Zippost

Recap for new readers: AI content tool for realtors and other 1099 professionals, built by a working realtor. Single digits everywhere: users, traffic, revenue. The gap since my last entry wasn't quiet, it was heads-down: App Store prep, a rebuilt onboarding, and the two lunches this post is about.

The operating pattern this month, discovered by accident: have lunch with a founder, ship their feedback before dinner.

Lunch one: a founder friend typed his email into my waitlist and asked "is this even live?" Fix shipped same day. The homepage is now an anonymous demo: paste your website, see your brand mirrored back in 20 seconds, colors and voice and all, before we ask for anything. Email only comes after "yes, that's me."

Lunch two: another founder said the site never names who it's for. Also true. Shipped same day: the homepage now walks three personas through their week. The agent story is literally my own Tuesday.

Today was infrastructure day for the boring-but-necessary pile: Zippost got a LinkedIn company page, my profile finally says Founder, and I re-cut the product explainer video so it tells the same story as the landing page (three personas, try-first flow, 31 seconds). Everything a stranger touches now says the same thing, which took embarrassingly long to achieve for a company whose product is literally about consistent messaging.

The deflating data, because this series doesn't skip it: since the anonymous demo went live, one person has run a scan. It was me, testing. The funnel is beautiful and empty.

The deflating data, because this series doesn't skip it: since the anonymous demo went live, one person has run a scan. It was me, testing. The funnel is beautiful and empty.

What I believe about the lunch pattern after two rounds: shipping advisor feedback same-day works when the feedback is about confusion, because confusion is universal. Neither founder is my ICP and neither told me what to build. They told me where they got lost. That's become my filter: confusion reports ship immediately, feature ideas wait for users.

But velocity on feedback is not distribution. The conversion machinery keeps improving and nobody arrives to be converted. The lunches fix the funnel. They don't fill it.

So the plan stays boring: warm outreach, two agent conversations in flight, posting through my own product daily. Five real weekly users, then I build only for them.

Question for the room: when someone smart but outside your ICP gives you product feedback, what's your filter for ship-now versus park-it? Mine is "confusion ships, ideas wait." Curious what survived contact with reality for you. that, tell me the number on your last posted entry and I'll anchor to yours instead.

Recap: solo founder, AI content tool for people who don't have time to post, five signups, growth funnel made of lunches, last experiment converted at exactly zero.

Here's the number that keeps me sane this week: 64 post kits have ever been generated in my product. 40 of them are for my own real-estate brand. My biggest user is me, and honestly, it's not close.

That's not a joke stat, it's the strategy while user count is stuck. My realtor brand, davidrayrealestate.com, now runs on Zippost output: the market analysis that did well on LinkedIn last week, the posts, the carousels. Every kit I generate for my real business surfaces something to fix in my software business. This week that loop produced a voice-samples feature (drop in your best old posts and drafts mirror them per platform) because I, the user, complained to me, the founder, that the LinkedIn drafts didn't sound enough like my LinkedIn.

Meanwhile there's a whole category of founder work nobody posts about: paperwork for storefronts that don't exist yet. I'm building the business pages, and I'm preparing the iOS app for the App Store, which is its own bureaucratic hiking trail: developer program enrollment, device testing, screenshots in six sizes, privacy labels that had better match what the privacy policy says, and Apple's review guidelines for anything with AI-generated content. The engineering has been done for over a week. The hurdles are all admin, and they're real days of work.

So the honest split right now: GTM stays the focus, because five real users is still the bar and nothing else unlocks until then. But there's a parallel track of unglamorous prep that has to happen anyway, and doing it during the quiet weeks means it won't block anything during the loud ones.

Some days both jobs pull at once. Tuesday I toured properties for clients in the morning and wrote an offer and then App Store metadata at night. I used to think building in public meant sharing wins. Turns out it mostly means documenting the weeks where you're the only user of the thing you believe in, still showing up to both jobs.

