Road21 blog

Harry's: 100,000 emails before the first razor

Gaurav — Tue, 27 Oct 2026 00:00:00 GMT

Harry’s collected 100,000 email addresses in seven days before it sold a single razor. No ads, no press tour. A two-page website, a team of twelve, and a pile of free shaving product. Co-founder Jeff Raider later published the entire playbook himself: the code, the email templates, the reward tiers, and the referral data. This is the campaign, the numbers, and where the company went from there.

The background. Before Harry’s was a company, Raider and Andy Katz-Mayfield spent the better part of two years researching the men’s shaving market. They found the nearly century-old German manufacturer that would make their blades, worked with industrial designers on a handle inspired by pens and knives, and laid the groundwork for a direct-to-consumer brand. Raider had co-founded Warby Parker before this. The central worry, in his words: “Were we going to launch to crickets?” A direct-to-consumer brand without customers is nothing, and Harry’s had no customers, no stores, and no audience.

The insight that shaped the campaign was about credibility. “The most powerful and effective way to be introduced to our new company was through a credible referral,” Raider wrote. They had watched friends do it: Everlane and Fab had both grown through referral mechanics. And the founders had spent months telling the Harry’s story to friends and entrepreneurs, building a list of maybe a couple hundred people. The campaign was designed to turn those people into a distribution channel.

The campaign. The whole thing was a two-page microsite, live for one week before the store opened.

Page one was a splash page. It showed a razor, the line “Respecting the face, and wallet since like right now,” a small drawing of a key, and a button that said “STEP INSIDE.” Nothing explained what Harry’s was. The mystery was deliberate: the first customers were insiders, and the copy made them feel like insiders. The only thing you could do on the page was enter your email address.

Page two was the referral engine. After signing up, you got a unique link back to the splash page plus one-click buttons to share it by email, Facebook, or Twitter. The pre-written tweet was simple: “Excited for @harrys to launch. I’m going to be #shaving for free.” A wooly mammoth carried the message: “Shaving is evolving. Don’t leave your friends behind.”

Then the rewards. “Invite Friends and Earn Product” was the one line on the page with no mystery in it. Five referrals earned a shave cream. Ten earned a razor. Twenty-five earned a shave set with the premium Winston handle. Fifty earned a year of free shaving. Two design decisions matter here. The first reward came easy, at five friends, so everyone could win something. And the team killed a planned tier of a lifetime of free product at 1,000 referrals, worrying it would make the game feel unwinnable. A progress tracker showed each user how many friends had signed up and which prize came next. Raider says some friends took the whole thing like a personal challenge.

The mechanics underneath were just as considered. The email address doubled as the tracking ID, so one field powered the whole referral graph. Every signup got a transactional email through SendGrid, and a bounce marked the address as fake. IP blocking stopped any single IP address from signing up more than two emails, to stop gaming. Unsurprisingly, Raider notes, most of the fraud showed up in the highest reward tier.

The seeding. A referral campaign needs someone to refer first, so the team of twelve seeded it themselves. Days before launch, every employee sorted their contacts into two Gmail groups: people who knew Harry’s and people who didn’t. They wrote personal, friends-and-family emails in advance, each with an explicit ask to spread the word, some with a product shot and the phrase “Harry’s is coming.” They wrote everything a day or two early so the team could simply hit send the moment the site went live, and set up email signatures linking back to the prelaunch page before mailing anyone. The founders also asked specific friends individually, especially the ones who could share the link with entire companies. They launched the Facebook page and Twitter handle the same day and answered mentions from personal accounts. Press was deliberately avoided. A referral from a publication would break the premise that the recommendation came from someone you know.

The results. Before launch, the team wrote their guesses on a whiteboard: 3,000 emails. 5,000. 7,500. One person said 15,000. They passed the highest guess on day one. By the end of the week they had collected, by their own count, over 85,000 valid email addresses, and over 100,000 in total. The launch week that followed brought a flood of tweets, emails, and, Raider’s favorite, customer phone calls.

The referral data is the part worth studying:

77% of the emails came through referrals. About 20,000 people referred about 65,000 friends.
The average referrer brought in more than 3 friends.
More than 200 people referred over 50 friends each and hit the top tier, mostly people close to the founders with big followings or access to company-wide email blasts.
About 3,000 people earned free product.

A week later, Harry’s opened its store to an audience that was already waiting. Reward winners got coupon codes redeemable on the live site, which funneled the most engaged referrers straight into the full catalog. On the first day of business, every employee manned the support queue, answering customers by email, phone, Twitter, Facebook, and text.

What came beyond. Harry’s went on to become one of the defining direct-to-consumer brands of the decade. In 2014, Harry’s raised over $100 million and bought the 94-year-old German factory that makes its blades, securing its own supply chain. It moved from internet-only into retail stores in 2016. In 2019, Edgewell, the owner of Schick, agreed to buy Harry’s for $1.37 billion. The Federal Trade Commission sued to block the deal in February 2020, describing Harry’s as a “critical disruptive rival” whose rise had forced the razor duopoly to cut prices. The sale was abandoned. A year later, Harry’s raised $155 million at a $1.7 billion valuation. By 2024 it had filed confidentially for an IPO, reportedly nearing $1 billion in annual revenue and profitable, and by then its brands included Flamingo, Lume, Mando, and Cat Person.

Why email worked. Every mechanic did two jobs at once. The email address captured the lead and tracked the referral. The reward was free Harry’s product, so every prize put the product in a future customer’s hands. The coupon code that delivered the reward also delivered the winner to the store. The one-week window manufactured urgency that a standing referral program lacks. And every recommendation arrived from a friend, which was the premise of the whole design: a credible referral instead of an ad.

One honest caveat, again from Raider: it is, he wrote, “very difficult to attribute its success to one specific variable,” whether the code, the tactics, or the idea. Harry’s also had a strong product, good timing, and well-connected founders.

The lesson for builders. You do not need customers to build an email list. You need a reason for people to tell their friends, a reward worth winning, and a first tier easy enough that everyone plays. Harry’s proved a list built before launch can be the demand at launch. Those 100,000 addresses were the first version of the business.

Building an email list before your product launches? Road21 runs email marketing, product-led growth, positioning and outbound for products that need to grow. Book a call and we’ll look at yours.

