Hey there,

Welcome back to another Don’t Feed The Algorithm, your weekly dispatch from Averi — where we explore the intersections of marketing, AI and the future of work.

Everything is moving this week. And I mean everything.

1.5 million people quit ChatGPT. Claude went from 42nd on the App Store to #1 — without spending a dollar. OpenAI panic-shipped two major model releases in 48 hours. Anthropic dropped a tool that lets you import your entire ChatGPT history into Claude, eliminating the last thing keeping people from switching. OpenAI's head of robotics resigned. Google's rolling out a core update that has webmasters in free fall. And NVIDIA GTC starts Monday with a chip reveal Jensen Huang says will "surprise the world."

A month ago, people were debating whether AI platform loyalty was even a thing. Now we're watching the first mass consumer migration between AI products in real time — driven not by features, but by trust.

That's the theme this week: The Great Migration.

Users are migrating between AI platforms. Traffic is migrating from traditional search to Discover and AI Mode. Rankings are migrating out from under people's feet mid-update. And the entire infrastructure layer is about to shift again when Huang takes the stage in San Jose.

For startups, the opportunity is the chaos. When platforms are stable, incumbents win. When everything moves at once, the companies that react fastest — with the right content, in the right places, saying the right things — capture market share that took incumbents years to build.

Inside this issue: the full news breakdown, a migration capture playbook you can run in 90 minutes, fresh data on AI citation hallucination rates (spoiler: even the best platform gets it wrong 37% of the time), and the Canva campaign that turned every designer's least favorite client feedback into a 14-billboard takeover of Europe's busiest train station.

We also shipped three new features — LinkedIn posting, auto-generated internal links, and Strategy Map cleanup mode. More below.

Let's build.

— ZC & The Averi Team

Here's what we kept seeing: teams publish a great article, then forget to share it on LinkedIn. Or they share it — by copying the intro paragraph, pasting it into a post, and hoping for the best. Meanwhile, their blog has 40 articles that don't link to each other, so Google thinks each one exists in isolation.

Both problems have the same root cause: the stuff that happens after you hit publish is tedious, easy to skip, and nobody's job. So it doesn't get done.

→ Turn any article into a LinkedIn post. There's a new LinkedIn button right in the editor. Click it, and Averi generates a ready-to-post draft from whatever article you're working on — hook, structured formatting, and engagement patterns based on what actually performs on the platform. No more manual repurposing. Every article you create now has a built-in distribution layer.

→ Auto-generated internal links. Click "Generate Internal Links" inside any article. Averi scans your content library, identifies related resources, and suggests contextual links throughout the piece — right anchor text, right placement. You review each suggestion and approve or dismiss with one click. No more digging through old posts trying to remember what links where.

Internal linking is one of the highest-leverage SEO moves you can make — it builds the topic clusters that signal authority to Google and helps AI search engines understand how your content connects. The problem is that doing it manually is tedious and easy to forget. So most teams skip it entirely. Now your content engine handles it.

We also shipped Strategy Map cleanup mode — you can now delete content pillars that aren't working and regenerate fresh options without starting over from scratch. Your positioning shifted? A pillar felt right during onboarding but doesn't fit anymore? Clean it up and keep moving.

All three are live in your editor right now.

— Averi Team

We just launched resources.averi.ai — and it's exactly what it sounds like.

431+ free templates, playbooks, guides, tools, and frameworks for content marketers and startup founders.

Here's what's inside:

→ 50 ready-to-use templates for content strategy, SEO, social, email, and editorial calendars — the kind of docs that usually live in someone's private Notion and never see daylight.

→ 20 tactical playbooks with phases, checklists, and timelines. Including "The First 90 Days" — a launch playbook for standing up content marketing at a new company from scratch.

→ 18 free tools — interactive calculators, content analyzers, and generators you can use right now without signing up for anything.

→ Complete pillar guides covering content marketing, SEO, AI content strategy, startup marketing, content ops, and lead generation. Each one is the kind of deep dive that would normally be a $49 ebook from someone's course funnel.

