Gameplanning the SGE Apocalypse

Lately, I've found myself fielding client questions related to Google's creatively named "Search Generative Experience (SGE)". This is hardly surprising, given that it uses generative AI, and generative AI has forced Gartner to redraw its hype cycle curve with a peak of inflated expectations that now extends about 50 miles into space. (I cringe to think what the trough of disillusionment is going to look like.)

In SEO circles, this has caused an absolute frenzy of panic and predictions, given that, like voice search, mobile, and other things before it, SGE poses a potentially existential threat to traditional SEO tactics. Outside of this tight band of technocratic minutiae, the technology probably just inspires some curiosity.

If you're not familiar, this is what we're talking about:

When you execute a Google search, the SGE experience attempts to answer your question right in the browser rather than simply serving up other results. Anyone who googles something like "chuck norris age" knows that this concept has existed for a while. But SGE expands it and... say the line, Bart... uses generative AI!

There are two questions that people are asking me:

  1. Do you think SGE is going to replace traditional search results?

  2. What should we do if it does?

And here are my answers, respectively:

  1. I don't know, and neither does anyone else.

  2. This is the question that actually matters, so I'm writing this post to answer it.

Why I Don't Know (Or Care): A Brief Note About Predictions

As an aspiring retired edgelord, I have a lot of contrarian takes. One of those is my opinion about predicting the future:

I think attempting to predict the future is a fundamentally silly and dangerous activity, and I try to avoid situations where my fate depends on guessing at things.

If you're interested in more detail about this, I'll elaborate in an addendum at the end of the post. But for now, let me explain it topically.

At the time of writing, an awful lot of people are wagering an awful lot of money on figuring out whether the Chiefs or 49ers are going to win a football game. They are, in fact, placing money on their ability to predict the future, based on their knowledge of football.

This weekend, some of them will win and some of them will lose. On a long timeline of betting on their ability to predict football games, they will all lose.

Who wins? The sports books. The ones who say, "I don't know or care who is going to win the football game, but I know how to make money either way."

In this post, I'm inviting you to leave the pundits to their professional buffoonery and join me in being the house.

Suddenly Missing Search Traffic: The Thought Exercise

I don't know to what extent SGE is going to displace traditional search and when. But that doesn't mean I can't prepare for what I would do.

So let's do a thought exercise. Let's assume that SGE is a wild success and that Google gambles its entire ad revenue business on replacing search results with direct question answering. This means that your search engine traffic drops to zero tomorrow.

What do you do?

Let's prepare for that eventuality. Let's assume that you suddenly have a deficit of 100K site visitors and 1K leads per month that you need to scramble to recover. If you prepare for that, then the more probable reality of a fractional-to-marginal traffic loss will barely register as a hiccup.

Here's how I'd do that.

1. I Wouldn't Worry About SERP Scraps

A lot of the current focus in SEO seems to be figuring out how to mimic Wikipedia and Gitlab in this screenshot. This makes sense, since it follows the general pattern of competing for featured snippets and other augmented search features that Google has added over the years. And SEOs have a rich history of speculative tea-leaf reading, since the entire industry is essentially a cargo cult, waiting for Google (and Bing, a little) to bless them with provisions.

But up until this point, all of these SERP features have existed to enhance the experience of directing searchers to results, except in trivial cases like Chuck Norris's age. The fundamental paradigm has not changed: searchers ask questions and Google helps them find third-party answers.

Except, the (hypothetical) SGE apocalypse now does change the game. In this scenario, Google is getting into the business of replacing its results with direct answers and competing with its platform partners in an unprecedented way. And given that Google is announcing loud and clear its intention to deemphasize third-party content, I wouldn't create a contingency plan that relies on appearing in these leftover scraps of search results.

(For anyone pointing out that I'm making a prediction, I'd argue that I'm actually doing a bit of subtle game theory here. The odds of you being able to figure out and execute whatever causes Google to give you the scraps are low, and the effort involved will be high, so I wouldn't play this hand until non-speculative evidence of its efficacy emerged.)

2. Find Whatever Replaces Google for Serving Human Content

I don't believe the world is universally enamored of ChatGPT as an answer to all questions it might have. In fact, the current zeitgeist rather reminds me of the emergence of Amazon's Alexa and the temporary proclamations that text search engines were over, since everyone would just voice search everything from now on.