Sources

Jeff Raider (Harry’s co-founder), “How to Gather 100,000 Emails in One Week,” Tim Ferriss’s blog, July 21, 2014: https://tim.blog/2014/07/21/harrys-prelaunchr-email/
Federal Trade Commission, “FTC Files Suit to Block Edgewell Personal Care Company’s Acquisition of Harry’s, Inc.,” February 3, 2020: https://www.ftc.gov/news-events/news/press-releases/2020/02/ftc-files-suit-block-edgewell-personal-care-companys-acquisition-harrys-inc
Forbes, “Harry’s Raises A $155 Million Series E At $1.7 Billion,” March 31, 2021: https://www.forbes.com/sites/stevenbertoni/2021/03/31/exclusive-harrys-raises-a-155-million-series-e--at-17-billion-a-year-after-the-ftc-blocked-its-billion-dollar-sale-to-edgewell/
CNBC, “Razor maker Harry’s files for IPO, sources tell Reuters,” March 7, 2024: https://www.cnbc.com/2024/03/07/razor-maker-harrys-files-for-ipo-sources-tell-reuters.html

How to write positioning for a product buyers don't understand

Gaurav — Tue, 27 Oct 2026 00:00:00 GMT

You explain your product. The buyer nods, then asks, “So what exactly is it for?”

That reaction often means the buyer put your product in the wrong category, or had no category for it. April Dunford, author of Obviously Awesome, has written a lot about this problem. This post uses her blog posts as the source, quotes a few key lines, and paraphrases the rest with credit. The working order below is my adaptation of her material, explained in a moment.

The worked example is hypothetical. I made it up for this post. It is not a client, and it is not one of Dunford’s cases. Her own examples are credited to her where they appear.

What positioning is

Dunford’s definition: “Positioning defines how your product is a leader at delivering something that a well-defined set of customers cares a lot about.”

She also separates it from things people confuse it with. In “An Introduction to Positioning” she says positioning is not messaging, a tagline, a brand story, or a vision. Messaging comes after. In “A Product Positioning Exercise” she says finished positioning becomes an input to the messaging document.

Why buyers get confused

Dunford describes positioning as context. Naming the market for your product, she writes, sets off assumptions about who it competes with, what features it should have, who it is for and what it should cost. Tell someone your product is “email” and they expect different things than if you say “chat.”

So a confused buyer has usually received one of two things: a category that points at the wrong competitors and the wrong price, or no category at all.

She gives examples of the first kind from early in her career. In one, a startup built by database experts made a fast analytics product and sold it as a database. Prospects told them they already had Oracle, and many would not take a meeting. The team repositioned the product as a business intelligence tool, which put fast analysis of large amounts of data at the center. In another, a company sold an enterprise CRM against Siebel, the market leader, and lost the comparison on almost every measure. In her account, the company then narrowed to CRM for investment banks, where its one different feature mattered most. In Dunford’s account, the company grew from under $2M to close to $80M over the next 18 months.

The order of work

Dunford lists five components of positioning: competitive alternatives, differentiated capabilities, value for customers, target customer segment, and market category. In the quickstart guide she explains that the order matters, and that she found the starting point has to be competitive alternatives.

Her separate exercise post has its own five steps: let go of where the product came from, isolate what is unique, find the value, find who cares most, and choose the market frame of reference.

I have merged these into six steps. That merge is my adaptation, not an official Dunford method. Step 1 comes from her exercise post, step 2 from the quickstart, and steps 3 to 6 follow the order both posts share.

The hypothetical product

To keep the example honest, here is everything the hypothetical product does, stated before any reasoning starts:

It is an invoice-reminder app for freelance designers.
It records invoices the freelancer has sent, with due dates.
It flags invoices that are overdue.
It drafts a reminder email for each overdue invoice. The freelancer edits and approves it before it goes out.
It lets the freelancer set how many days after the due date each client gets a reminder.

It does nothing else. It has no bookkeeping, tax reports or bank feeds. The founder’s original idea was “simple accounting software for freelancers,” so that is the category the app sits in today.

Step 1. Let go of where the product came from

Dunford says weak positioning is often the result of holding on to a default market position rooted in the product’s history.

For the hypothetical app, the history is accounting software. A buyer who hears that term expects bookkeeping, tax reports and bank feeds, compares the app with well-known accounting tools, and finds it thin. The founder’s first job is to stop defending that label and describe what the app does today: it helps a freelancer chase late invoices.

Step 2. Competitive alternatives

Dunford’s test question: “What would a customer do if your offering didn’t exist?”

She warns against listing every company that could compete. She calls those “phantom competitors,” meaning companies that could compete in theory but that you never meet in deals. She advises listing what your customers actually see as alternatives. Doing nothing counts. In her words, that means sticking with the current way of solving the problem, such as a spreadsheet, a manual process or an intern.

For the hypothetical app, the likely alternatives are a spreadsheet of invoices, calendar reminders, and writing each chaser email by hand. A large accounting suite goes on the list only if freelancers say they considered one when choosing.

Step 3. What is unique

Dunford says to list what you have that the alternatives do not, without yet judging whether anyone cares.

Compare the app’s five capabilities with the manual setup of a spreadsheet, calendar reminders and hand-written emails. Two things stand out. The app drafts the overdue email for the freelancer to approve, where the manual setup needs a new email written each time. And it keeps invoices, due dates, overdue flags and per-client reminder timing in one place, where the manual setup spreads them over a sheet and a calendar. A determined user can build parts of this with spreadsheet formulas or automations, so the difference is effort and convenience rather than something a sheet could never do.

Step 4. Value

Dunford writes: “Customers don’t care about your unique features, they care what those features can do for them.”

Turn each unique item into something a customer would notice. A drafted email that the freelancer approves means no blank-page moment when writing to a client who is late. One place for invoices and timing means less checking across tools to see who owes what.

Step 5. Who cares most

Dunford writes: “Many kinds of prospects care about what you deliver but only some of them REALLY care.” She suggests asking who you would focus on if your product’s future depended on selling as much as possible this month. Her reasoning is that well-suited customers understand your value, buy faster and ask for fewer discounts.

For the hypothetical app, a reasonable first guess is freelance designers who send many invoices and have regular trouble getting paid on time. Someone with one retainer client has little use for reminders. This is a guess within the example. With a real product you would check it against real sales conversations.

Step 6. Market category

Dunford calls this the hard part: which market frame makes your value obvious to your best prospects. She notes it often is not the market the idea started in.

For the hypothetical app, three candidates:

“Accounting software for freelancers” invites bookkeeping comparisons the app cannot win.
“Invoicing app” invites comparison with every tool that creates and sends invoices, which the app does not do.
“Invoice reminder tool for freelancers” matches what the app does and who needs it.

Dunford advises caution about inventing a brand-new category. In the quickstart guide she argues that category creators often lose in the long run to later companies, and that it is more common for startups to begin in an existing category. Her example of a good fit is “embeddable database for mobile devices”: buyers already understand “database,” and the added words narrow it.