→ Benchmarks, comparisons, industry breakdowns, and a 75-term glossary — because half the battle in content marketing is just knowing what "good" looks like for your stage and vertical.

Not sure where to start? There's a 2-minute Content Strategy Quiz that gives you a personalized recommendation based on your stage, team size, and goals.

Someone just asked ChatGPT: "What's the best project management tool for remote startup teams?"

Your SaaS wasn't in the answer. Your competitor was.

That person didn't click a single Google result. They didn't read your G2 reviews. They didn't visit your beautifully optimized landing page. They typed a question into an AI assistant, got a recommendation, and signed up for your competitor's free trial… all in under ninety seconds.

This is happening millions of times a day. And it's accelerating.

The question isn't whether AI product recommendations matter for your SaaS. The question is whether you'll engineer them deliberately or leave them to chance.

This is the playbook for deliberate.

We all know the industry loves to pretend it’s got it all figured out.

But here?

We tell the truth.

Every week, we’re asking you to share how things really work behind the scenes. Because getting better starts with getting honest.

Don’t worry, it’s a safe space.

Drop your vote.

Canva turned the most hated client feedback in marketing history into a 14-billboard takeover of Europe's busiest train station.

London Waterloo. 100,000+ daily commuters — a disproportionate number of them agency creatives, in-house marketers, and brand managers. Canva and Stink Studios took over all 14 billboards in the Waterloo gallery, and every single one was a different designer nightmare made physical.

The Canva logo bursting out of its frame — because someone asked to make the logo bigger. A landscape ad crammed sideways into a portrait space — because the 16×9 was meant to be 9×16. A real 3D e-bike mounted to a poster — drag and drop, literally. A billboard with its background erased to reveal the raw brick wall behind it — the product is the execution. And one ad covered entirely in sticky notes with conflicting client feedback, mapping the path from "brief lands" to "humbly accept standing ovation."

The campaign extended across YouTube and LinkedIn through summer 2025, with Stink Studios continuing the partnership into 2026. Media by OMD. Special builds by Talon and Grand Visual.

Why we love it:

This is a B2B SaaS company that understood something most B2B SaaS companies never will: your audience doesn't want to see your UI. They want to feel seen. Every billboard is a product demo disguised as an inside joke. Canva didn't say "our resize tool adapts content across formats." They showed a sideways ad and let every social media manager in London wince in recognition. The targeting wasn't algorithmic — it was geographic and deeply human. Right joke, right people, right station.

The algorithm could never because every execution looks wrong on purpose. Logo too big. Format sideways. Background missing. Ad covered in chaos. An optimization engine would tell you to put a clean value prop and a CTA on a billboard. It would never greenlight mounting a bicycle to a poster or erasing the ad itself. The wrongness is the hook — and you can't A/B test your way to an insight that specific.

Your industry has a "make the logo bigger." The shared pain everyone complains about but nobody campaigns around. Find yours.

→ See the campaign on The Drum · Full breakdown on LBBOnline · All executions on Ads of the World

The bots are citing you. They're just making stuff up about what you said.

The Columbia Journalism Review tested eight AI search tools on their ability to accurately cite news sources. The results should concern anyone who cares about being cited by AI:

→ Perplexity hallucinated on 37% of citation tasks — the best score in the test

→ ChatGPT got it wrong 67% of the time

→ Gemini missed on 76% of queries

→ Grok 3 hallucinated on a staggering 94% of citation attempts

The failure mode is worse than you think. Perplexity doesn't invent fake URLs — it cites real sources with fabricated claims. The link looks legitimate. The publication name is real. But the information attributed to that source? Made up. For your content strategy, that means even when the bots cite you, they might be putting words in your mouth.

Meanwhile, the citation landscape is consolidating fast:

→ Only 11% of domains get cited by both ChatGPT and Perplexity. The platforms are pulling from almost entirely different source pools. Optimizing for one doesn't mean you're visible on the other.

→ 85% of brand mentions in AI answers come from third-party pages, not your own website. The reviews, articles, and Reddit threads written about you matter more than the content on your site.

