Good morning.
We don’t know about you, but over here, spring arrived and brought sun for exactly two days.
Then it remembered it had somewhere else to be. Now it’s gray, it’s raining, and the only thing blooming is our disappointment.
If we wanted to get rugged this hard we’d buy NFTs in 2022.
Gemini referral traffic has grown so quickly, it’s lapped Perplexity
Last year, Google was certainly feeling the holiday spirit.
Something clicked for Gemini in December: Referral traffic surged 51%, then another 42% in January, per SE Ranking’s data across over 101,000 sites. Eight months of nothing, then this.
The catalyst was Flash. Google’s final Gemini 3 release, which became the default across the Gemini app and AI Mode for Search, and traffic climbed almost immediately.
If a single product launch can boost a platform’s growth rate that dramatically, your AI visibility strategy needs to be nimble.
Brands that were already showing up in Gemini’s responses caught that wave. Brands that weren’t missed entirely.
And Perplexity didn’t see it coming: As of January, Gemini is sending 29% more referral traffic than Perplexity globally.
Five months ago, Perplexity was sending 3x more than Gemini. Oh, how the tables have turned.
If you’ve been tracking Perplexity as a referral source, Gemini may have already overtaken it in your analytics. It’s worth checking before your next reporting cycle.
And some numbers behind the numbers:
- ChatGPT’s traffic share fell from 86% to 64% over the past year
- Gemini climbed from 5% to 21% over the same period
ChatGPT still dominates AI referral traffic at around 80% of all visits. And though its lead over Gemini is shrinking, it’s still roughly 8x. For now.
Relying solely on ChatGPT optimization is starting to look like relying solely on one search engine. Track all the major LLMs, not just the one topping the charts today.
ChatGPT ad business is growing fast, but is it ready for everyone yet?
Every new ad platform has a golden window before costs climb and competition crowds in.
That’s ChatGPT right now: OpenAI has crossed $100M in annualized ad revenue in just six weeks.
That milestone comes from a platform still reaching a fraction of its potential audience. With roughly 85% of Free and Go users eligible to see ads.
A few more reasons to pay attention:
- Over 600 brands are on the platform and all through a managed pilot, not self-serve.
- Self-serve access is expected in April, which is when the broader market actually gets in.
- International expansion is in the works, with Canada, Australia, and New Zealand in the pipeline.
And users are finding ads relevant. OpenAI reports fewer than 7% of ads are rated low relevance by users. A solid start, though they’re still actively working on it.
Advertising isn’t a side experiment for OpenAI. They hired a former Meta ad executive to lead sales and have an IPO to answer to.
ChatGPT is where users are actively seeking recommendations and solutions. That’s high-intent territory. When self-serve opens and advertisers pile in, costs will climb.
On the other hand, managed-only access, limited inventory, and an audience that’s still warming up to ads means this isn’t a plug-and-play channel yet.
April’s self-serve launch is the real moment to watch.
Keep it on your radar. Just don’t reshuffle your budget for it quite yet.
AI search is the fastest-growing discovery channel
Your customers ask ChatGPT, Perplexity, Claude, and Gemini for recommendations daily. Is your brand the answer?
Peec AI shows you exactly where you stand:
- Track your visibility and sentiment across all major LLMs
- Benchmark against competitors to see your share of voice
- Get step-by-step guidance on improving your AI visibility
Turn AI search from a black box into a measurable growth channel with clear metrics and a scalable strategy.
How to track your AI prompts on a budget?
Let’s talk about prompt tracking. Or how you monitor your brand’s appearance in AI-generated answers across a saved set of prompts.
Instead of chasing Google rankings, you’re tracking mentions, sentiment, and how often a model surfaces your brand for relevant queries on different LLMs like ChatGPT or Gemini.
Naturally, tracking prompts is a different beast than tracking rankings. Tracking every prompt for every keyword inflates costs, muddies your data, and burns resources. A big ouch.
And Briie Anderson breaks down how to build a prompt tracking strategy that generates real progress without draining your budget.
Prompts that don’t align with your business goals reduce ROI. Prompts that don’t fit the tool’s specs produce false positives.