Following that pattern, “invoice reminder tool” is an existing, understandable category narrowed by “for freelancers.” I have not tested whether it is the best choice. Only conversations with real buyers can show that.

A positioning description you could test

Putting the six steps together, here is one hypothetical description for the hypothetical app:

A reminder tool for freelance designers who often wait on late invoices. It tracks what you have sent, flags what is overdue, and drafts the reminder for you to approve, so you can follow up without writing each email from scratch or checking a spreadsheet and calendar.

It uses only capabilities defined above. It names the category, the buyer, the manual alternatives it replaces (the spreadsheet, the calendar and the hand-written email) and the value. It is a first draft to put in front of buyers.

Write it down, then change the business to match

Dunford recommends recording the result in a dated document so you can see why the positioning is what it is and when it was last reviewed.

She also warns that it is bigger than a copy change: “A change in positioning is a fundamental change in your business.” Buyer expectations about competitors, price and sales channel shift, and internally the roadmap, skills and partnerships have to follow. For the hypothetical app, moving from “accounting software” to “invoice reminder tool” would change what gets built next, with more work on reminder timing and email drafts and none on tax reports.

Test it, and expect it to age

Dunford describes the output as a best guess to test with customers. She also writes that “Great positioning often becomes weak over time,” because markets move and unique features become standard.

Skip the positioning statement

Dunford argues against the fill-in-the-blanks positioning statement. Her reasons in “A Product Positioning Exercise”: it assumes there is one way to position a product when there are several, it produces an awkward sentence nobody uses, and it gives a false sense of being finished. Her alternative is the sequence above, worked out with evidence from customers at each step.

Your turn

Take a sheet of paper and write the six headings. Ground your answers on alternatives and unique value in what customers say, not in what your team believes. If you cannot list your real alternatives, finding that out is the first job.

If you want a second pair of eyes on your answers, you can book a 30-minute call with me.

Sources

All quotes and paraphrased claims about Dunford’s views come from her own writing:

“A Quickstart Guide to Positioning”, March 15, 2021: https://www.aprildunford.com/post/a-quickstart-guide-to-positioning
“An Introduction to Positioning”, May 4, 2020: https://www.aprildunford.com/post/an-introduction-to-positioning
“A Product Positioning Exercise”, October 17, 2019: https://www.aprildunford.com/post/a-product-positioning-exercise

Obviously Awesome is the book these posts summarize and is not quoted directly. The invoice-reminder app is hypothetical, created for this post.

Viktor: the team-wide adoption loop

Gaurav — Tue, 27 Oct 2026 00:00:00 GMT

In May 2026, Fryderyk Wiatrowski closed his talk at a New York AI conference the way a salesman would: a QR code on the screen, and the line “Everyone in this room has a hundred dollars in free credits. No string is attached.” Three months earlier, his company had launched what it treated as an experiment. It was no longer one.

Background. Wiatrowski studied math at Oxford, worked as an engineer at Meta, and spent time in quant trading. His co-founder, Peter Albert, had joined Meta’s AI ranks for one specific reason: to find someone to start a company with. Their evenings went into agents that could automate tedious knowledge work. The team now works from Warsaw and Munich, behind a Delaware entity called Zeta AI, Inc.

History. Viktor is their fourth swing. The first, in 2023, was a browser agent that, on Wiatrowski’s telling, reached the top of the Web Arena benchmark but completed only “three to five steps reliably,” with sixty percent reliability compounding at each step. Impressive on a leaderboard, unusable as a product. The second, JACE, was announced in June 2024 with a $2.9 million pre-seed from Daniel Gross and Nat Friedman. The third, Jace, was an email agent launched in February 2025; it still runs. Then, in February 2026, they put a fourth experiment into early access: an AI employee that lives in Slack and Microsoft Teams. “Zero expectations of growing at all,” Wiatrowski says. “It was actually an experiment, and it surprised all of us.”

The layup. Viktor is not a chatbot you visit. It sits inside the company workspace and does jobs across the tools a team already uses: pulling reports, auditing ad accounts, following up on CRM entries, building small internal apps, running scheduled automations. The company claims 9,800+ integrations. The pitch is blunt: “Viktor is not a tool. It’s a hire.”

Putting it in Slack was the first sharp decision. “You don’t interact with human employees in web apps,” Wiatrowski points out. “You interact with them in Slack.” And chat resets expectations: “If you ping someone on Slack and tell them to build an app for you and get an answer in 10 minutes, you are shocked. No teammate has ever built you an app in 10 minutes.” A wait that would feel broken on a web page feels fast in a chat window.

The loop. Then comes the mechanism this episode is about. Viktor is a company agent, not a personal one. “It’s sufficient for one person from the company to connect an integration,” Wiatrowski explains. “Viktor will inherit the permissions… and then the whole team has access to it, so you don’t need to connect them a hundred times.”

The pricing pulls the same way. There are no seats. A team pays $100 a month for 40,000 shared credits, on the principle of “you pay for output, not seats.” The free tier is up to $100 in trial credits, no card required, and they never expire. One advocate can introduce Viktor through the Slack App Directory.

And Viktor performs in public. It posts its finished work back into channels. It reads channels and suggests automations. When one team debated sign-up growth in a thread, Viktor checked the connected analytics and reported back: true, but “not statistically significant.” Every task it finishes in a channel is a live demo for colleagues who never installed anything.

That is the loop. One person hires it. The whole team watches it work. The watching can turn into usage. There is no per-seat purchase to negotiate for each new colleague.

Results. The numbers moved fast. Dealroom, sourcing Viktor itself, recorded $1 million in ARR at launch. Product Hunt named it Launch of the Day on March 3. On May 19, Fortune reported a $15 million annualized revenue run rate across more than 2,000 organizations, about ten weeks after launch. The homepage now claims 74,000+ teams and shows a 4.9 G2 rating; what counts as a team, and how many pay, is not defined, so treat those as company claims. On May 19, Accel led a $75 million Series A - its partner flew to Warsaw within weeks of launch. The angel list includes Stewart Butterfield and Cal Henderson, the founders of Slack itself.

The honest incidents. A product that acts in front of your whole company also fails in front of it - the team tells these stories on itself, and one of them reached Fortune. During an internal layoff announcement, Viktor reacted to the message with a skull emoji, in a public channel. At one large e-commerce company, an admin connected his personal Gmail as the first team integration, “and then suddenly the team started speaking to Viktor about this guy’s emails.” On day one at another rollout, Viktor “starts DM-ing everyone and then participating in the threads, and the security is going crazy.” Two of these changed how the product ships. The Gmail leak produced per-person integration scoping. The day-one chaos produced their standing advice: roll Viktor out to a few users first.

