This week’s Field Notes is about how to optimize and increase your brand visibility in LLMs.
A lot of what I write here attempts to abstract out frameworks, strategies, or underlying mechanisms related to organic growth. This, I believe, is underrated in nowadays. We live in a world of short soundbites, quick hits, tips, tactics, and hacks. I don’t personally enjoy the TLDR of it all!
The problem with this is that a tactic means nothing in isolation. If you do the same thing I do, it might have completely different results due to execution, context, and even a misunderstanding of what the tactic is doing in the first place.
However, I’ve learned over many years spent in stakeholder meetings, and through attending hundreds of concerts, that both executives and concert attendees sometimes just want you to play the hits. A little red meat for the base, if you will.
So I’ll give you a bunch of tactics you can use to optimize for LLMs. I still implore you to think critically and apply the right tool to the job; don’t just blindly follow a list of tactics!
Quote of the Week
“Knowledge is of no value unless you put it into practice.”
– Anton Chekhov
Field Notes
Alright, so AI is leading us to the new frontier of organic growth.
ChatGPT usage is growing, leads from “AI” are accelerating for most businesses (particularly if you use self-reported attribution in addition to visibility and traffic measurement), and even classic search is merging into a monstrous AI-search Frankenstein due to AI Overviews and AI Mode.
What the hell is a brand to do about all of this?
(Call us of course)
Certainly, it can’t simply be to do what has been done before – pick a high volume keyword from a content gap analysis, run it through Clearscope, and write some generic SEO book report with 200 more words than your competitors’ blog post.
We’re all learning more every day. This stuff changes like the wind.
But here are some ideas for what you can actually do to bring some self-efficacy, and internal locus of control, in the face of such tumultuous uncertainty (and opportunity).
Content Planning in the AI Era
Content planning needs to a reboot.
As alluded to above, the dynamics of AI are altering user behavior, and possible the value of certain types and formats of information.
On the surface, people self-qualify and type longer and more specific queries into ChatGPT than they do Google Search.
We don’t really have accurate prompt data like we do keyword data. Some tools say they do; however, this data isn’t incredibly useful or trustworthy (yet – and that’s a good thing).
How to plan and prioritize organic & content in the AI era, where user behavior is evolving and the data is obfuscated or missing?
The most effective content strategies now triangulate three dimensions instead of relying solely on search volume.
- Buyer research: What language do prospects actually use in sales calls? What are their pain points and jobs-to-be-done?
- Product research: How do you fit into existing categories and comparisons? How does your product solve customers under-met needs?
- Channel research: How can we quantify opportunities and optimize for distribution based on the most propitious channels (Google, ChatGPT, beyond)?
We’ve used this three-prong model well before ChatGPT entered the scene, but now I see the buyer research dimension as being imperative and holding a lot of competitive alpha.
Here’s a cool tactic I’ve been using recently that directly ties qualitative customer / buyer research with product and channel data and dimensions.
I built a CustomGPT (and a custom product, though I’ll show you the GPT so you can do this too) that analyzes unstructured voice of customer and turns it into nascent content ideas as well as categorical prompts to track in AI.
I simply upload a sales transcript, kickoff call, or another artifact that dives into detail about their audience or market.
Eventually, I pull through a set of prompts to track in AI, and I use Peec (or Profound or Search Party) to audit the brand’s visibility.
Then, I look into the citation sources that are influencing the results of these prompts.
This is very handy when it comes to off-page SEO and brand mention campaigns, which I’ll cover in a bit.
But it’s also a cool source of topic and keyword research. You can click into the individual source URLs, export them, and then run a bulk analysis in Ahrefs, Semrush, or your SEO tool of choice.
This reverse-engineers the most influential URL sources against your relevant prompts, and then gives you classic search keyword ideas.
I note even the pages cited that appear to have zero classic search value or keywords.
Some of my most successful content pieces this year have had zero detectable search volume. I discovered them through pattern matching pain points in sales conversations and through this reverse engineering process.