Question for the room: for those who dogfooded through the quiet stretch, did being your own power user actually produce a better product, or did it bias you toward building for yourself? I genuinely can't tell yet which one I'm doing. The realtor roadmap says "for myself" is the point. The zero conversions say maybe I'm the only one who wants it.

Recap: solo founder, AI content tool, spent all week going deeper on real-estate agents (the roadmap literally says "realtor engine"). Five signups, two active, growth strategy made of lunches.

Yesterday I ran a deliberate experiment against my own strategy. My LinkedIn posts had all been agent-to-agent. This one opened "I'm a realtor, here's how I live" and then swung the aperture wide: attorneys, accountants, coaches, anyone with a 1099 whose content is their storefront. The hypothesis: my network knows me as a realtor, but most of my network isn't realtors. Maybe the niche was costing me my own warmest audience.

The result, 30 hours later: warm engagement on the post itself. And zero tracked events on the site. Not low. Zero.

What I take from that, in order of confidence:

1. Feed warmth doesn't convert without a direct ask. This is now the third time I've measured it. People will comment on a post before they'll click a link in one. The comment costs three seconds and pays social credit. The click starts a commitment.

2. Broadening the message didn't change the physics. Niche or broad, a LinkedIn post is a billboard, and billboards don't work at n=5. Conversations do. The aperture question turns out to be the wrong question at this size.

3. Honest caveat: professionals run ad blockers, and my analytics almost certainly miss some real visits. I'm shipping a fix for that blind spot this week (first-party proxy) because flying without instruments is how I'd misread everything above.

So the experiment answered a different question than I asked. I wanted to know "niche or broad." The data said "neither, go back to the lunches." Which is fine, because the lunches are working: the calendar-based funnel from day 7 is still producing my only real conversations.

The plan stays boring: DM every person who commented on either post with a personal note and the link. Follow up on the quiet emails. Keep the meal funnel full. Five real weekly users, then I build only for them.

One thing I'd ask the room: has anyone actually converted feed engagement to product usage at tiny scale without taking it to DMs? I'm starting to believe the answer is structurally no, and I'd love a counterexample.

Recap for new readers: solo founder, AI content tool for real-estate agents, 5 signups, 2 active, every cold channel has bounced me so far.

Here's what "growth" actually looks like this week. Tomorrow: lunch with my neighbor, a top agent who's sold more units in our shared building than almost anyone. This week: a call with a friend from my old brokerage who answered my launch email with "congrats, let's chat." Those two calendar events are the entire pipeline.

The mechanics of how I got even this far: I mined my own sent mail and my LinkedIn data export, found the 168 agents I actually know, ranked them by real message history, and wrote every email around something true between us. The open house where we met. The deal memos I keep requesting. The building we share. Fourteen sent. One reply. From my closest contact.

What manual outreach is teaching me:

The funnel stages at my size are named after meals. Email, reply, coffee, lunch, user. Every conversion is one human deciding you're worth an hour.

Personalization gets you opened, not answered. The emails were good. But an agent's inbox is leads and escrow deadlines, and a colleague's product launch is nobody's urgent.

The lunch is secretly a usability test. The last time I watched someone use the product across a table (day 2 of this series), I walked away with six bug reports and a re-prioritized roadmap. Tomorrow I'll watch Sergio point the Mirror at his real brand, and I'll say nothing while he squints.

I kept waiting for this to feel beneath the ambition of "startup." It isn't. This is the whole game at this size, and pretending otherwise is exactly how I wasted week one on filters and feeds.

The bar from day 6 stands: five real weekly users, then I build only for them.

For those past this stage: what was the most embarrassingly manual thing that actually worked, and how long before you could stop doing it?

Series so far: deleted my approval flow, built an outreach machine, shipped LinkedIn publishing into a void, ran my product's output on my professional account.

This morning's honest numbers: 5 signups total, 2 activated, 8 pageviews in the last three days and every one of them is me. I've sent 14 personally-written emails to agents I actually know or researched deeply. One reply so far. It was warm ("congrats, let's chat"), and it came from the person I know best. The other 13 sit unread or unanswered.

