He is the growth architect behind 3 of the fastest-growing companies (Superhuman, Wispr Flow, and Viktor).
He built their $100M growth engines.
I just watched his 1 hr. 21-min podcast with Harry Stebbings (@HarryStebbings).
Here are my top 30 takeaways:
1.) Matt (@MattSwulinski) believes the E-commerce playbook is the correct model for SaaS. Every dollar of ad spend must trace back to an acquisition event. Just like every cent in E-commerce must lead to a purchase or add-to-cart.
2.) Start paid advertising immediately. Not after you have built organic brand presence. Paid is the fastest way to validate that your product-led growth funnel actually works. Because you can test messaging, funnels, and positioning within a week rather than waiting months for organic signals.
3.) 3 core acquisition channels for any early-stage company are Meta, Google, and lifecycle (email, SMS, or push). You can scale to your first $1 to $10 million in ARR on just those three before adding anything else.
4.) Before spending a single dollar on paid, get your analytics and conversion tracking infrastructure right. He estimates 90% of companies skip this step. Result: the ad platforms cannot identify your best customers properly. And you end up with a high cost per acquisition.
5.) SaaS has no out-of-the-box attribution tool equivalent to what e-commerce uses (like Triple Whale). SaaS teams must build their measurement stack from scratch with an analytics leader and a developer who can wire marketing signals back into each platform.
6.) If you want best results early on, optimize purely for your single most important acquisition event... whether that's a download, trial start, or a completed onboarding step.
7.) Retention is the product team's job. Growth's job is bringing in quality users at an acceptable cost.
8.) A 1:1 LTV-to-CAC ratio is acceptable early on if you have raised enough capital. Because speed of distribution matters more than efficiency when the market is crowded with clones. The long-run target is roughly 3:1.
9.) For AI or usage-based SaaS products, LTV-to-CAC alone is not enough. You must look at LTV on a gross-profit basis, accounting for token costs, inference costs, and all other variable usage expenses. Otherwise your unit economics will look healthier than they are.
10.) Creative is now the targeting on Meta. A Meta algorithm update called Andromeda removed manual audience targeting controls. Meta now analyzes the creative itself to find the right audience. This means the media buyer's job has become almost entirely creative strategy.
11.) At the scale of a $100K monthly Meta budget, you need 400 to 500 new creatives per month to avoid plateauing. Diversity of age, gender, setting, and hook style in creatives is essential.
12.) At Viktor, they run a creator program that pays creators a percentage of ad spend to produce UGC videos. Teenage creators aged 17 to 19 are earning $20,000 to $30,000 a month making a handful of ads. And as long as a great ad keeps spending, the creator does not need to make another video for a while.
13.) The top 20% of creators drive roughly 80% of ad performance. The best-performing creatives share a pattern-disruption quality: rough camera movements, an unexpected or messy opening, or a conversational tone that makes the viewer stop scrolling.
14.) YouTube requires its own dedicated creative program, separate from Meta UGC if you want to scale. A practical workaround for initial testing is to take story-format videos, drop them into a landscape static template with logos, ratings, and a CTA in Figma. And export that as a YouTube ad.
15.) At Whispr Flow, Google Ads was the single best-performing channel.
16.) When you hit audience fatigue on a channel, keep unlocking the next audience segment. Map your ICPs in order. Run creative and landing pages targeted at each one. Prove the economics. And then move to the next.
17.) Matt's approach to scaling budgets is to go as hard as possible until things break. Then pull back and use the failure data to learn where the ceiling is. Trying to figure this out gradually takes too long.
18.) If turning paid spend way down causes revenue to collapse, you have neglected the other half of the job: SEO, AEO, YouTube reviews, PR, and organic word of mouth.
19.) To run a successful referral program, the key is to make the reward tangible, immediate, and surface the offer at the exact moment of friction (such as when a user is about to hit a usage limit). Superhuman's "give one month, get one month" program generated 100s of referrals per power user.
20.) He sees free trial credits as a fully loaded marketing cost. If you are not adding the value of credits given away to your CAC calculation, you are not being accurate with your acquisition cost.
21.) AEO follows similar principles to SEO. But the most important external citations come from YouTube reviews, Reddit threads, and credible press mentions. AI crawlers pick these up when someone asks "what is the best tool for X." Traditional PR matters less for traffic and more for generating those external citations.