Why it worked. My read: Viktor spreads one workspace at a time, not one seat. Free credits drop the cost of the first “hire” to zero. Shared integrations and workspace pricing let one advocate equip everyone. And because the output lands in channels, every finished task advertises the product to the next colleague. The distribution is the product’s own output.

The counterweights are real. Microsoft and Salesforce can bundle rival agents into contracts their customers already signed. Usage credits invert the usual freemium curve: the teams that love Viktor most pay the most, and Wiatrowski says some now spend more on it than on a junior hire. Nothing public proves which channel actually drove the growth.

The lesson. If your product does its work in front of a team, the team is the funnel.

Building a product that should spread on its own? Road21 runs email marketing, product-led growth, positioning and outbound for products that need to grow. Book a call and we’ll look at yours.

Sources

Bootstrapped Weekly, October 25, 2026

Gaurav — Sun, 25 Oct 2026 00:00:00 GMT

Bootstrapped Weekly is the news from companies that grew without outside money. This issue covers Monday, October 19 to Sunday, October 25, 2026. Each item says who made the claim, because most of the numbers come from the founders. It’s another short week.

Hex-Rays closes its purchase of HexAI

Last week we reported that Hex-Rays, the Belgian maker of the reverse-engineering tool IDA Pro, was acquiring Israeli startup HexAI for $30 million to $35 million, per Calcalist. This week the deal closed. In a company announcement dated October 22, Hex-Rays said the former HexAI team now works as Hex-Rays employees in one engineering and product organization, and that HexAI’s existing customers keep the same service. The two companies plan to release a joint AI-agent platform for security research in early 2027.

Hex-Rays’ announcement gives no price and says nothing about funding. The price and the “no outside capital” claim are still Calcalist’s, from last week’s report.

ReddGrow says it passed $1M in eight months, with no employees

Yahav Fuchs wrote on LinkedIn on October 20 that ReddGrow AI, which he started alone eight months ago, “just crossed $1M,” without saying whether he means ARR, annual revenue or something else. His next target is $5M. ReddGrow helps companies get mentioned in AI search answers by working through Reddit.

Fuchs previously co-founded Orbb, which raised more than $6M. ReddGrow is a different company. Mako, the Israeli news site, reported on October 16 that Fuchs started it in February 2026 with no employees and no investors, and that it has more than 400 customers and is profitable. Mako describes the revenue as annual recurring revenue of more than $1M. His own post says nothing about investors or profit.

Infinity Global Sports says it crossed $3M ARR with $0 in funding

Joaquin Cordoba, who says he joined Infinity Global Sports after it started, posted on LinkedIn on October 20 that the company crossed $3M in ARR with no funding. He says it came “purely from our outbound muscle”: no paid ads, no investors. Since he joined, he says, the company added more than 25 hires, 4 offices, 5 destinations and a new business line. Infinity sells sports training and travel experiences. Cordoba isn’t described as a founder, and the figure is his own.

Also this week: an interview, not news

The revenue figure below is the founder’s own.

On October 20, the SaaS That App podcast published an episode with Adam Sandman, founder of the software-testing company Inflectra. The episode notes say Inflectra has grown from roughly $1M to more than $20M in self-reported revenue, and describe it as bootstrapped. The episode covers why he skipped venture capital.

Growing a product and not sure which channel to start with? Road21 runs email marketing, product-led growth, positioning and outbound for products that need to grow. Book a call and we’ll look at yours.

Sources

Hex-Rays, Hex-Rays Completes Acquisition of HexAI, Oct 22, 2026
Calcalist, 25-year-old Israeli founder sells stealth AI startup for up to $35 million without raising capital, Oct 18, 2026
Yahav Fuchs, LinkedIn post on ReddGrow, Oct 20, 2026
Mako, Without employees and with more than 400 customers: Yahav built a company with over a million dollars a year (Hebrew), Oct 16, 2026
Joaquin Cordoba, LinkedIn post on Infinity Global Sports, Oct 20, 2026
SaaS That App, $20M, Zero VC in 20 Years: The Bootstrapped AI Playbook, Oct 20, 2026

Cold email for SaaS founders with no sales team: how I run it with AI agents

Gaurav — Wed, 21 Oct 2026 00:00:00 GMT

I run go-to-market alone for a product that helps teams run AI agents. There is no sales team. So cold email has to be small, and most of the work has to go to an agent.

This is how I run it.

Start with who, not what to write

Before any email, I set the target. For this product the first one is tech and agency companies with 15 to 50 people, in Europe without the DACH countries. The Baltics and the Nordics each get their own list.

A list is only useful if I can act on it. My bar: each list ties to one campaign goal. Every company has a named contact and an angle. And there are at least 200 well-defined targets, so the test has enough data to mean something.

I had a hand-picked list of 123 companies. I threw it out. It missed that bar. It was a set of companies that matched a filter. It had no named contacts and no angle for any of them. I couldn’t act on it. On October 13 I changed the targeting to a tighter profile.

The ideal customer was easier to name because we already ran a pilot with an agency like this. That gave me a real company to picture when I wrote the profile.

Principle: set the target first. Without one, an agent will build you a big list of the wrong people, quickly.

What the agent does

I send my assistant short instructions on WhatsApp. A campaign goal and who it is for.

It builds the list. It finds the contacts and verifies their emails. Then it visits each target’s live website and reads what the company does. It writes the sequence from that. It enrolls the contacts and leaves the campaign paused.

The agent works inside my tools and acts on my behalf. It does the prep, the research, the building and the enrolling. I do the approving. It uses APIs where it can, but mostly it works through the browser, using the tools the way a person would.

I spend nothing and send nothing without approving it. I approved each credit spend myself. Sending, or turning a campaign on, stays my decision. The Baltic and Nordic sequences were built this way: contacts enrolled, campaign paused, waiting for me.

Before anything goes live I read examples. I tell the agent what is wrong with the copy and it redoes it. We repeat until I would be fine receiving it.

Principle: the agent does the legwork, you make the calls that matter. Who to write to, and what to say.

The sequence

Every email is written for one company. The agent reads that company’s own site, so the copy refers to what they actually do, and it is tailored to the role of the person receiving it.

Principle: personalize from the company’s own site. A line about their business shows you looked. A generic line does not.

The sequence has four steps. Each email is short, under 300 characters. Each one opens with a value proposition. Each one uses a different use case, taken from the company’s own website. So step two is not a reminder of step one. It says something new.

The subject line is the same every time: “{{first_name}}, quick question?”

No links. A first email from a stranger with a link in it looks like a template.

