Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. Not rough transcription you have to clean up. Actual polished text — ready for email, Slack, or any app.

Speak the way you think. Go on tangents. Change your mind mid-sentence. Flow strips the filler, fixes the grammar, and gives you text that reads like you spent five minutes writing it.

89% of messages sent with zero edits. Millions of professionals use Flow daily, including teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, and iPhone.

Start flowing free

In July 2025, David Protein started selling boiled cod.

Not a bar. Not a shake. Frozen wild-caught fillets, four to a box, its only retail door a single grocery store in New York, with the boxes wheat-pasted around the city like a band poster. The copy read: "Our commitment to protein has led us to a strange place. Boiled cod. We're selling it now."

The logic was almost mathematical. David's Gold bar gets 75% of its calories from protein. Boiled cod gets 92%. So the brand publicly bowed to the thing that beat it on its own metric. The company says Instagram engagement was up 1,000% in the first week and engagement on X was up 14,000%.

They ran it again in May 2026 with tinned cod, Renaissance packaging, and a launch film shot like a Steve Jobs keynote.

And two weeks ago, the playbook got one turn stranger. HallPass a separate low-sugar chocolate brand under Peter Rahal's Medici umbrella, not David itself put out launch images on August 11 riddled with AI-generated misspellings. "Suger." "Pacaage." Garbled text everywhere. When people piled on, the brand's cofounder Michael Tierney said the errors had been left in on purpose: "We left them there for a little bit of internet rage bait." His follow-up was two words: "It's worked." The product hits Walmart shelves this Saturday.

It works. It's fun. Half the operators we talk to have a version of it on a whiteboard somewhere.

It's also the cheapest thing David Protein has ever done, and it is not why the company is worth what it's worth.

About seven weeks before the cod went on sale, on May 29, 2025, David announced a $75M Series A led by Greenoaks at a $725M post-money valuation. (By this June, Inc. was calling it a billion-dollar playbook.)

A big chunk of that round did not go into marketing. It went into buying a company called Epogee. David has never disclosed the price.

Epogee makes EPG: a patented, chemically modified plant oil that behaves like fat in a formulation but is mostly non-absorbable. 0.7 calories per gram, against 9 for real fat. It is the entire reason a David bar can be 28g of protein at 150 calories and still taste like a candy bar. Without it, the product doesn't exist.

Here's the part that should stop you:

→ Epogee was, by virtue of its patents, the only commercial producer of EPG in the world.

→ David was ~90% of Epogee's revenue.

→ And David's demand was already running at 120% of Epogee's total capacity, headed for 150%.

Rahal's own words: "Last summer, we had to go out of stock on all items."

So one of the fastest-growing brands in the fastest-growing category in snacking was structurally short its own hero ingredient, and its only supplier was a small company that couldn't make more. They bought it, then expanded capacity five-fold.

Rahal has been blunt about the tradeoff: "Betting on EPG instead of just making another protein bar. It made the company much harder to build, but much more defensible."

And then the line that is really the whole issue:

That's the frame we keep coming back to with clients, and David is the cleanest illustration of it we've seen in years.

The cod bought trial. Attention, earned media, a reason for someone to reach past a $2 bar.

EPG bought repeat. The reason the second bar gets bought, and the third, at roughly $3.25 against a $1.70–$2.10 mass-market shelf (Sacra's estimate David doesn't publish it).

Last week we wrote about Alex Cooper shutting down Unwell Hydration one day after her media company was valued at $500M. That brand had the biggest audience in women's podcasting, a Nestlé manufacturing partner, and Target from the start. Maximum trial. It lasted under two years, because retail is a velocity business a shelf is a rented obligation to sell X units per store per week, and trial that doesn't convert to repeat gets reset out.

Same lesson, opposite end of the telescope.

And this month the market priced this distinction twice, in the same category, twenty days apart:

August 4 P&G agreed to buy Thorne for $3.8B. Thorne is not a hype brand. No Julia Fox campaign, no stunts. Its moat is the healthcare-practitioner channel and NSF Certified for Sport a third party vouching for it. Look at the arc: roughly $525M at its 2021 public listing, $680M when L Catterton took it private in 2023, $3.8B now. About 5.6x in three years for the least talkable brand in supplements.

