Paul C. Chambers

Recurring revenue guru. Helping leaders and companies find purpose in the $2-trillion subscription economy via @Sub_Summit. Firefighter in my spare time.👨🏻‍🚒

[Paul C. Chambers @paulchambers]

You can't retain your way out of acquiring the wrong customer. A subscriber you bought on a discount hunt was always going to churn, no matter how good your flows are. The cheapest retention lever most brands ignore is who they target in the first place.

[Alex Wyatt | Meta Ad Creatives @Alexwyatt47]

Customer acquisition costs have increased 40-60% since 2023. Average CAC is now $68-$84. 88% of subscription brands are seeing it climb even further.

But the answer to rising CAC isn't spending less on acquisition. It's spending smarter.

60% of DTC revenue comes from returning customers. Loyal customers convert at 60-70% vs 5-20% for new prospects. Retention costs 5x less than acquisition.

So the question isn't "how do we lower our CPA." It's "are we acquiring customers who are actually worth acquiring."

Which product are you pushing? Which desire are you targeting? Which landing page are you sending them to? All of that determines whether someone buys once and disappears or comes back in 60 days and spends 3x more.

Spend aggressively on acquisition. But spend intentionally. Push the products that create repeat buyers. Target the desires that attract high-LTV customers. The brands winning right now aren't spending less. They're acquiring better.

[Paul C. Chambers @paulchambers]

@ecom_nicolas Measuring a retention agency on opens and attributed revenue lets it grade its own homework. The number that matters is whether order two happens without you paying to acquire the same customer twice. Everything else is email production, not retention.

[Paul C. Chambers @paulchambers]

@AdamKitchen_co The cancel flow is the most honest research you own. People show you exactly where the value stopped. And a pause beats a discount every time, because a discount just reschedules the same cancel 30 days out.

[Paul C. Chambers @paulchambers]

@resexhof The reorder feeling automatic instead of a monthly decision is the whole skill. People don't cancel a routine, they cancel a purchase they keep having to re-justify. Telehealth just makes that math impossible to ignore.

[Paul C. Chambers @paulchambers]

It doesn't even drive me crazy anymore how expensive Quickbooks has gotten. What drives me more crazy is the ad for additional services they sell me on every click.

Give me an easy conversion path, feature parity, and SOME cost savings and I'm switching so fast.

[Paul C. Chambers @paulchambers]

@KodyNordquist Of everything on this list, the replenishment offer is the one that compounds. An accepted upsell is a one-time margin bump. An accepted subscription changes the customer's default from deciding to staying. Same 60 seconds, very different LTV.

[Paul C. Chambers @paulchambers]

@yojimmykim Time to value is the whole game in subscription. The clock that matters is felt value vs the next charge. Bill before they've felt the first one work and you've created the cancel yourself. Pace delivery to the win, not the calendar.

[Paul C. Chambers @paulchambers]

@ruben_vdzr Broken attribution is real, but most brands fix the front of the funnel and never measure the back. You can argue all day about which ad gets credit for order one. Order two either happens or it doesn't, and that number isn't fiction. Retention is the cleanest data you own.

[Paul C. Chambers @paulchambers]

@HNilsonne @Neko 100K is the number everyone celebrates. The one that decides the next 100K is whether scan two feels more valuable than scan one. Members don't renew on the mission, they renew on the felt upgrade each cycle. Looks like that's exactly what you're building.

[Paul C. Chambers @paulchambers]

The lesson here isn't sales. It's coverage. Every team has a 40% problem: the win-back that never goes out, the churn signal nobody works, the reorder nudge that waits till next quarter. Agents win where there's a coverage gap, not a skill gap.

[Jason ✨👾SaaStr.Ai✨ Lemkin @jasonlk]

Our 1.25 humans + 20 AI agents closed 140% of what our all-human sales team did last year.

That's a real number. But I'm not sure it tells you what you think it does.

Here's what we actually changed:

1️⃣ First, we concentrated every qualified lead into our best closers

When you go from a bench of mixed-quality reps to 1.25 humans who are your best people, close rates go up. That alone probably moved the number 15-30%. Maybe more.