The email list is verified emails only. I also leave out catch-all domains. A catch-all domain accepts any address, so you can’t tell if the person is real. Skipping them means a smaller list and fewer bounces.

Why this much care? Gmail asks all senders to keep their spam rate below 0.3% in Postmaster Tools, and recommends staying under 0.1% (Google’s sender guidelines). I send small volumes, so I am far from that line. But bad lists are how you end up near it.

The first examples the agent wrote were too long and too generic. I cut them down and asked for plainer wording. The agent redid them, and the second round was much closer to what I would want to receive. I keep doing this on every new campaign. The first draft is never the one I send.

Also, every commercial email to US recipients needs a working opt-out (FTC CAN-SPAM guide). Put it in.

Mailboxes, warm-up and volume

Good copy will not save you if the mail lands in spam. This part is plain setup, and it takes longer than the writing.

Use a separate domain. Don’t send cold email from your main company domain. Snov.io tells users to create separate domains for outreach, to protect the main domain (Snov.io). It also says to buy them at least 30 days before use, because new domains need time to build sender reputation, and to stick to .com, .org or .net.

Use a few mailboxes, not one. Snov.io says to put up to 5 mailboxes on one domain, and warns that more than 5 raises the risk of the whole domain being flagged. It also suggests a mix of Google and Microsoft mailboxes, 2 Google for every 1 Microsoft.

Set up authentication. Google says all senders must set up SPF or DKIM. Senders of more than 5,000 messages a day to Gmail need SPF, DKIM and DMARC, and DMARC can be set to “none” (Google). Microsoft’s Outlook.com has the same 5,000-a-day rule for SPF, DKIM and DMARC, enforced since May 5, 2025 (Outlook.com postmaster). You will send far less than that, but set all three up anyway. I would, and Google recommends it.

Warm up before you send. A new mailbox has no sending history. Snov.io says to wait to send campaign emails until warm-up finishes. Its ramp-up: in week one, 15 campaign emails a day for a Gmail mailbox or 5 for Outlook, plus 40 warm-up emails. Each week after that, add 5 to the campaign limit and take 5 off the warm-up limit. After about a month you reach 40 campaign emails and 10 warm-up emails a day per mailbox (Snov.io).

Keep each mailbox low. The tools give different numbers.

Instantly recommends 30 campaign emails per mailbox per day (Instantly). Its slow ramp option adds 2 emails a day until it reaches your limit.
Snov.io recommends no more than 50 a day per mailbox, campaigns and warm-up together, and says to add mailboxes instead of sending more.
Hunter starts each connected account at 15 emails a day and says to raise it by 5 to 10 a day (Hunter).

These are vendor recommendations. None of them is a rule from Google or Microsoft.

Know the provider limits, then stay far below them. Google Workspace allows 2,000 messages a day per user, 1,500 for mail merge, and 500 on trial accounts (Google Workspace limits). Consumer Gmail stops you sending for 1 to 24 hours after more than 500 emails in a day (Gmail). Exchange Online allows 10,000 recipients a day and 30 messages a minute (Microsoft). Hunter’s own maximum daily limit in its tool is 400 for Gmail and Google Workspace and 250 for Outlook and Microsoft 365. That is Hunter’s cap, not the provider’s. Google and Microsoft set hard limits but give no “safe” cold email volume. For that, use the tool recommendations above.

Do the math on mailboxes. At 30 emails a day per mailbox, 5 mailboxes send 150 a day. A four-step sequence to 200 contacts is 800 emails. Work out how many mailboxes you need before you start, and remember you won’t be at full volume during warm-up.

Keep the list clean. Verified emails only. Leave out catch-all domains. Put an unsubscribe or opt-out line in every email, and stop sending to anyone who opts out. Google requires one-click unsubscribe for marketing messages from bulk senders.

Tools

You need a way to find and verify emails, and a way to send and follow up. Some tools do both. I use a prospecting platform for search and a sequencing tool for sending.

These are other options worth a look. Prices change, so check before you buy.

Finding and verifying emails

Hunter. Free plan: 50 credits a month. Starter: $49 a month, or $34 a month billed yearly, with 2,000 credits. One credit finds one email. Verifying one email costs half a credit. If no email is found you pay nothing. Pricing
Skrapp. Free plan: 50 credits. Professional: $29 a month billed annually, with 2,000 credits a month. You are only charged for emails marked valid or catch-all. Pricing
Prospeo. Free plan: 100 credits a month. Starter has 2,000 credits a month per user. A verified email costs 1 credit. Credits, plans

Skrapp charges for catch-all results. If you leave catch-alls out like I do, check how your tool counts them.

Sending and follow-ups

Instantly. The Growth plan is $47 a month, or $37.6 a month on yearly billing. It includes 5,000 emails a month, 1,000 uploaded contacts, unlimited email accounts and warm-up. Pricing
Snov.io. One tool that finds, verifies, warms up and sends. The Starter plan is $39 a month, or $29.25 a month on annual billing. It includes unlimited follow-ups and warm-up for 3 mailboxes at a time. It charges per new person you contact, and follow-ups to that person in the same billing period are free. Pricing

What I would do this week

Name one target. Write down one campaign goal.
Build a list of at least 200 with a named contact and an angle each. Have an agent do it.
Verify the emails. Drop the catch-alls.
Write four short emails, each with a different use case from the company’s site.
Read the examples. Fix the copy. Then decide whether to send.

The Baltic and Nordic campaigns are built and paused. I decide when they go.

If you want help setting this up, book a call.

Grammarly: how going free built a $13B company

Gaurav — Tue, 20 Oct 2026 00:00:00 GMT

Grammarly nearly died as a paid product. Then it gave the product away for free, and became a $13 billion company with 40 million daily users. This is how the turn happened.

The background. Grammarly started in 2009 in Kyiv. Three Ukrainian founders: Max Lytvyn, Alex Shevchenko, and Dmytro Lider. They had just sold their previous startup, a plagiarism checker called MyDropBox, to Blackboard, and they put almost all of it, under $1 million, into the new company. As non-native English speakers, they knew the problem personally: writing without mistakes takes time, effort, and what Lytvyn calls psychological pressure. They wanted to make it easier.

When they described the idea, people split into two camps. The first said nobody needs this, everyone already has a spell-checker. The second, the people who understood the ambition, said it could not be done. The second group was closer to the truth. The founders wanted real-time feedback on writing, and that was not technically practical in 2009. Lytvyn’s view was that the underlying technology had to progress on its own to make the vision possible, so they built what they could and waited for it to catch up.

Six years of survival. The first version was slow. You uploaded a document, pressed a button, and waited five to ten minutes for the result. The founders took no salaries until 2011 and nearly ran out of money several times.

