Some thoughts on travel content sites in the AI + Google HCU + whatever hit comes next realm
was posted by
Stuart McDonald
in
Discussion
,
AI
,
Writing
,
Blogging
.
Featured on May 19, 2024 (5 months ago).
Some thoughts on travel content sites in the AI + Google HCU + whatever hit comes next realm is rated
5/5 ★
by 1 member.
+ 382 others joined this month
Comments
Posted this on LI and Ian suggested may be of interest to some of you TravelMassivers, so here we go:
Thinking about “travel guide like content sites” in the AI + Google HCU + whatever hit comes next realm, and a few thoughts on going forward:
EMBRACE AI NO CHARGE
Go full indexing in the hope there’s enough resulting traffic to remain a going concern. Past iterations of this (Google Rich Snippets etc) indicate that while some traffic will result, it will be less than whatever preceded it. Prepare for a substantial—and steepening—business downsizing.
EMBRACE AI WITH CHARGE
In a fantasy world where the AI platforms licence content, it will be a poison pill. Milk it for the year or so they pay up, then prepare for the above-mentioned downsizing.
BLOCK AI ENTIRELY
Risk substantial to near-total loss of SEO-derived traffic—depending on how search shakes out over the next year or so, maybe sooner.
PAYWALLS
Have been asked about this a few times as moved my main site (Travelfish www.travelfish.org) to a hard paywall years ago. The hit to the hip pocket is real, as is a fall in rankings. The potential of this route depends on the site. If it has a strong community, or some very hard to find/compelling/niche content, then can be a smart move, but I can think of only one site in Southeast Asia that is well positioned for this (Vietnam Coracle www.vietnamcoracle.com), most sites are too hooked to the drive-by traffic that aggressive SEO delivers to make this move and stay afloat financially. LP nuking Thorntree is the all-time-classic head-desk move in this area.
PRIVATE COMMUNITIES
A potential halfway house to a paywall. For most publishers this most likely necessitates building on another platform like FB or whatever, which entails its own challenges, but think can be a good indicator of the feasibility of a paywall approach down the track.
PRIVATE NEWSLETTERS
Ties in with the previous two. If the newsletter is going to be substantial in value, it is a truckload of work, and so will need to at least have a paid stream. As with paywalls, if the site doesn’t already have an engaged audience, getting drive-by traffic to pay for a newsletter is a big ask. Newsletter fatigue is real, doubly so for paid ones. Hard yards, very long game.
PRIVATE COURSES
The good old “if you can’t make money from it, teach it,” concept has popped up a few times from publishers I know. Think is a very high-risk approach, but is something I’ve looked at in depth a few times in recent years. Aside from it meaning I’d need to shave more often and not work naked, it also requires an entirely different skillset, a hell of a lot of work to do well, and conceptually is quite removed from the actual provision of traditional travel information. Most examples I’ve seen that appear to have some traction are focused more on the generic pre-trip stuff rather than the trip itself.
PRIVATE PODCASTS
Harder to make money from the newsletters. Discovery is an absolute nightmare.
SOCIAL MEDIA
No idea. I gave up and pivoted to mostly photos of my dog a few years ago.
CLOSING THOUGHTS
Interesting times!
Thanks for sharing, Stuart! Some good thoughts to consider.
1. On "Block AI Entirely" — I've blocked ChatGPT in our robots.txt (probably a bit too late, but anyway). And yes, Google and others will train AI on our content but the cat is out of the bag on that. Blocking ChatGPT is a symbolic protest.
2. On "Private Communities" — if you don't own your community platform, your data is probably being trained by your provider (e.g. Facebook, MightyNetworks, etc). It looks like Slack is already doing it! arstechnica.com/tech-policy/2024/05/slack-defends-default-opt-in-for-ai-training-on-chats-amid-user-outrage/
(This is one of the reasons I developed the Travel Massive platform for our community and why we host and operate our own independent platform.)
3. On "Podcasts" — I'm not sure podcasts are safe from AI. I've seen a few indiehacker projects that are using AI to transcribe all the content in the podcast and make it indexable / searchable. Spotify, Google, etc could also be doing this but are yet to unleash it. If I can recall the project I'll come back and share it here.
Yeah, I think it is a fair assumption that anything public-facing will be hoovered up—why I designated podcasts etcetera up top as private, but yes, unless you roll your own community platform one should assume the platform you use is ingesting everything. Many AI platforms don’t obey robots, so I block at Cloudflare level as well, but it is like a game of whackamole.
Thanks for sharing Stuart
Some tactical thoughts/questions I've been mulling over:
Beta SGE vs what's been actually rolled out is VERY different, and I assume (/hope) because testing demonstrated users didn't really want a giant Mr Clippy on top of their search results. "AI Overviews" are really just pimped up featured snippets at this stage, although yes will obviously change over time as user behaviour adapts.
I've not seen any definitive documentation anywhere that says you need to allow Gemini access to your data in order to feature in the website link cards. Has anyone else? I *think* at the moment the website link cards are coming from the regular search algorithms. This seems like a good thing?
There will be an order of magnitude loss of SEO traffic, but I think the impacts will be very uneven. Users will still want premium, human curated expertise, and Google (and other LLMs) will still need their data to ingest. It might be wishful thinking but I can't imagine SEO going to zero.
As so who will get a smaller slice of a much smaller pie... who knows. Everyone thinks their baby is the cutest, every website owner thinks their content is truly high value. We're doubling down on the only thing we know how to do, and until it's demonstrated that you need to allow Gemini access in order to feature in the results we're also blocking all the bots in Cloudfare, although keeping that decision in constant review.... Perplexity might be an exception.
As far as I understand it, blocking google-extended blocks Gemini (or whatever they are calling it today) and Vertex, but not the search crawl. See the May 14 entry in their changelog here: developers.google.com/search/updates#updated-the-description-of-the-google-extended-product-token
Link cards can be controlled via nosnippet
The issue is they seem to be intent on cramming so much crud up top, you’ll need a new level of intestinal fortitude to reach the organic links—sorry, I mean links to Reddit. See Rafat’s post on LI for a pretty amusing example.
I don’t think SEO will go to zero—there’s already a tonne of posts about on how to game for the AI results if you want to play that game, but yeah I think many may see somewhat of a lifestyle re-alignment.
As for quality content, I’m gutted to see our post of where to buy second hand women’s clothes in Bangkok is no longer ranking well—do not ask me why we ever had that post—at one stage it out-trafficked the entirety of our Java coverage—FML.
"What is (or was) your #1 ranked blog post?" could be an interesting discussion!
No idea. I don’t check views/traffic at all anymore, disconnected GA a year or so ago and can’t figure out how to access the old data as GA4 was designed by people who are clearly not human. Traditionally the itineraries got a fair swack of traffic, a few million views at least, and for years we ranked very strongly for some weather queries that delivered a truckload of (mostly worthless) traffic, though we slowly lost those rankings. It’s a big site—around 70-80,000 pages (depending on what you want to call a page)—so more a sum of its parts than any particular one-hit-wonders. I nuked the second hand clothing one a while back.