These pieces rode nascent waves, got cited by AI systems (as well as ranked in search), and now drive substantial traffic, too. Most importantly, they came up directly in sales conversations with prospects (always close the loop with sales and self-reported attribution)
Be the Source
We introduced a framework for LLM optimization with three swings you can run for tactical playbooks:
- Be the source
- Be included in the source
- Replace the source
The most impactful and lowest hanging fruit you can do is simply create content on your own website that answers questions and communicates messaging about your brand, product, and use cases.
Much of what influences AI outputs, beyond the commonly talked about juggernauts of Reddit, Wikipedia, and YouTube, is the more boring, banal, and expected source: your own website.
Yes, content creation is a path for success in GEO as well as in SEO. Who’da thunk it?
Here’s a list of common content formats I find are high leverage in the GEO world:
I always like to start with the holy trinity, if it doesn’t already exist for a topic / query: a product page, a listicle, and a piece of product-led content. Bonus for a case study.
So, if I wanted to be included in outputs related to “SEO agencies,” I’d create a services page, a listicle of the top SEO agencies (possibly splintering that by industry, like “best B2B SEO agencies” or “best SaaS SEO agencies,” and an article about our methodologies for B2B SEO strategy.
Depending on your product category, documentation could play heavily into your visibility. We work with several highly technical products, some of which are open source, so you’re going to see a lot higher leverage with documentation and GitHub here.
Sometimes, a brand or product serves several distinct personas or industries. In this case, it may warrant a highly niche page created for service or product pages. Think:
- SEO services for fintech
- SEO services for HR tech
- SEO services for dev tools
And on and on. Most of these won’t show up with any search volume in Ahrefs, so it’s sort of the triumphant return of product marketing and sales enablement content. The difference now is that there is inventory and incentive to create them, as ChatGPT is now your sales person.
Here’s an easy one for startups and challenger brands who have less red tape: create competitor comparisons.
Your larger incumbents have a ton of legal and PR red tape, and they already have existing brand demand. By associating yourself with your larger competitors, you build a connection between them, yourself, and your category – long term associations that will likely increase AI visibility. You also get the advantage of appearing when people directly search for [competitor] alternatives in search or LLMs.
Another very underrated tactic is to produce original research that relates to your buyer pain points.
This is the new TOFU in many ways, as users start their journey researching a topic broadly (maybe they used to click on “what is [topic]” pages, but this is largely answered by AI Overviews).
Where one can inject proprietary data that alters or owns the narrative in a category, one is cited, mentioned, and can nudge users down the ol’ customer journey towards their solution.
Be Included in the Source
“Brand mentions are the new backlinks.” Okay. How do you get brand mentions?
First, focus all your efforts on being a brand worth mentioning.
Like, great product solving undermet needs, marketing worth talking about naturally, etc.
You’re probably best ignoring all the tactics that follow and just focusing on that, honestly.
But you want outreach tactics, so here you go.
Everyone’s sending the same templated emails asking for list inclusion. The brands getting mentioned are doing something different. You’ve got to stand out.
- Method one: Do something nice for them first. Instead of asking to be added to someone’s list (cold), ask to interview them for your content. Bring ’em on your pod, webinar, whatever. Collaborate on research. IDK, there are a million ways to add value. Content marketing becomes relationship building. The value flows to them first, and mentions often follow naturally.
- Method two: The frenemy approach. Your competitors in sales cycles aren’t always your competitors in search. Partner with complementary tools to create better resources together. Two smaller players teaming up can often outrank the incumbent.
- Method three: Get creative with value exchange. Send pizza to target accounts. Hire Cameo celebrities for personalized outreach. Host intimate meetups that bring prospects and partners together. The more memorable your approach, the more likely you are to cut through the noise.
Overall, relationships trump outreach tactics. The best placements come from genuine connections, not cold emails begging for links or inclusions. Try not to use AI automated outreach sequences, either.
