Issue 15 - AI, SEO, and Ugly Ads

New agents and search careers

Author

The DTC Times
May 19th, 2024

AI.

It swept aside the Crypto hype cycle and has grabbed center stage in Silicon Valley. All of the big platform players are now rushing to build out their various LLM’s and agents, hoping this particular gold rush isn’t of the fool’s variety like much of the previous cycle (Bitcoin’s current pierce notwithstanding). 

Do you think Google has been suspiciously quiet when it comes to AI? That was the calm, and Google’s Keynote last Wednesday may be the start of the storm. Hold on to your hats because Google might be overhauling our experience of the internet soon. 

In this week’s issue: 

  • Is SEO about to become much less important?

  • Google’s boring AI assistants will be everywhere

  • A dive into how to make your ad creative simpler

Also, check our quick hits to learn about Amazon ads, Netflix getting into ad tech, and a new $500k DTC giveaway. 

Move over, SEO

@Levelsio, serial builder and founder of:

Saw that his website Nomad List, which ranks #1 on Google, is now suddenly absent from Google’s new AI LLM results:

These top-of-page, AI-generated texts are part of what Google calls Search Generative Experience (SGE), which seeks to accelerate user information search.

It does a variety of things, including:

  • Aggregating information from various sources

  • Providing quick links

  • Suggests follow-up questions

All of this appears in an “AI snapshot” in the most prime real estate on the page.

Image shared by @levelsio

The information provided, if sufficient, will negate the need to click any links. However, it also provides a few links which SGE deems most relevant.

  • Significantly, these differ from the usual ranking. Simply put, there is a new ranking that supersedes SERPs. 

  • Does it make SEO and SERPs obsolete? No, but it likely ushers in a new field of search optimization. That of AI SEO. AIO. SGEO? 

  • Let’s call it AI SEO for now. 

What do we know about AI SEO?

  • Honestly, not much. As with most algorithms, we’re dealing with a black box 

  • Google’s LLM (called PaLM 2) is a closed model, meaning it is not publicly available for use, perusal, or modification. This makes deciphering its logic even harder.

Nevertheless, SGE’s progressive worldwide release has been ongoing for about a year, starting with the US. Here are some of the early AI SEO learnings:

  • Leverage long-tail keywords: apparently, there is a trend of more specific searches, so these are valued.

  • Use natural language: conversational tones, less jargon, as well as question-and-answer formats within your content seems to work well. 

  • Implement structured data: rich snippets, schema markup, and JSON-LD format. 

  • Focus on unique information: SGE seems to try to aggregate all the information it can find for you. Bring net new information, or simply come at it with a new angle and you are likelier to be featured. 

  • Focus on opinionated, expert content: this is how you avoid being replaced by the two-paragraph summaries SGE will offer. 

  • Improve readability & directness: ever see Google highlight result text that matches your search? That’s what you want to offer. Example: 

Takeaway: 

SEO is poised to give way to “AI SEO”. Early tactics are emerging, but intelligent experimentation and volume will continue to shed light on its inner functions. New job openings/career fields incoming.

Google’s “boring” AI

Beyond SGE, Google has been working on a boatload of new AI features that integrate seamlessly (or so they promise) with the rest of the Google suite. 

Many of these were announced in the Google Keynote on Wednesday this past week. 

According to @boringmarketer:

These features might explain it:  

  • AI email agents.

Google’s AI can intelligently extract information relevant to your query from your inbox’s depths. Not only that, but it can then transfer the information in an orderly fashion into sheets. It’s basically an email virtual assistant. 

  • Chat record-keepers

AI that can access the full record of your company’s chatrooms. It can figure out where old conversations were left, where you’re at as a team regarding certain projects and if roadblocks were ever addressed. Amazing, if true. 

  • Everyday assistant

Google’s Gemini can access websites, log in, update your information, order, and do basically everything for you online. 

  • Ad agents

AI-generated headlines, copy, creative, and analytics. It remains to be seen whether these are any good. 

  • Sheets assistants

We mentioned Gemini could automatically transfer email information to sheets. It can also help you with in-sheet operations, including project management and trackers. This should democratize spreadsheet wizardry, which, to be honest, is a big deal. 

  • Personalization assistants

Because Gemini is sheet-fluent, it can also create content based on sheet information. This could be social media content, or even marketing content based on lead information. 

  • Video and image generation

Google demonstrated its new AI Video generator, Veo. It looks great and you can sign up for the waitlist right now. 

  • Up to 2 million tokens per prompt

LLMs break down prompt elements into tokens to process them. One word could equal one token, but the better ones break words down into multiple tokens to further improve comprehension. The “up to” part of this statement leaves us dubious, but ChatGPT currently handles 4096 per prompt max, so it still looks pretty good. 

Takeaway: Actually useful AI features that integrate and bridge the gaps between sheets, work chatrooms, and email sound like they’ll blend into our everyday life pretty fast.  

Lowering your ads “cognitive load”

Have you ever stared at your screen for hours trying to come up with the wittiest possible ad? The well-known intimidation factor of the blank page. 

It's a classic mistake. We get bogged down trying to stand out. 

Sam Ross, founder of @NumeralTax, said thinks it doesn’t have to be so complicated. 

The essence of his take is that ads should be intuitively understandable. If it takes more than a few seconds, you’ve lost. 

This is one of the 5 things Sarah Levinger removes from her customer’s ads.

  • If the eye doesn't know what to focus on, it'll usually skip straight past whatever it's looking at.

  • That’s why it’s important to work with designers who understand visual hierarchy, i.e. sequentially directing viewer attention from the most important to least important elements.

Sam’s advice from the thread above: 

→ Start with your product in-frame 

→ Ask a thought-provoking question 

→ Call out your ICP and/or their pain points

  • shorter: make it HARD for viewers not to get your ad immediately. 

There is another tactic to get around this problem, however - that of Barry Hott’s “ugly ad”. 

  • Simply, these are just ads that look and feel exactly like typical user content. As a result, you watch an ad, without feeling like you are.

  • When using “ugly” ads, the problem stops being about visual hierarchy or cognitive overload and becomes how well you can emulate a trend (or whether you manage to find the right content creator to work with).

  • Distilling your value prop or offer down into its most simple terms so it seems both natural and relatable is the trick here.  

Takeaway: Keep your ads simple. Sam’s framework should help you get there, but if it’s still a headache, consider the “ugly ad”. These are best achieved through affiliates, which you’ll find advice on how to pick in our Issue #10: Beyond the Metaverse

Quick hits

Reply

or to participate.