Before spending everything, they tested demand with two “painted door” tests. First, they put up a consumer website advertising the product, but the checkout showed an error and asked people to come back later. The first purchase attempts came within hours. Second, they went to an education conference in San Francisco and pitched universities directly. Several showed purchase intent. That was enough proof to finish building.

So Grammarly launched as a paid subscription, sold mostly to students, and it had revenue from day one. The founders used it to fund development. “Our users paid for building the rest of Grammarly,” Lytvyn says. By 2011 the company had 300,000 paying subscribers. Brad Hoover, an investor at General Catalyst who had come across the company, joined as CEO that year. Grammarly was cash-flow positive. It was also small: a niche tool for people patient enough to wait ten minutes for a grammar check.

The layup. Two things changed by the middle of the decade. The technology caught up: checking could finally happen in real time, as you typed. And the users had already shown the way. People were using Grammarly for everything, personal emails, professional writing, social posts, not just student essays. The market was not students who could pay. The market was everyone who writes.

The turn. In 2015, Grammarly made two moves at once. It went freemium: the core product became free, with paid plans on top. And it released a free browser extension, first for Chrome, then for Safari, Firefox, and Edge.

The extension was the distribution genius. Instead of asking people to visit a website and paste in text, Grammarly went to where people were already writing: email, LinkedIn, Google Docs, social media. Every time you wrote anything online, Grammarly sat there underlining your mistakes. The product demonstrated itself constantly, for free, inside the tools people used every day.

The results. Growth took off immediately. One million daily users in 2015, the year of the switch. 6.9 million daily users by 2017, most of them free. Thirty million by 2020.

The free tier was not charity. Free users get spelling and basic grammar. Paid plans, around $12 a month, add style, tone, fluency, vocabulary, and plagiarism checks. A heavy user hits the ceiling of the free tier alone, and the upgrade sells itself at that point. Grammarly took its first venture round, $110 million, only in 2017, eight years after founding, and it was already profitable. “Growing faster than anything we normally see in San Francisco,” one of its new investors said at the time. By 2019 the company was valued at over $1 billion. In November 2021 it raised more than $200 million at a $13 billion valuation.

The same bottom-up motion worked inside companies. Individuals brought Grammarly to work, and teams followed. The company launched Grammarly Business, and by 2021 it served 30,000 professional teams, with enterprise customers including Zoom, Cisco, and Expedia. Distribution kept widening: an add-in for Microsoft Office, desktop apps for Windows and Mac, a mobile keyboard for iOS and Android, and a deal with Samsung that put Grammarly’s suggestions inside the phone keyboard itself, no app install needed.

What came beyond. As of May 2025: over 40 million daily users, more than $700 million in annual revenue, still profitable. General Catalyst put in $1 billion of non-dilutive financing, essentially a bet on the customer-acquisition machine the freemium model built. The company acquired the document platform Coda in 2024 and the email app Superhuman in 2025, brought in a new CEO, and in October 2025 renamed the company Superhuman. The writing product is still called Grammarly. Management says it plans to go public eventually.

Why free worked. Grammarly checks three boxes that make freemium powerful. First, it is a daily-use product. You write every day, so the product becomes a habit, and habits spread by visibility. Second, the free version is genuinely useful, not a crippled trial, so people recommend it without being asked. Third, the extension turned every text box on the internet into a demo. The product was the marketing.

There is also a quieter point. Grammarly spent six years as a paid product before going freemium, and those years were not wasted. Selling to students taught the founders who needed the product most, and it kept the company alive on revenue instead of venture money. Freemium worked because there was a working business underneath it.

The lesson for builders. Prove people will pay first, then remove the paywall. Grammarly sold to everyone for six years and nearly died. Then it gave the product away and grew from one million to thirty million daily users in six years. A free tier works when you already know the product is worth paying for.

Sources

Tom Huddleston Jr., CNBC Make It, 2 Ukrainians launched a San Francisco startup worth $13 billion—they nearly went broke ‘multiple times’ on the way (January 28, 2025)
Max Lytvyn, Grammarly, A History of Innovation at Grammarly (April 1, 2021)
Katie Roof, TechCrunch, Grammarly raises $110 million for a better spell check (May 8, 2017)
Aisha Malik, TechCrunch, Grammarly raises $200M at a $13B valuation to make you an even better writer using AI (November 17, 2021)
Jordan Novet, CNBC, Text-checking software maker Grammarly is worth $13 billion in latest funding round (November 18, 2021)
Brad Hoover, Grammarly, Grammarly Raises over $200 Million to Empower Effective Communication (November 17, 2021)
Grammarly, Announcing Grammarly for Samsung Keyboard (October 26, 2021)
Krystal Hu, Reuters, Grammarly secures $1 billion from General Catalyst to build AI productivity platform (May 29, 2025, via Yahoo Finance)
Grammarly, Grammarly Announces $1 Billion Growth Financing With General Catalyst (May 29, 2025)
Aisha Malik, TechCrunch, Grammarly acquires productivity startup Coda, brings on new CEO (December 17, 2024)
Ivan Mehta, TechCrunch, Grammarly acquires AI email client Superhuman (July 1, 2025)
Ivan Mehta, TechCrunch, Grammarly rebrands to ‘Superhuman,’ launches a new AI assistant (October 29, 2025)

Groupon: the email list that built a $12.7B IPO

Gaurav — Mon, 19 Oct 2026 00:00:00 GMT

Groupon built the biggest deal email list in the world, rode it to the largest US internet IPO since Google, and then lost most of its value within sixteen months. It is the strongest case study in email marketing, and the warning that comes with it.

The background. Groupon started as a failure called The Point. Andrew Mason was three months into a public policy graduate program in Chicago when Eric Lefkofsky, a local investor, offered him about a million dollars to build a website instead. The idea was collective action. A user could say: I will do something, but only if a critical mass of other people do it with me. Boycott a cable company together. Fund a cause together.

Nobody knew what to do with it. In the early days the team mostly sat around reading academic books on collective action, Saul Alinsky and the like. After nine months without growth, the investors were close to shutting it down and handing the money back. Mason laid people off to bring the burn down to about $40,000 to $50,000 a month, and used the runway to experiment. One of the experiments was group buying. Instead of an open marketplace, find one deal a day and email it to a list of people who might want it.

The first deal. Groupon launched in November 2008, in Chicago only, and stayed a one-city operation for its first six months. The site was called getyourgroupon.com, because groupon.com belonged to someone in England. The first deal was two-for-one pizza at Motel Bar, the bar downstairs from the office. It sold about twenty. The team earned those twenty the hard way: standing in the building lobby with printed postcards, walking strangers through the signup one at a time. Then a sushi place sold about five hundred, and they knew they had something.