Here’s another one you could do (only if it applies to your own industry and data): buy a niche site that has tanked in traffic but shows up frequently as a source in LLMs. Inject your brand in historical content.
I’ll be honest here: many of the off-page “tactics” and techniques will feel like whack a mole or Sisyphus, unless you’re actually a brand worth mentioning. There are just too many sources and websites to run tactical outreach campaigns and really move the needle. How are you going to dominate a review site, let alone several review sites, with few customers willing to give you good reviews, for instance?
Seriously, if you do the underlying things that make your product naturally worth mentioning, you won’t need to worry a ton about the tactics here. You’re not just trying to get mentioned once (or twice or ten times). You’re building the foundation for the Matthew Effect—where success breeds more success, and mentions lead to more mentions.
The other thing with off page is it will require cross-functional support, which SEOs are notoriously not great at. Look at this list below of common off page citation sources. Editorial lists may be influenced by link building approaches, but review sites, partner pages, social media, forums – all of these will require coordinated strategies with other teams like brand, PR, customer marketing, possibly affiliate, and social.
So maybe your best tactic is to go get coffee with someone outside your team.
Replace the Source
This is the swing that confuses people. What’s the difference between “be included in the source” and “replace the source?”
Here’s my theory: the information marketplace in LLMs is very imperfect right now.
There are a ton of inefficiencies, and if we’re to believe that the user behavior will change with adoption of LLMs and AI search, then there will emerge the need and incentive for new content that better answers user questions (and novel behavior like highly personal context, knowledge docs, memory, and multi-prompt conversations).
Also, in many industries, a lot of citation sources just suck. The content covering a topic is just not great – it’s generic, high level, ultimate guides, etc.
Finally, the broader web needs to be seeded with your brand while co-occurring with your category (training data + real time search benefits).
Lots of tactics to replace the source, most of which I would summarize as “just great brand marketing.” I.e. do stuff that is worth talking about.
Here’s the big actionable tactic: original research. Data-driven content marketing. Many emerging and non-consensus industries are in desperate need for good data and narrative shifting content.
You can create it. And you can distribute it through a service like Stacker or through your PR team.
You can also create a specific page or answer where previously there were only ultimate guides and generic content. This is constantly the case in my analysis of AI visibility.
These, for example, are the sources influencing the results of a prompt related to “replacing your legacy ABM stack with AI first tools” (I’m paraphrasing):
A few decent sources, namely the “AI tools transforming ABM” one. But we can do better. Run the playbook – create site content, get included in these, and then leverage third party platforms to drive the message home.
I’m talking: podcast appearances, Reddit threads, YouTube videos (underrated for B2B), and social media (decentralized LinkedIn campaigns). All related to replacing legacy ABM stacks with AI-native tooling.
Replace the source with better answers.
Upstream vs. Downstream
These are all downstream tactics. Upstream of these tactics likes the hard to fake signals that will drive cumulative value: good product, customer experience and support, effective positioning and messaging, cross-functional alignment and goals allowing you to deploy resources to effectuate the best tactics.
Otherwise, a brand mention here or there, a listicle in isolation – it ain’t gonna move the needle.
We were approached months ago by a massive brand – think Superbowl commercials, widely known, household name.
They have a horrible reputation. Cheap quality, bad product, poor customer support. And they wanted us to help remedy their sentiment and messaging in LLM outputs.
You see, however, it is not up to us how LLMs talk about them. It is up to the hundreds of thousands of data points that, together, inform the LLMs and cultivate the message to users that “this is not a good product.”
We could reach out to 1% of these and perhaps get them to tweak the message, but again, it’s whack a mole, because as soon as we do that, another thousand will pop up on Reddit, social, etc.
Ultimately, you have to fix the problem at the source if you want to play the long game. So I hope these tactics are useful, but I also hope you think from first principles and before launching a listicle series, ask, “why should an LLM recommend my brand?” And then lean really hard on your intellectually honest answer to that question.
Want more insights like this? Subscribe to Field Notes