August 24 Mondelēz launched Grenade nationwide in the US. Twenty grams of protein, one gram of sugar, an Oreo flavor. Per Nielsen data cited by Grenade, protein bars are now the fastest-growing category in US snacking, growing 4x faster than protein snacks overall. That's the number that pulls a $38B strategic into your aisle. Note how Grenade entered: GNC, Vitamin Shoppe, Bodybuilding.com and Amazon the specialty door first, borrowing credibility before buying velocity. And note its differentiator is flavor and a license, not a mechanism. That's the test David's moat is about to face.

The honest version of the David story, by the way, is not a victory lap. Whey has gone from $7 a pound to nearly $12 since they launched, and Rahal's stated plan is "just, like, survive versus changing anything." A labeling class action alleging the bars tested well above their stated calories was filed in January and voluntarily dropped in March without prejudice, and without any ruling on the merits, so nothing there has been settled either way.

And three smaller brands that built products on EPG sued over access in June 2025. A judge dismissed the case in February on market-definition grounds; the plaintiffs amended and it's still going. What's interesting for you isn't the outcome, it's David's argument: it said those brands could have had supply, but never signed long-term agreements the way other Epogee customers had "nothing but a case of bad business planning." The plaintiffs dispute that characterization, and no court has ruled on it.

Contested or not, that's the cheapest lesson in the entire story. Go find out today whether you have a written long-term agreement for your single most important input. Most brands your size don't.

Right. Nobody reading this is spending a $75M Series A on their supplier. Vertical integration at $3M in revenue is a fantasy, and any newsletter telling you otherwise is selling you something.

But the principle trial is rented, repeat is owned is size-independent. And at your size, the repeat side of the business isn't a supply chain. It's your subscription mechanics, your cancel flow, and whether you can actually read your own numbers.

So here's what we've found running retention for subscription brands this year. These are our accounts, anonymized, with the caveats intact.

Mechanism one: skips. Most cancel flows offer "skip your next shipment" as the first save. A skip isn't counted as a cancel, so your churn number looks great. In one subscription brand we run, we pulled the cohorts and found the skippers were mostly just... cancelling later. The current month looked clean. The liability moved to next month.

Mechanism two: the denominator. This one is worse, and it's the trap every brand scaling paid falls into. When you're pouring new subscribers in the top, your churn rate falls even if your churn behavior is identical the denominator is growing faster than the numerator. Then ad spend comes down and the real number surfaces.

The client's own analyst said it better than we could: "I think we got a false positive. Churn rate went down to 8% because it includes basically the old churn rate but was masked by the massive influx of new subs. Now that ad spend is down, it's kind of exposing that we didn't necessarily change it that much."

Do this: stop reporting churn as a rate. Report it by cohort of the subscribers who joined in March, how many are still active in month 2, month 3, month 6? A cohort curve can't be diluted by acquisition. A rate can.

We dug into which cancel-flow offers produced the longest-lived subscribers on that same account. The gap was not close:

→ Offered a skip: roughly 20% made it to their next shipment.

→ Offered a discount: roughly 75% did.

Honest caveat, because it matters: the discount cohort was small when we pulled it, so treat the whole comparison as directional rather than gospel the gap is real, the exact figures will move. We changed the splash-screen offer from "skip" to "get 20% off your next two orders" on August 6 and the early read is positive more reactivations, more recoveries but we don't have a full churn cycle on it yet. We'll tell you when we do.

The structural finding is the more durable one: the splash screen is where all the juice is. It's the first thing someone sees when they click cancel, and it's where essentially all the saves happen. Everything after it is scraping the bottom of the barrel. If you optimize one thing in your retention stack this quarter, optimize that single screen.

Related, and free: build dynamic offers keyed to the stated cancel reason. If someone says "too expensive," offer money. If they say "too much product," don't Skio's own data says about 1% of "I have too much product" cancels ever resubscribe, so that person needs a winback in eight weeks, not a discount today. On one account we inherited, the flow was making no offer at all.

