or two channels really well. And one of those channels is probably going to be your blog on your website. So it's worth experimenting with a number of different ones to see which one fits. Though there are alternatives to consider such as podcasts. Producing your own is definitely an investment!”

“Regardless, content distribution can be a slow burn that requires you to be consistent and disciplined. Just like with a new software product, you might not expect to launch and get thousands of signups in just the first month or two. After you launch your product, you need to do a bunch of work around marketing to get users and engagement. Again, in many ways, it's just like launching and marketing a new product. So, the right way is to do some experiments.”

“An important experiment to consider for distribution is understanding who your audiences are, and how they are segmented. Which helps you figure out which channels map best to those segments. For every community, there are going to be certain things that are considered to be of higher value than others. And you get to know that by asking people you think are in your target audience. You figure out where they hang out, what and whom they listen to, and where they get their information. And you might learn that there's this one Python podcast that everybody who's into data engineering listens to. Those are the finds that can be like gold for you. Now, not every audience has a segment like that with a great channel, but if you do enough research, a lot of times you can come up with at least a few avenues to explore that are more likely to get you traction because your audience already cares about them. In some cases, it can be valuable to get that granular, instead of just trying everything at a high level and not necessarily seeing results.”

“Finally, when considering your choice of channel, it can also be valuable to consider the nuances of that channel and how to succeed there. For example, LinkedIn’s algorithms seem to have a tendency to reward users who are early adopters of new features. It also seems to reward users who exhibit their preferred behavior, such as not including a link in your initial post, or ensuring you engage with other users’ posts.”

More Resources: How to Amplify Your Content With Smart Distribution

• Video: Using Data To Create Content That Spreads with Rohin Dar
• Article: Building Contentful’s Global Advocacy Program by Ted Carstensen
• Video: Building A Content Marketing Machine with Hiten Shah

How to Make the Most of a Potentially AI-Driven SEO Landscape

• Content Is a Cost-Efficient Tactic for Startups That’s Hard to Avoid: The outsize value that high-quality content provides probably outweighs the admittedly annoying prospect of Google and other AI vendors using your content to train their models.
• Google’s SGE and Other AI Vendors Training on Your Content Could Be a Win: Google and other top vendors will likely train their AI models using the best content in the space. If the future of content will be dominated by AI-generated results, it’s probably best to have your message and your content be what dominates those results.

So the future of web content may be that Google’s AI algorithm gobbles everything up and that it surfaces as much of its own AI-generated content as results as it possibly can. Does this mean that content is done for? Does this mean startup founders should abandon ship and try other channels? The ércule team doesn’t think so.

“It can be frustrating to think that you’re devoting all this time and effort to creating high-quality content that Google is just going to feed to its AI algorithm and spit out itself. And there are, of course, other ways to go to market–you can hire a fleet of salespeople, you can avoid organic and go the paid route, and invest significantly in costly digital advertising, for instance. But for early-stage startups who are still wondering whether content marketing is a good strategy for them, we’d gently point out that going with a purely paid strategy or a purely sales-led approach can be too expensive for startups to try at the very beginning.”

For founders who are worried about Google, we recommend that, once you have high-quality content that’s worth distributing, you then distribute that content as much as you can.” - Justin Dunham, Partner / ércule

“At some point, you probably will end up doing some kind of content. Whether that's having a blog you purposely build for marketing, or whether that’s great documentation for your product or some other thing, you're still going to be doing content of some kind. So there isn't really a way around that. The best advice we can give is to produce content that doesn't just rely on Google to be successful. You’re going to want to produce content that your audience really wants to read. Which means you can distribute it successfully. Yes, on Google, but also in your newsletter, and in your nurture tracks when people sign up for email updates, in updates to your investors so they can send it to all of their friends and portfolio companies. So for founders who are worried about Google, we recommend that, once you have high-quality content that’s worth distributing, you then distribute that content as much as you can.”

“What if Google takes your content and turns it into AI training data fodder? First of all, that’s a wonderful thing if it happens because it means that your content was one of the top results–which again, is what Google will presumably use to inform its AI algorithms. And again, it’s not all that likely to happen anytime soon because of the complexity of the content. But the way around that risk of Google returning AI search results powered by your content instead of links to your website, if it is a ‘risk,’ is to produce content that you're using in lots of different places.”

“In terms of blocking your website to prevent AIs from scraping your website’s content to train them, we’re not sure we’d recommend that kind of thing as a priority for an early-stage startup founder. Instead, it’s probably a better idea to focus on producing the actual content. It’s not necessarily a great use of time to worry about controlling exactly how it gets used or picked up by anybody–because frankly, if it does, that’s potentially a win. What if Google serves the same AI-generated search result to everyone, and that result contains your message? Maybe Google’s AI search results will effectively become the new featured snippet.”