In both cases, it seems likely that a decent cross section of humanity will continue to seek out human answers to their questions. And if that's true and Google stops providing those answers, something else will emerge to pair human-written results with user questions.

So the first part of my contingency plan would be to keep an eye out for that. Is anyone developing and advertising a contra search engine for Luddites? Are new Q&A paradigms and sites emerging?

If they are, I'd want them in my rolodex ahead of the SGE apocalypse.

3. Syndicate Existing Organic Content En Masse

I haven't looked at their media kit lately, but I imagine that dev.to has monthly site visits in the tens of millions. Some of that obviously comes from search, but they also have a prodigious amount of direct navigation.

Wouldn't you like to get in front of those folks?

At the time of writing, I've posted 168 articles there, mostly through syndication. That nominal effort, with posts not specifically intended for the platform or promoted at all, earned me about 5.8K followers and 50K post views.

Imagine what would happen if you created an organization page and syndicated your entire content corpus there. If this content is specifically aimed at developers (and not a hodgepodge of stuff about content marketing and management consulting), the traffic figures to be substantial. And you could achieve this with a marketing intern and a couple of person-days of elbow grease.

Of course, this doesn't have to be dev.to specifically. You can do this on multiple sites and pick ones that have community members whose interests align with your content.

And remember, in this hypothetical world, Google has removed our ability to use the internet search feature to consume non-AI content. So these sites will likely see massive upticks in use of their internal search functionality.

Marketing automation and measurement isn't as seamless as on your own site, but referred traffic from these sites will qualify as leads at a much higher rate when they click through to your site.

4. Repurpose to Q&A Sites and Communities

Content that does well in search is generally Q&A-style content: What does "DevOps" mean? How do you get started with Python and MySQL?

Whether you think of it this way or not, if you have 100K search visitors per month, you've answered a lot of questions that people are asking.

So go find where else they're asking these same questions.

This is a little different from my point two in that, there, I'm calling for participation in a hypothetical search engine alternative. In this example, I'm suggesting that you build a profile on Q&A sites and in communities by using your existing content to answer questions in medium-appropriate ways.

This will require some trimming and modification. You're not going to answer a StackOverflow question with a 3,000-word pillar post without getting downvoted into oblivion. But you can certainly extract snippets and feature links back to your content where appropriate.

5. Create Email Courses and Move Earlier in the Buyer Journey

Atlassian's "Agile Coach" is a huge organic traffic earner and a great way to structure organic content. So if Google suddenly withdrew 100% of the traffic that it earns, I'd personally think about reconstituting it as an email course to promote and make available to aspiring agile coaches.

Call it "Agile Coach in 45 Days" or something and drip it out for free to anyone wanting to sign up. You could then promote it with ads, partnerships, and distribution in communities or Q&A sites.

Anyone embarking on this journey is quite likely to be a qualified lead as a JIRA user at some point, if not now. Thus you wouldn't need to convert all 100K (or whatever) monthly visitors to email readers—just the 1K subset of qualified leads.

6. Create Distributable Marketing Assets for Authority

Atlassian could also turn their Agile Coach content corpus into a book. It's dead simple to turn material like that into an ebook in the major stores, and it's not even especially difficult to generate a print book. (Hit Subscribe does both of these things for clients at times.)

For a brand, I'd recommend making the book free in as many places as possible. And when you have to sell it due to store requirements, you can just very publicly donate all proceeds to charity.

The idea here is to have written the book on being an agile coach (or whatever applies to your situation). Until Amazon replaces all books with Alexa making things up, you can count on their distribution network to put the book in front of people.

This is doubly true if you run a successful enough launch to earn a best seller in a category, which isn't as hard as you might think. I managed to do this with a book once, years ago, and it still earns me a healthy supply of beer money even years later, with zero promotion effort.

What About New, Pre-Apocalypse Content?

Up to this point, everything I covered answers the question, "What would I do if I had an existing traffic stream that went away?" I should also address the question of what creating new content might look like, both now, as a hedge, and post-SGE-apocalypse.

I'll address the latter first, since it's easier to comment on. Once the apocalypse happens, you'll have much better data on what media and distribution channels work best, so I'd cross that bridge when you come to it.

The problem of hedging today against a hypothetical apocalypse is a bit more interesting. Given that organic search is such a CAC-effective, easy distribution channel, I wouldn't recommend deemphasizing it at all based on speculation. Instead, I'd recommend hedging bets by doing things that would make the recovery methods that I outlined easier to implement if and when the time comes.