Some hypotheses about why, because guessing in public is the point of this series:

1. Agents don't live in email the way founders do. Their inbox is leads, escrow, and listing alerts. A thoughtful note from a colleague still waits days. My one reply took 12 hours and came from a friend who'd have replied to anything.

2. Content pain is chronic, not acute. Nobody's feed fell over today. There's no forcing function that makes "try it now" beat "sounds cool, someday."

3. Professionals adopt what colleagues vouch for. I have zero testimonials, so every pitch asks someone to go first. Going first is expensive.

4. Borrowed audiences keep rejecting me (documented in earlier posts: filters, link suppression). Earned audiences respond. I only have one of those.

So the next play leans all the way into that last line. Tomorrow is my birthday, and I'm posting one ask to my whole network: no wishes, just try the thing and tell me what's broken. It's the one day a year you're allowed to ask a favor at scale, and I intend to spend it completely.

Beyond that, the plan stays stubborn and small: founder-led everything, personal follow-ups on the 13 quiet emails after a respectful gap, favors called in one at a time. The goal isn't traffic anymore. It's five real users. Not signups: people who generate posts weekly and complain to me directly.

The moment I have those five, I stop marketing and start building for exactly them.

The roadmap's already pointed there: listing posts, market updates, property pages. Things an agent needs weekly, built with the five people who'll tell me when it's wrong.

Question for the room: when you were at zero-ish, did you set a "first N real users" bar before changing behavior? What was your N, and what did you do differently once you hit it?

Context from the series: 5 signups, warm outreach working, cold channels dead, 2 pageviews on a bad day.

New experiment. I'm a working realtor with a personal brand account that has more reach than my product will for months. So I had Zippost write a genuine market analysis (why SF holds while Phoenix and Austin correct), generated the carousel with it, disclosed "includes AI-generated content" in fine print, and posted it as regular professional content. No product mention.

Results: 15 reactions, 8 comments, 2 DMs from people wanting to talk SF strategy. Real engagement from my actual professional audience, including other agents. One comment called it "the clearest take on the correction I've read." The disclosure line? Not one person asked about it.

Then I posted the reveal: that was my product doing its job. [early reveal-post numbers if available]

What I think I learned: "AI content" isn't a quality tier, it's a workflow. The same audience that scrolls past obvious slop engaged with this because the take was specific, local, and in my voice. The tool did the research and drafting; I did ten minutes of taste.

Also: founder-led distribution might just mean using the reach you already have. I spent three days trying to borrow audiences (Reddit filters, LinkedIn suppression). The audience that actually responded was the one I built over years in my day job.

For the folks selling tools to a profession you belong to: did your personal professional credibility end up being the main channel? How far did it carry you before you needed something else?

Series so far: day 1, three visitors and one signup (me). Day 2, deleted my approval flow. Day 3, built an outreach machine and admitted traffic was zero.

Yesterday's traffic: two pageviews. I've stopped flinching at the number.

What actually happened this week is that I went fully founder-led on both ends of the business.

Outreach: 26 personalized emails to real-estate agents, written one at a time. Not a template blast. I mined my own inbox and my LinkedIn archive, researched every person's actual content, and wrote each email around the video series they run or the open house where we met. Warm relationships remain the only channel with a pulse, so I'm pressing harder on exactly that: my neighbor who sells units in my building, the broker whose deal memos I keep requesting, the founder who once tried to recruit me. Every email has my name, my license number, and my phone number at the bottom.

The dogfooding got recursive this week, which is my favorite development. Zippost now publishes to LinkedIn directly: connect your account, compose, schedule. The first thing it published was my LinkedIn post about Zippost. The product markets itself through itself, and my entire LinkedIn presence this week is its output. I review, I tweak, I hit schedule.