→ Brand search volume — not backlinks — is now the strongest predictor of AI citations (0.334 correlation). Backlinks show "weak or neutral correlation" with LLM visibility. Decades of SEO wisdom just got inverted.

→ The average domain age of ChatGPT-cited sources is 17 years. If you're a startup, you're fighting an incumbency bias baked into the model. The workaround: multi-platform presence. Brands appearing on 4+ platforms are 2.8x more likely to show up in ChatGPT responses.

→ Citations drift 40-60% month over month. The source that got cited in February might not get cited in March. This isn't a set-it-and-forget-it game. It requires ongoing monitoring.

What this means for you: Getting cited by AI is only half the battle. You also need to verify what the bots are saying when they cite you. Run your brand queries across ChatGPT, Perplexity, Claude, and Gemini this week. If the citations are wrong — and statistically, there's a decent chance they are — you now have a content gap to fill: clearer, more extractable answers that leave less room for the model to improvise.

→ Read the CJR study · Dig into the full citation research · See the 2025 AI Visibility Report

Strategy You Can Steal

The "Great Migration" Playbook — How to Win Customers When Platforms Lose Trust (90 minutes)

1.5 million users just left ChatGPT. Anthropic launched a memory import tool the same week. Google's core update is reshuffling rankings. Everything is in motion — and when everything moves, the companies that have their arms open win.

Here's how to run a "migration capture" play in 90 minutes:

Block 1: Identify the Exodus (20 min)

Every platform disruption creates refugees. Your job is to find them.

Right now, the biggest migrations happening:

Pick the migration most relevant to your audience. Where are your potential customers moving from — and what are they looking for when they get there?

Block 2: Build the Landing Zone (30 min)

When people switch platforms, they're actively searching for alternatives. This is the highest-intent content window you'll get all year.

Create ONE piece of content for each of these formats:

→ The comparison page. "[Your product] vs [thing they're leaving]" — whether that's a competing tool, an old workflow, or a platform they've lost trust in. Comparison content converts 2-3x higher than generic landing pages.

→ The migration guide. "How to switch from X to Y in [timeframe]." Anthropic literally built a migration tool. You can build a migration guide. Make the switch feel frictionless and you own the decision moment.

→ The trust signal. Write a post that says what you believe. Not your product features — your actual position on something your industry is debating. Anthropic's "no" to the Pentagon was worth more than any ad campaign. What's your "no"?

Block 3: Capture the Conversation (20 min)

Migrations happen in public. People post about switching. They ask for alternatives. They share frustration with the thing they're leaving.

→ Reddit: Search your category + "alternative" or "switching from" or "looking for." These threads are where AI citation begins and where purchase decisions happen.

→ X/Twitter: Monitor the hashtags and conversations around whatever disruption affects your space. Reply with genuine help, not pitches.

→ AI platforms: Ask ChatGPT, Claude, Perplexity, and Gemini: "What are the best alternatives to [competitor]?" If you're not showing up in those responses, your migration capture will leak.

Block 4: Fortify Your Position (20 min)

If you're the thing people are migrating to, you need to hold them.

→ Onboarding speed matters more than ever. Anthropic's memory import tool gets users productive in 24 hours. How fast can a new user see value in your product?

→ Stack your switching costs positively. Don't lock people in — make leaving feel like a loss. Integrate with their workflows. Store their context. Become the system of record.

→ Tell the story of who's switching. Social proof during a migration window is disproportionately powerful. If notable users are switching to you, say it. Loudly.

The output: A migration capture playbook with one comparison page, one migration guide, one trust-signal post, and a list of communities where the conversation is happening.

The companies that win during platform upheaval aren't the ones with the best features. They're the ones standing at the exit door with open arms and a clear reason to stay.

Know of a role that should be here? Reply to this email and we'll feature it next week.

Did You Know? —  The same brand's AI citation volume can differ by 615x depending on which platform you check — same content, same month, wildly different visibility.

The bots are reading everything. They're just not all reading the same things.

Til next time,