2️⃣ Second, AI agents gave us something we never had: 100% coverage

Our human team responded to less than 40% of inbound leads. Not because they were lazy. Because humans have meetings, they eat lunch, they sleep, they take PTO. Our AI agents respond to 100% of inbounds within minutes. Every single one.

3️⃣ Third, the AI agents worked every past prospect in our CRM

Thousands of old leads that no human rep wanted to touch. No cherry-picking. No "I'll get to those next quarter." The agents just worked the list. All of it.

4️⃣ Fourth, AI as a market exploded

Our content is now all AI agent content. Our community is AI + B2B. Demand went up because the market went up. That's a tailwind, not a strategy.

So did we close 140% because of AI agents?

Partly.

Did we close 140% because we concentrated talent?

Partly.

Did we close 140% because the market was on fire?

Partly.

We can't run the A/B test. I can't go back in time and try the same period with a full human team selling into the same AI-fueled market. Nobody can.

What I do know:

20 AI agents + 1.25 humans costs a lot less than a full sales team. 140% of last year is 140% of last year, whatever the cause. And 100% response rates on inbound beats 40% every single time, regardless of who or what is doing the responding.

If you're thinking about deploying AI agents in sales, don't do it because you think AI is magic at closing deals. Do it because you have a coverage problem you can't solve with humans alone.

That's the real unlock. More at-bats for your best batters. Not better batters.

[Evan Luthra @EvanLuthra]

🚨TESLA JUST FOUND A WAY TO BUILD THE WORLD'S BIGGEST AI SUPERCOMPUTER WITHOUT BUILDING A SINGLE DATA CENTER

The answer was sitting in millions of driveways the whole time… your parked car.

The entire AI industry has hit a wall.. And it's not chips.. It's power..

Building AI data centers now means waiting years for grid connections.. The Stargate project from OpenAI and Oracle is spending up to $500 billion to build 7 gigawatts of capacity.. And it'll take years to come online..

Tesla just realized it already has 7 gigawatts.. Sitting in its Supercharger network.. Already built.. Already connected to the grid.. Already permitted..

So on June 18, 2026, Tesla quietly filed a trademark for something called MEGAPOD.. Modular AI data center hardware designed to drop straight into existing Supercharger sites..

No land to buy.. No years-long grid queue.. No new power plants.. They just bolt compute onto infrastructure they already own..

But that's the small idea..

Here's the radical one..

The average car sits parked and unused about 95% of its life.. And every modern Tesla already has a powerful AI chip inside it.. Built for self-driving..

So Tesla wants to link millions of parked cars into one massive distributed supercomputer..

The math is staggering.. If Tesla hits 100 million vehicles, and each contributes about 1 kilowatt of compute.. That's 100 gigawatts of AI processing power..

That dwarfs every data center on earth combined.. And the real estate, the power, and the cooling were all already paid for.. By the people who bought the cars..

Your Tesla is liquid-cooled.. Plugged in overnight.. Doing nothing.. It's basically a sleeping computer in your garage..

And Tesla's plan is to let you rent it out..

Owners could earn passive income, free Supercharging, or discounts on Full Self-Driving in exchange for leasing their car's idle computing power while they sleep..

Your car stops being a depreciating asset.. And starts earning money while parked..

This is the part competitors can't copy..

OpenAI has to spend half a trillion dollars and wait years for power.. Tesla already has the grid connections, the batteries to stabilize them, the chips, and millions of cooled computers sitting idle in driveways worldwide..

Everyone else is trying to build a giant brain in one place..

Tesla is turning the entire planet into one.

[Paul C. Chambers @paulchambers]

@_umair_sheikh The skip option is the whole game. A pre-charge email that only says 'you're about to pay' takes control away. One that says 'skip, delay, or swap this one' hands it back. Same email, opposite outcome. Control keeps people subscribed, not lower prices.

[Paul C. Chambers @paulchambers]

@karri_tweets @alecsandrull Two reasons it's brutal: great beans are available everywhere, and most roasters ship faster than people drink. You end up paying for a pile of bags on the counter. The subs that survive sell discovery and rotation, and they pace delivery to actual consumption.