On product, I've stopped pretending to be horizontal. I sell real estate for a living, my first users are agents, and the roadmap now says so out loud. Shipped this week: LinkedIn publishing and scheduled posts, an iOS flow that stages your kit and deep-links into each platform's app, a writing-quality split where a cheap model filters topics and a strong model writes every word a human actually reads, and user-manageable radar sources. Next up, built for my agent early users: listing posts, weekly market updates, property pages. The tool my own work week needs, basically.

The thesis I'm testing: at zero distribution, the only things that compound are product velocity and relationships. So the plan stays the same, just more ruthless. Ship for realtors daily. Put my name on every email. Keep publishing through my own product, in public.

For the folks who niched down hard, one profession instead of "creators" or "small businesses": how long before the niche started pulling you instead of you pushing it?

Continuing the series (day 1: three visitors, one signup, it was me; day 2: deleted my approval flow). The scoreboard as of this morning: 5 real signups, 2 fully activated accounts, 48 post kits generated. Ten pageviews in the last 48 hours, and most of those are me.

Here's what we actually did since the last post. Shipped the auto-approve change from day 2, so signing up now lands you straight in the product with a welcome email in under a minute. Fixed every major bug from my friend's feedback session within a day and told him so, which felt better than any metric this week. And I built a real outreach machine: mined my own inbox and my LinkedIn export for the 168 agents in my network, deep-researched 46 agent-creators on YouTube and Instagram, and wrote 26 personalized emails where every single one references the person's actual content. A video series they run, a market report they publish, the open house where we met. About half have gone out so far.

What's working: warm beats everything, and it isn't close. Every signup so far traces to a real relationship. Also working: dogfooding. Most of those 48 kits are me using Zippost to write Zippost's own marketing, including the posts and emails I just described. The product is its own content department, which is either poetic or concerning depending on the day.

What's not working: anything cold. LinkedIn gives me comments without clicks. Reddit's filter ate my post before anyone saw it. Public posting has produced, as far as I can measure, zero visitors.

So the honest status is: product improving daily, feedback loop tight, founder still personally hand-delivering every user like it's 1995. I know the answer is probably "keep going, the compounding is invisible at n=5." But I'm asking anyway: what was the first channel that reliably brought you strangers instead of friends, and how long did it take to show up?

Follow-up to the day 1 post (3 visitors, 1 signup, it was me).

Current count: four signups. My cousin is one of them. How many are active users? I honestly can't tell yet, and that's kind of its own answer.

Watching the funnel taught me something uncomfortable. My flow was: sign up for a waitlist, wait for me to approve you, come back, sign in. Three steps, and two of them existed for my comfort, not the user's. At my scale an invite gate isn't scarcity. It's friction wearing a scarcity costume. Someone lands on the page excited, joins, and by the time my approval email arrives they've moved on with their life.

So today I removed it. Signing up now goes straight to a welcome and the product. I still see every signup in my admin panel, the approval just happens instantly. If there's ever a real reason for an invite gate, it's one env var to turn back on.

Also shipped since yesterday: all the big bugs from my friend's feedback session are fixed and live (the mobile sideways-scroll, the cropped X images). Took about a day.

For people who've run early-stage funnels: did a waitlist ever measurably help you, or is it theater below a few hundred signups? I genuinely can't find evidence for the first case.

Shipped my AI content studio (Zippost — paste your website, it learns your voice, watches your industry, writes your week of posts) into early access yesterday. Posted on LinkedIn and Reddit. Got comments, reposts, encouragement.

Then I checked PostHog: 3 unique visitors. Reddit's spam filter killed my post before anyone saw it (new account). LinkedIn gave me engagement but zero clicks — turns out comments are free, clicks are not, and LinkedIn buries posts with external links anyway.

What I'm changing this week: warm DMs to my actual ICP (I'm a realtor; my first users are realtors), native video with a comment-for-invite mechanic instead of links, and building Reddit karma before reposting.

Two questions for people who've been through this: what actually moved your first 100 visitors — and did a waitlist help you or just add friction? I'm one tap away from making access instant and I can't decide.