22.) Traditional PR like TechCrunch is still worth doing at launch. But mainly as a citation-building exercise for AEO rather than as a direct traffic driver. It is table stakes for credibility.
23.) TikTok and X ads have not worked for any of the 3 companies Matt has been at. X in particular is polluted with identical launch videos. X ads don't work. He is yet to meet a SAS head of growth or performance marketer that says that X ads print money.
24.) The biggest gap in teams is the absence of systems thinkers. Many people use AI to answer individual questions. But still do their jobs manually. A true systems thinker can map every input, output, and inter-relationship in their role. And identify exactly where agents can replace manual steps.
25.) Matt's interview test for systems thinking: Ask how a candidate uses AI workflows in their personal life, not just at work. A bad answer describes a single chat thread. A good answer describes a self-improving feedback loop where agent output is critiqued, refined, and eventually runs autonomously with only human approval gates.
26.) Matt built an autonomous sponsorship workflow at Whispr using Claude Code: the agent reads incoming newsletter sponsor emails. Negotiates rates. Researches audience quality. Drafts copy. Creates tracking links. Sends everything to the partner. Ingests performance data. And decides whether to renew, with only 2 human approval steps in the whole process.
27.) He recommends a "session end" skill inside Claude Code that automatically distills every working session into an Obsidian vault... logging decisions, open tasks, and learnings. This creates a compounding memory layer so future sessions can surface connections across weeks of work that a human would otherwise forget.
28.) His view on team composition is that most marketing teams should be dramatically smaller. Each role should be filled by one AI-native systems thinker who can manage what previously required 3-5 people. A single person with the right agentic setup could manage 100,000 influencers a month.
29.) His hiring hot take: probably fire most of your marketing team. The gap between A-players who have made themselves AI-native and B-players who have not is widening rapidly. And it's clearly visible in their output.
30.) Matt's broader prediction is that within 3 years, companies will function more like a board of directors. Humans will provide 20% strategic direction. And agents will handle 80% of execution. The marketing and growth leaders who understand how to architect and supervise those agent systems now will have a structural advantage over incumbents who do not.
if you're selling software:
> share your earned secrets for free
> charge for the software
if you're selling agency services:
> share your earned secrets for free
> charge for the service
if you're selling coaching/consulting:
> share your earned secrets for free
> customize earned secrets for your paying clients
who coined the term 'earned secrets'?
Ben Horowitz
his definition of an earned secret:
you did something in your past to solve a hard problem and learned something about the world that not a lot of other people know.
AI companies are going from 0 to $500M in a few months/years. But how?
Jeddi (@antinertia), who is the growth architect behind Polsia and HeyGen, has the answers.
I just watched his 53-min podcast with @wallstreetpapii.
Here are my top 20 takeaways:
1.) Gamma, Lovable, Replit... all of these product-led companies are using the same growth playbook to go from 0 to $500M+ in a few months/years.
2.) 4 Pillars of their growth playbook:
>Paid ads
>Influencer marketing
>UGC farms
>Product launches and in-house content
3.) Myth: If your product is good, it's going to grow by itself.
Maybe it was true 10, 15, 20 years ago... when you could put your app on the App Store and become #1. But not anymore.
There are 10 platforms and 100 products doing almost exactly the same thing.
3.) All the fastest-growing "product-led companies" are using paid ads. But they don't like admitting that publicly.
4.) Meta ads used to be about picking the right interests and behaviors. Now the game is volume of creatives.
5.) 60%-70% of your total acquisition can come from paid ads. Rest from influencers, affiliates, and owned content. But if you only rely on Meta, you are playing a dangerous game. They can update their algorithm. And that can put a hole in your growth.
6.) Influencer marketing means you are renting someone's audience and trust. You are giving them a flat fee. And they are making content about your product.
7.) YouTube influencers convert far better than TikTok and IG ones. Because YouTube viewers actually click description links.
8.) UGC farming is the game of volume. You know most reels won't go viral. But a few will. And they are enough to give you those revenue spikes.
9.) Study Higgsfield for viral feature launches.
10.) X ads were bad for a long time. Now they are worth testing again. He knows a guy who's printing with X.
11.) Outreach for influencer marketing: He gets like a 40% reply rate. Wants to get higher still.