The layup. The daily deal was built for email. A deal lasts one day, so it needs a channel that reaches people that morning and pushes them to act before midnight. Groupon emailed its subscribers discounted offers for local goods and services every day, and it bought subscribers aggressively, mostly through paid search and social ads, at $5.53 each by early 2011. Restaurants and spas made up a slight majority of the offers. The list was not a marketing channel for the business. The list was the business.

Everyone else noticed. Once Groupon hit the tech press, new daily deal clones launched every day. Groupon’s answer was speed: new cities, telesales teams signing merchants, and a bigger list than anyone else could assemble. By late 2010 the list had grown to 35 million subscribers across 300 markets, and the company was pulling in more than $1 billion a year. In December of that year, Google offered a reported $6 billion to buy it, which would have been Google’s largest acquisition ever. Groupon said no.

The results. The IPO filing in June 2011 showed what the list had become:

1.8 million email subscribers at the end of 2009. 50.6 million at the end of 2010. 83.1 million by March 31, 2011.
Revenue of $713 million in 2010, up 2,241% from 2009, and a $2.6 billion annual run rate by the first quarter of 2011.
30.3 million Groupons sold in 2010. By March 2011, 15.8 million people had bought at least one.
In mature markets like Boston and Chicago, about a third of all email subscribers had bought a Groupon.
Groupon kept 40 to 50 percent of each coupon’s value, and 54% of revenue already came from outside North America.

It was among the fastest-growing companies in history. It was also unprofitable on paper, with a $389 million net loss in 2010, though it generated $72 million in free cash flow, because customers paid up front while North American merchants typically waited about 60 days to be paid.

On November 4, 2011, Groupon went public at $20 a share, a $12.7 billion valuation, the biggest US internet IPO since Google in 2004. The stock closed up 31% on day one and hit a record closing high of $26.19 ten days later.

What came beyond. The peak was the end of the story. Under pressure from the SEC, Groupon had already dropped a custom accounting metric that had made the company look profitable in its filing. Through 2012, growth slowed and the clones, plus Google and Amazon, squeezed the model from every side. By early 2013 the stock had lost about 80% of its value. In February 2013 the board fired Mason. His memo to the staff opened: “After four and a half intense and wonderful years as CEO of Groupon, I’ve decided that I’d like to spend more time with my family. Just kidding - I was fired today.”

The company he left was worth about $3 billion. Two years earlier, it had turned down $6 billion. Under new leadership the business stabilized enough for the stock to rise 88% in the first year after Mason, though it never came close to the IPO price again.

Groupon still exists. In 2024 it reported $492.6 million in revenue and 15.4 million active customers, a fraction of its 2011 run rate.

Why email worked. Groupon understood that a deal with a deadline is an email product. The morning email creates urgency that no website visit can. And the economics compounded: a subscriber cost $5.53 to acquire, about a third of them bought, and every buyer kept receiving the email for free. Distribution, conversion, and retention ran through one channel. That concentration is also what made the company fragile. When the novelty wore off and the clones piled in, the inbox that built Groupon belonged to everyone else too.

The lesson for builders. An email list can be the entire engine of a company, not just its newsletter. But a channel builds you faster than it protects you. Groupon proved both halves at full scale.

Sources

Groupon S-1 (June 2011): https://www.sec.gov/Archives/edgar/data/1490281/000104746911005613/a2203913zs-1.htm
Lee Hower, reading of the S-1 (agilevc): https://agilevc.com/blog/2011/06/05/groupons-s-1-from-zero-to-like-billions-in-30-months/
Andrew Mason interview, New York Magazine (2018): https://nymag.com/intelligencer/2018/10/andrew-mason-on-groupon.html
Andrew Mason, Y Combinator Startup School: https://www.ycombinator.com/blog/startup-school-andrew-mason
NYT DealBook, Google offer (Dec 2010): https://archive.nytimes.com/dealbook.nytimes.com/2010/12/03/groupon-said-to-reject-googles-offer/
Reuters, IPO day (Nov 2011): https://www.reuters.com/article/world/groupon-shares-surge-but-concerns-linger-idUSTRE7A404X/
TechCrunch, Mason fired (Feb 2013): https://techcrunch.com/2013/02/28/groupon-stock-goes-up-in-after-hours-trading-on-news-that-ceo-andrew-mason-has-been-fired/
NYT DealBook, Mason’s tenure (Feb 2013): https://archive.nytimes.com/dealbook.nytimes.com/2013/02/28/remembering-the-long-strange-trip-of-groupons-now-fired-chief/
Fortune (Feb 2014): https://fortune.com/2014/02/28/groupon-up-88-in-first-post-mason-year/
BBC News (Feb 2013): https://www.bbc.com/news/business-21625280
Groupon FY2024 results (SEC exhibit): https://www.sec.gov/Archives/edgar/data/1490281/000149028125000006/a2024q48-kxexhibit991.htm

Bootstrapped Weekly, October 18, 2026

Gaurav — Sun, 18 Oct 2026 00:00:00 GMT

Bootstrapped Weekly is the news from companies that grew without outside money. This issue covers Monday, October 12 to Sunday, October 18, 2026. Each item says who made the claim, because most of the numbers come from the founders. It’s a short week.

HexAI sold for up to $35 million with six people and no outside capital

Hex-Rays, the Belgian maker of the reverse-engineering tool IDA Pro, is acquiring Israeli startup HexAI in a deal valued at $30 million to $35 million, according to Calcalist, published October 18. HexAI was founded in 2025 by 25-year-old Elon Gliksberg. It builds an operating system for AI agents that do security research. Calcalist says the six-person company did not raise outside capital, and that all six employees will join Hex-Rays.

A founder with a $1.04 billion exit is bootstrapping his next company

Entrepreneur ran an interview with Wesley Lones on October 13. He helped build Byte, acquired for $1.04 billion, and now runs Daybreak, a telehealth company that treats sleep apnea at home with a custom-fit oral appliance instead of a CPAP machine. Daybreak has treated more than 20,000 patients in roughly its first 28 months, according to Lones. Entrepreneur says it was built without outside funding.

Outreach Today says it passed $3M ARR

ARR Club reported on October 15 that founder Max Korolev said on October 14 that Outreach Today, an email infrastructure platform for cold outreach, crossed $3M in ARR at the start of 2026. Korolev describes the company as fully bootstrapped and fully owned. This is the founder’s own statement, and ARR Club doesn’t say it checked the figure. The milestone dates from early 2026. The announcement is from this week.