[Paul C. Chambers @paulchambers]

The biggest AI line item here isn't engineering. It's client delivery and operations. That's the tell. AI stopped being something you build with and became something you run the business with. Spend landing in the work, not the dev team, is adoption that actually stuck.

[Taylor Holiday @TaylorHoliday]

Here's our AI spend at CTC, broken down by department year to date.

$31K/month. $276/person average.

Our biggest spender isn't engineering.

It's client delivery at 44.5% of total spend.

Didn't expect that when we started tracking it. But it’s ideal.

I want costs to follow revenue as much as possible.

Our growth team (we call them Prophit Engineers) manages client P&Ls, forecasts, creative strategy, media allocation, measurement and more.

$13,759/month. Nearly half of all AI spend.

Prophit Engine: $320
Technology: $222

Heaviest per-capita users are operations and leadership. Nobody mandated that. It just happened.

Every single one uses AI to some degree.

Retention: $80/person
TikTok Shop: $94/person
Finance: $56/person

Some of these are tiny. But the breadth is interesting. Adoption spread on its own once a few teams proved it out.

Honestly not sure if that's high or low relative to other companies our size. We're still figuring out the right expectations.

But the pattern we keep seeing: AI spend concentrates where revenue gets generated.

Sharing because I wish more companies published this data.

I saw @marty_kausas do a post and found it very helpful.

What are you all seeing?

[Taylor Holiday @TaylorHoliday]

Here's our AI spend at CTC, broken down by department year to date.

$31K/month. $276/person average.

Our biggest spender isn't engineering.

It's client delivery at 44.5% of total spend.

Didn't expect that when we started tracking it. But it’s ideal.

I want costs to follow revenue as much as possible.

Our growth team (we call them Prophit Engineers) manages client P&Ls, forecasts, creative strategy, media allocation, measurement and more.

$13,759/month. Nearly half of all AI spend.

Prophit Engine: $320
Technology: $222

Heaviest per-capita users are operations and leadership. Nobody mandated that. It just happened.

Every single one uses AI to some degree.

Retention: $80/person
TikTok Shop: $94/person
Finance: $56/person

Some of these are tiny. But the breadth is interesting. Adoption spread on its own once a few teams proved it out.

Honestly not sure if that's high or low relative to other companies our size. We're still figuring out the right expectations.

But the pattern we keep seeing: AI spend concentrates where revenue gets generated.

Sharing because I wish more companies published this data.

I saw @marty_kausas do a post and found it very helpful.

What are you all seeing?

[Paul C. Chambers @paulchambers]

@markdmei 'Renting revenue' nails it. The number under RPR is order 1 to order 2. Most brands with weak RPR aren't losing loyal buyers, they're losing everyone right after the first purchase. Win the second order and the rate fixes itself.

[Paul C. Chambers @paulchambers]

Appears that @Lunchly was short-lived?

As a parent I can tell you this was never a good product to begin with. Still just as unhealthy for the kids, if not worse, and a terrible value.

If in fact on its way out, I’m happy with that.

[Austin Rief ☕️ @austin_rief]

Mr. Beast and Logan Paul just launched a Lunchables competitor called Lunchly.

Each box will contain PRIME and Feastables.

This is going to crush it.

[Paul C. Chambers @paulchambers]

@IstvanicMarin The fix most brands miss: a discount on the first order anchors your price forever. A discount that rewards the second order builds the habit instead. Same margin spent, opposite signal. One teaches 'this is cheap,' the other teaches 'this is worth reordering.'

[Paul C. Chambers @paulchambers]

@ecommcowboy @damondidier That's an acquisition truth. Subscription flips it. Your audience already said yes and the offer already closed. The email's only job now is to make the next charge feel earned, not to sell again. Most brands keep blasting offers at people who already bought.

[Paul C. Chambers @paulchambers]

@RomanEcom Cleanest case for recurring revenue I've seen all year. That multiple was never paying for revenue. It pays for revenue that comes back without you buying it again. Paid acquisition rents your top line. Real rebill behavior is the part you actually own.