You need a smarter list.
Start with your audience: Before spending anything, check GA4 to identify which LLMs are already sending you traffic.
Research shows LLM preferences vary by gender, age, and country, so confirm where your audience spends time before setting up tracking.
In GA4, go to Reports > Acquisition > Traffic acquisition. Switch the primary dimension to Session source/medium and filter for LLM referrers. Simple as.
Don’t limit yourself to ChatGPT and Gemini, as you might miss smaller but growing referral sources.
Prepare your brand topics before building prompts. This helps you validate branded terms and catch regional differences.
Validate competitors using entity signals. Search your brand on Google and review the People Also Search For panel.
That list reflects how Google’s Knowledge Graph actually maps your competitive landscape.
Separate prompts into three types: Branded, comparison, and non-branded. Branded prompts inflate visibility metrics without telling you if you’re being recommended organically.
The real opportunity is non-branded prompts where the model recommends solutions unprompted.
Prune aggressively. Remove anything off-topic, commercially irrelevant, or tied to fast-changing topics your tool updates too slowly to capture.
Finally, treat prompt tracking like CRO: Go in with a hypothesis, test in clusters, and iterate.
That’s the short version. If you want a full breakdown, check the full post here.
Transform raw recording to professional content within minutes
Vidocu turns one video upload into a full content engine:
- Translate subtitles and generate localized voiceover in 65+ languages. No re-recording required.
- Auto-generated guides with screenshots and SEO-ready headings, straight from your video content.
- Built-in editing, annotations, zoom effects, and team collaboration keep your whole crew aligned.
See it in action here, from raw video to polished, edited content, auto-generated help article, and multi-language translation.
Why do people stick around for your brand?
Brand values? We like to think that’s why.
But more often than not, it’s gamification:
The number one reason consumers remain loyal (at 41%) is simply that the brand offers a loyalty rewards program. The golden handcuffs, if you like.
Program can be bigger than the product: While 33% stay for high-quality products, the program itself is the stronger tether for this specific segment.
Only 7% stay because the brand aligns with their values, and only 5% because the brand makes them “feel special.”
This data contradicts the “love the brand” narrative.
For many, loyalty is transactional. They are loyal to the rewards, not necessarily the logo. If the program stops, the customer stops.
The takeaway: If you don’t have a loyalty program, you are losing 41% of your potential retention power. Start simple: a points-for-purchase system is better than nothing.
AI EDUCATION: ChatGPT, Claude, Gemini, Midjourney… So many names, but what’s actually useful for you in your work? There’s a newsletter called The Deep View that exists to sift through all the noise and get you up to speed on what’s actionable with AI products, and it’s free. Join 512,000+ subscribers with one click and let AI empower you.*
GOOGLE: Demand Gen can now turn static images into video ads automatically using its Veo AI model, covering YouTube Shorts, Discover, and Gmail in one go. With campaigns already delivering 26% more conversions per dollar, leaning into the AI creative generation is a natural step.
REDDIT: …is forcing suspected bots to prove they’re human through passkeys, Face ID, or government IDs, after already removing around 100,000 bot accounts every day. With Cloudflare predicting bots will outnumber humans online by 2027, you should be asking whether your Reddit engagement metrics are as real as they look.
META: Instagram Plus is nearly here, with features like caption links, Story Extend, and offline Reels reportedly in the mix. Meta Verified already has an estimated 35M paying subscribers, so there’s clearly an appetite for premium perks. Worth testing if your audience is creator-heavy.
LINKEDIN: A little prep never hurt anyone, right? From June 22, 2026, spontaneous live streams are out on LinkedIn. You’ll need to schedule in advance, even if that’s just minutes before going live. Given that event ads already drive 31% more viewership on average, treating LinkedIn Live like a proper broadcast rather than a last-minute thought makes good sense.
*This is a sponsored post.
“My thunder comes before the lightning;
My lightning comes before the clouds;
My rain dries all the land it touches.
What am I?”
You can find the answer here.
Interested in advertising in one of our newsletters?
Connect with over 100,000 of the world’s best marketers who read a newsletter by Stacked Marketer.