MAVI says it reached $1M ARR in eight months, and has since raised

ARR Club reported on October 17 that co-founder and co-CEO Molly Liu announced on October 15 that MAVI reached $1M in ARR in eight months with a team of fewer than five. She and co-founder Aman Puri started with $800 of their own money. The same article says MAVI is now close to 50 people and has raised $4M. So the company bootstrapped to $1M ARR and has outside funding now. It is not currently a bootstrapped company.

Sources

Calcalist, 25-year-old Israeli founder sells stealth AI startup for up to $35 million without raising capital, Oct 18, 2026
Entrepreneur, This Founder Sold His Company for $1.04 Billion and Planned to Take 2 Years Off. His Retirement ‘Lasted 7 Minutes’, Oct 13, 2026
ARR Club, Outreach Today crossed $3M ARR as a fully bootstrapped company, Oct 15, 2026
ARR Club, MAVI bootstrapped to $1M ARR in 8 months with a team under five, Oct 17, 2026

How to find your ICP when you only have a few customers

Gaurav — Thu, 15 Oct 2026 00:00:00 GMT

Search for how to define your ideal customer profile and you get the same advice everywhere: look at your best customers and find the pattern.

That’s fine advice if you have 200 customers. You have three.

Three customers don’t have a pattern. They have a coincidence. So what do you do?

The standard advice breaks at small numbers

Golden Section’s ICP guide says to build it from your best existing customers, scoring each one on things like contract value, churn, support load and sales cycle. It also lists a case where this doesn’t apply: when “fewer than a handful of paying customers exist,” it says to use interviews and a projected scorecard instead. That’s the honest answer. But the guide spends almost all its length on the other case.

Outbound Republic has the better framing for an early company. Its post on building an ICP with almost no customers says early customers are “usually a reflection of your network, not the market.” Then: “at this stage you don’t have an ICP – you have hypotheses.”

I agree with that. But most of what ranks for this question stays vague on the steps. Here are mine.

Step 1: write down why your best customer bought

Not their industry. Not their headcount. The reason.

What was broken before they found you? What happened that week that made them act? Was there a deadline, a bad quarter, a new hire?

That’s the trigger, and it’s the most useful thing you can know. Two companies of the same size with different triggers are different customers. Two companies of different sizes with the same trigger are the same customer.

April Dunford’s book Obviously Awesome starts its positioning process in the same place. Step 1, “Understand the Customers Who Love Your Product,” says:

“The first step in the positioning exercise is to make a short list of your best customers.”

(Quoted from the book as excerpted by Commoncog’s summary and in a reader’s Kindle highlights, Kindle location 711.) She defines best customers as the ones who understood the product quickly, became raving fans, referred you and acted as a reference. With a few customers, that list is short. Good. Start there.

Step 2: ask the people who love it

You have a few customers. Call them. Ask who else they think would get the most out of what you do, and why.

Superhuman’s Rahul Vohra did a version of this at scale. In his First Round Review write-up he asked users how they’d feel if they could no longer use the product. He then looked only at the “very disappointed” group. His reasoning:

“you could just narrow the market with preconceived notions of who you think the product is for, but that won’t teach you anything new.”

(The page has a stray space in “you”; the quote above is otherwise exact.)

The point: let the people who value you tell you who you’re for. Don’t guess from a slide.

Wes Bush at ProductLed gives a blunt way to spot them. In his piece on ideal user profiles he writes:

“You can spot your ideal users by looking at which customers use the product the most, upgrade the quickest, and leave the most testimonials.”

Note he says users, not buyers. His point is that the person who uses the product daily is not always the one who signs the check. With three customers, that’s a good prompt: who actually uses what you built, and who only approved it?

You won’t run a survey with three customers. A 20-minute call with each does the same job. Ask what they’d do without you. Listen to how they describe the problem. Their words will beat anything you write.

Some colleagues of mine were selling an agentic solution to manufacturers: user agents that could potentially do almost anything, or at least be set up to. They led with blueprint error detection, request for quotation, and order intake. But when they talked to the manufacturers, they found the companies had very different pains. They had thousands of SKUs, and their prices were tracked in an Excel sheet. That sheet needed pricing updates more than five to six times a year. So they built that for them.

Step 3: pick two or three segments, not one

Now you have a trigger and a few guesses. Turn them into two or three segments. Specific ones. “B2B SaaS companies moving from sales-led to self-serve” is a segment. “Startups” is not.

On how narrow to go, Dunford is blunt in a post on her blog (not the book) about selecting a target market she writes:

“In my experiences the smaller and tighter you can get on your segmentation, the easier it is to get early traction. You can always go broader later.”

She also says “SMBs” is “just too big to be a practical target market for a startup.” Same goes for “startups” or “SaaS.”

Outbound Republic suggests lists of 50 to 100 contacts per segment, run at the same time, so you can compare. I think that’s about right. Small enough to be quick. Big enough to hear back from.

Step 4: let the market answer

Send each segment a short, specific message about the problem. Then read the replies, not the open rates.

Outbound Republic’s test is a good one: when you’ve hit the right segment, prospects “reference the problem without you having to explain it.” When you haven’t, conversations are polite but slow. It suggests running each segment for two to three weeks.

This is the step most founders skip. They spend months debating the ICP in a doc. Three small campaigns would have told them in a month.

A caveat that I’d add and the sources don’t: small tests are noisy. If one segment replies twice and another replies once, that proves nothing. Look for a difference you can’t explain away. Prospects who answer fast and ask real questions are signal. A slightly higher reply count is not.

Step 5: write the one-page version, then revise it

When one segment pulls ahead, write it down on one page:

Who they are (industry, size)
The trigger
The role you talk to
Who looks like a fit but isn’t

The last one is the one people skip. Saying no to a prospect who looks fine on paper is what makes the ICP real. When your pipeline is thin, it’s also the hardest thing to do.

Treat the page as a guess you keep improving. After ten customers, redo it with real numbers.

The short version

With three customers, don’t analyze. Talk to them. Find the trigger. Test two or three segments with small outbound runs. Keep the one that answers fast. Write one page. Revise it.

If you want help running those tests, with the lists, the inboxes and the sequences, book a growth call.

Sources

Golden Section, Defining an ideal customer profile
Outbound Republic, How to build an ICP when you have almost no customers yet (June 30, 2026)
Wes Bush, ProductLed, How to identify your Ideal User Profile (July 27, 2023)
April Dunford, Obviously Awesome (Step 1), as excerpted at Commoncog
April Dunford, blog post, How To Select A Target Market (Feb 16, 2017)
Rahul Vohra, First Round Review, How Superhuman built an engine to find product-market fit