For instance:

  • Create thematic corpuses of content (e.g., the Agile Coach, content glossaries, tutorial exchanges, etc.), rather than treating your blog as an episodic journal.

  • Establish and measure other distribution channels besides the search engine.

  • Cultivate efficiency at repurposing your organic content for other media.

  • Start measuring direct traffic and branded searches and treating them as KPIs. (For instance, "atlassian agile coach" earns 170 global searches per month, indicating that people are looking for that content and it would continue to earn post-apocalypse direct visitors).

Relax and Take Deep Breaths If You're Worried

Let me wrap by hopefully calming your frayed nerves a bit, if you're concerned about traffic loss or the viability of search as a channel. Here is the fundamental question that I'd ask myself:

Does it seem likely that everyone in the world would soon rather have generative AI answering all of their questions than humans?

If you think the answer to that is "yes," then I'd recommend going straight to ChatGPT, asking it to type you up a GTM strategy, and immediately and unquestioningly executing whatever it spits out. If, on the other hand, you think that people might still want to hear from one another on occasion, then you can relax.

All search engines really are, at the end of the day, is Q&A writ large. And all SEO really is at the end of the day is figuring out what "Qs" your audience has and "A-ing" them (on a website that doesn't suck to visit). So as long as they have questions and are looking for helpful, human answers, getting those answers in front of them is just a tactical detail that we'll be able to figure out for you.

Addendum: My Take on Predictions in More Detail

Over the last number of years, I've started to view predicting the future as an increasingly fraught and often self-sabotaging activity for those that do it.

At their most harmless, future predictions tend to take the form of empty conversational calories: half-drunk uncles at Thanksgiving arguing about who will win the Super Bowl or podcasters predicting what social media will be "big" next year. Everyone makes up nonsense in an accountability-free zone and then promptly and mercifully forgets about all of their past predictions, save the occasional one that actually comes true.

But often, those situations resolve to less harmless states. Uncle Steve loses his 401K being wrong about the Lions this year, or a company folds because of a CMO hunch that Google Plus is the future of GTM strategy. These things are what happens when people become too attached to their own predictions. Guesses evolve into suckers' bets.

I'm old enough that I watched the rise of the agile movement against the incumbent "waterfall" methodology. Waterfall was predicated upon the fundamental premise that if you just practiced enough, did enough analysis, isolated enough variables, and did enough squinting, you could predict the exact future costs of eight-figure software projects. Agile's fundamental premise, on the other hand was, "Nah, that's dumb."

Or perhaps less glibly, the fundamental premise of agile was "We're never going to be good enough at predicting the future, so let's instead become good at responding to the evolving present." A lot of subsequent methodological changes embrace this same line of reasoning. SRE, for instance, from an outsider's perspective, seems to emphasize mitigation (a responsive activity) as much as prevention (a predictive activity).

When people make predictions, they become attached to them. Doubly so if they pride themselves on being good at prediction. So if you're some kind of analyst predicting that crypto is a fad or that Meta and its AI-verse or metaverse or whatever will crash and burn, you start rooting for those outcomes, betting on them, and even making them part of your identity.

And I don't want that to be true of me. It precludes opportunistic and rational action, and it blurs good prioritization.

Is Meta going to crash and burn? I don't know. And I don't own any of their stock, so I don't care. Why waste time thinking about it, let alone bloviating about it?

Prediction is a sucker's play, particularly when there are stakes involved. The theory is that, by knowing the future, you can create arbitrages. But the reality is that you don't know the future, so you're just gambling, often blindly, without understanding important game theory concepts like stakes, risk, and expected value.

In the above post, I mentioned sports books, and the same logic applies to casinos, stock brokerages, and any number of other institutions. If you look at how those outfits make their money, it's essentially by monetizing game player hubris.

They don't know which stocks or teams or hands are going to win, but unlike their customers, their monetization strategy doesn't hinge on knowing. Rather, when setting up their businesses, they're asking themselves, "How can I come out ahead regardless of whether the coin lands heads or tails?" Heads they win, tails you lose.

So in emulating that mindset, when I'm asked to predict something in a business context, I school myself to stay away from guessing. Instead, I think of outcome probabilities, cost/benefit to the outcome, and then, logically, "How would I come out ahead in the case of any of the possible outcomes?"