Klaviyo showed about 10% of total revenue attributed on a subscription-heavy account we run. Our read is the real contribution is closer to 20%, because roughly half of that business is recurring orders Klaviyo doesn't claim credit for the retention work keeps the subscriber alive, and the subscription platform books the order. Add SMS and we'd put the combined number at 25–30%.

That 20% and 25–30% are our estimates, not measured figures. But the direction is not in question, and it matters because we've watched founders cut retention budget on the strength of a 10% number that was never the real number.

Loyalty credits. We sent one campaign to a small segment holding more than 100 points and it did ~$5k from email alone at 5x our average click rate on credit those customers already had. The pattern in the cohort data: people who redeem credits spend materially more and cancel far less often.

It reads like a promo. It isn't. Nothing was discounted; we just reminded people what they already owned. We now run it monthly, put a dynamic "you have $X waiting" block across the flows, and added "award credits" as a save in the cancel flow.

→ Smart Sending. We found it switched on inside an abandoned-checkout email, which meant that email was skipping nearly every profile. An entire message in a revenue flow, silently not sending. Go audit every flow message for it.

→ Conditional splits. If your flows use nested conditional splits to filter out purchasers, rip them out and put filters on each individual message instead. Splits create a customer journey nobody on your team can read, which is how a flow stays broken for six months.

→ Your pop-up question. If your pop-up asks something you can't tie back to LTV, you're collecting decoration. On one apparel account, segmenting by the answer to "what are you shopping for?" surfaced a cohort spending 40%+ more over their lifetime with an 80% higher AOV and it was one of the smallest audiences on the list. That's not a segmentation insight, that's an acquisition targeting instruction. On another account the same report came back flat, and the right move was to change the question, not the flow.

MUD Jeans, the Dutch circular-denim brand, went bankrupt in early August the CEO announced it on August 5. Projected 2026 revenue: about €3.1M. That's a brand most of you can see yourselves in.

Here's what makes it worth your ten seconds: the turnaround had worked. By the company's own account it reached positive EBITDA and break-even in Q1 2026. And it got there by killing "Lease A Jeans" the famous subscription model that made them talkable in the first place and moving into retail, repair and verified resale. The CEO's explanation: "the financial burden of historic debt ultimately proved too great to overcome."

Profitability is a P&L event. Solvency is a balance-sheet event. Fixing your unit economics does not retire debt you took on at worse ones. And if your subscription requires you to own the asset the customer is using, you're a leasing company with a brand attached, and you need leasing-company money.

Here's the thing. We're genuinely obsessed with this stuff. Graham and I spend our days inside these accounts, and most of what's in this issue came out of a normal Tuesday.

So: hit reply.

Not a form, not a Calendly, not a funnel. Reply to this email and tell us what you're working on whether you're trying to crack the second purchase, unlock the next level of growth, or just get your attribution to stop lying to you.

We're not going to pitch you. We'd love to work with you if it's a fit, and we'll say so plainly if it isn't. But mostly we'd like to be friends of your business. Tell us what's broken and we'll tell you what we've seen work, and if it turns out we're the right people to help, we'll figure that out later.

If enough of you reply, we'll pull a few operators together on a call next month and just talk shop no agenda, no deck. Say the word in your reply if you'd want in on that.

Parker & Graham

Sources: AgFunderNews on the EPG capacity expansion · AgFunderNews on the Epogee acquisition and Series A · Businesswire: David closes $75M Series A · Inc. on the David playbook · Inc. on the cod · Bakery&Snacks: Rahal on why food tech beats branding · Fortune on protein costs and GLP-1s · Modern Retail on the EPG lawsuit · AgFunderNews on the February dismissal · Inc. on the class action being dismissed · Sacra on David's pricing · HallPass and the rage-bait typos · Nosh: HallPass announcement · CNBC: P&G to acquire Thorne · Nutritional Outlook on the Thorne deal · Food Dive: Mondelēz launches Grenade in the US · NL Times: MUD Jeans declared bankrupt · Modern Retail on celebrity brand shutdowns