12.) Outreach tool of choice: Lemlist (for multi-step LinkedIn and email sequences)
13.) Documenting founder's journey in daily or frequent video content can build massive trust and audience. Look at Ben from Polsia. He might become the first person to document a 0-to-billion-dollar journey.
14.) A personal brand compounds regardless of the business outcome. Even if the underlying company fails, a founder with a large audience still retains a lot of leverage and will get more opportunities.
15.) "Do I start content or business first?" His advice: if you have something to say, start saying it today. The primary goal is to earn attention and trust. Once you have that, you will find a way to monetize that.
16.) AI-generated UGC ads are now close to indistinguishable from real people (he helped a Texas insurance company print money with that).
17.) Find the shortest path to feedback from your actual ICP: for students, stand outside universities. For plumbers, call Google Maps listings. For digital products, use tools like Ahrefs to check and spot unmet demand.
18.) He validated a side business in the health space through tools like Ahrefs. Spotted strong demand with no strong supply of answers. Built the product and started generating revenue without ever doing a customer interview.
19.) Reddit can give you a useful signal for early idea validation. Because it's structured around Q&A. You can see problems people are trying to actively solve before you build anything.
20.) He is a big fan of World of Warcraft.
share this with your founder friend who has big ambitions.
>half-life of X posts: 52 minutes
>half-life of TikTok posts: 0 minutes (exceptions apply)
>half-life of Instagram posts: 18.27 hours
>half-life of LinkedIn posts: 23.22 hours
>half-life of YouTube posts: 10.60 days
post accordingly
They started with 2000 followers.
After 5 years, they have 6 million+ followers. And a Netflix documentary.
Bryan Johnson (the man who wants to live forever). She is the content genius behind that brand.
I just watched her 50-min podcast with Colin and Samir (@ColinandSamir).
Here are my top 20 takeaways:
1.) Kate Tolo (@_katetolo) sees the algorithm as a nightclub bouncer whose only job is to keep people inside happy.
2.) Do not focus on gaming the system. Focus on creating content that makes people feel something deeply (hope, shock, fear, etc.)
3.) Failure is invisible when you are small. So there's no real downside to posting more and experimenting.
4.) Test. Test. Test. Do not assume. Even when you're experienced... test. Because you don't know which headline or cta will win.
5.) Text-first approach even if you're creating for a visual platform. Get the story right in text first.
6.) She uses a 3-step writing process:
>State the idea as simply as possible.
>Rewrite it from the reader's perspective (what value it will give them).
>Identify the most "gossipy" or evocative element and move it upfront.
7.) Gossipy element just needs to be a remarkable and repeatable fact. Example: 90% of visible skin aging is caused by everyday sun.
8.) Know the "ticket price" for each platform. For YouTube, it's the thumbnail and title. For email, it's the subject line. For app notifications, it's the first 3 words.
9.) Create tension. And then give a lot of value. If a headline triggers fear or urgency, follow it with a practical solution or protocol. This makes the audience feel you're on their side. And not just there to alarm them.
10.) Speak simply. Example: an academic study about dog ownership and dementia can become "You a cat or dog person? Own a dog + exercise: 63% lower dementia risk. Own a dog + don't exercise: 0% benefit. Own a cat: 0% benefit."
11.) Repetition works. This same tweet... "Sleep is the most powerful drug available. For performance, mental health, and well-being." got 220k views (2025) and then 348K views when posted again (2026).
12.) When something works, iterate on it minimally. Find a post that outperforms your average. Make 1 small change. Republish. And see the result to isolate which element is driving the result.
13.) Review content in the right context. A tweet needs to be seen like a tweet with actual line breaks. A product label needs to be seen on the shelf next to competitors.
14.) Do not stockpile content. Publish fast. You are missing a learning cycle when you hold back a piece of content.
15.) Quote from Erik Torenberg: "Content either needs to be timely or timeless." Spending 2 weeks on content that's neither relevant nor built to last ends up being pure waste of time.
16.) Use "yes, and" approach to deal with negative or challenging comments.
17.) Viral content almost always produces both negative and positive reactions simultaneously.
18.) Emotional copywriting >> rational copywriting
19.) Taboo or uncomfortable topics are often the highest-value content territory. They deliberately cover topics like erectile dysfunction as a cardiac predictor that many others avoid.
20.) Authenticity and care for your audience are the key drivers behind virality at scale. Bryan spent 2 months drafting his relationship announcement because he sees the audience as his family.