Honest monthly recap from a solo developer marketing one privacy-first Android app entirely in public. Real numbers, two rules that survived the month, one that just died, and a competitor that made me rewrite my own pitch.
A month ago, my first build-in-public post reported 35 articles and 185 total dev.to views. As I write this, the catalog is 71 articles and 820 views — plus the first 6 reactions and first 12 comments this account has ever earned. Roughly double the catalog, four and a half times the traffic, and the first signs that anyone on the other side is actually a person and not a search crawler.
That's the headline. The interesting part is everything the numbers don't say.
What actually moved the needle
The single biggest lever this month wasn't writing more — it was changing where the writing landed first.
For the first eight weeks, dev.to was my canonical surface: publish the article, then spin off a Twitter thread, a Quora answer, a LinkedIn post. Pillar first, derivatives after. The problem was that new dev.to publishes hit a Day-1 traffic wall — a feed-pickup throttle that quietly shuts off discovery for new submissions on a low-follower account, while older catalog pieces keep trickling views from search.
So I inverted it. For privacy-anxiety topics, the rule became Quora first, X long-post at +72h, dev.to canonical at +7d. The pillar now lives where its audience is already searching — a Quora question like "How do I know if my free security camera app is collecting data about my home?" is itself the feed. The dev.to article became the durable destination people follow after engaging, arriving as referral traffic rather than feed traffic. Referral traffic isn't throttled the way feed pickup is.
The catalog lift since the pivot started (253 views on 5/19) to today (820) is real, and the shape of it is the tell: it concentrated on the cross-linked middle of the catalog — the 5-to-10-view pieces external referrers land on — not on each day's fresh publish. That's exactly the fingerprint you'd predict if the mechanism is "land readers on existing canonicals," not "finally beat the feed."
This rule is now operationally bound. It's not an experiment anymore; it's the default routing for every privacy-anxiety pillar.
The rule that just died
Here's the part build-in-public is for: reporting the experiment that failed.
I had a sub-hypothesis that the build-in-public archetype specifically should reverse even further — publish to LinkedIn first, then X at +48h, then dev.to canonical at +7d — on the theory that a week of LinkedIn referral runway would let the dev.to canonical finally break its Day-1 zero.
The test piece was last week's Week-4 verdict, published to dev.to on 6/06. The bet: a genuine Day-1 view count (≥1, with real read-time) would confirm the LinkedIn runway worked.
Seven days later, that article sits at 10 views with zero reactions and no measurable read-time — the same flat, backfill-shaped floor I see on throttled publishes that got no referral runway at all. The LinkedIn-first ordering produced no detectable lift over just publishing normally.
Verdict: falsified. Build-in-public reverts to dev.to-first, LinkedIn as a +48h cross-post. This very article is the demonstration — it's going to dev.to first.
The lesson I'm taking: the channel-pivot rule works because it moves the pillar to a higher-intent surface (Quora's question graph). LinkedIn-first failed because LinkedIn, on a tiny account, isn't a higher-intent surface for this content — it's just a different low-traffic one. Reversing the order only helps when the new front-of-line surface is genuinely better at finding the right reader. Order is a proxy; intent is the thing.
The day a competitor shipped my moat
The recap I didn't want to write: FadCam is on the Google Play Store.
FadCam is an open-source, ad-free Android recorder with background recording, screen-off capture, live streaming, remote camera control, and no data collection. If you read that list and thought "that sounds like your app" — yes. It overlaps a large part of what I've spent a month positioning Background Camera RemoteStream around. And it's open-source, which means on the single axis I lean on hardest — trust me, the data stays local — FadCam can do something I can't: let you read the source and verify it yourself.
A weaker version of me would bury this. The honest version: a competitor shipping your differentiator is the best forcing function you'll get for finding out what your actual moat is.
After a week of sitting with it, here's where I landed. "Privacy-first background recording" is no longer a moat — it's now table stakes in this niche, and an open-source rival arguably owns the purist end of it. What's still mine is the streaming-and-remote-viewing stack built for non-technical users: one-tap streaming to YouTube Live (not just a local network), a built-in web server so you can watch the feed from any browser with zero setup, and a no-signup, no-account flow that someone repurposing an old phone can finish in five minutes. The moat isn't "we're private." The moat is "we're the easiest way to turn a spare Android into a live, remotely-viewable camera without trusting a cloud."
That's a narrower claim. It's also a truer one, and FadCam is the reason I now know which sentence to lead with.
What's next (Month 2)
Three things.
First, lean into the narrowed pitch. Every comparison piece and landing surface gets rewritten around YouTube Live + browser-based remote viewing + zero-setup, not generic "privacy." Where FadCam is the right call for a reader (pure local recording, wants to audit the source), I'll say so — credibility compounds faster than concealment.
Second, find out whether comments are a channel. Twelve comments in a month, all clustered on one thread, is a small number with a big asymmetry: it's the first two-way traffic this account has had. Month 2 tests whether author-level dialogue on high-traffic third-party posts converts to durable referrals.
Third, keep the channel-pivot rule bound and stop re-litigating it. It works. The discipline now is execution cadence, not more ordering experiments — that lesson cost me a week and a falsified hypothesis, and the data was unambiguous enough that I don't need to run it twice.
If you're building something solo and marketing it in the open, the meta-lesson from Month 1 is dull and useful: most of your leverage is in where and to whom, not how much. I wrote roughly the same volume all month. The month moved when I changed the addressing, not the output.
Month 2 starts now.
Background Camera RemoteStream turns an Android phone into a privacy-first camera that records with the screen off, streams to YouTube Live, and serves a live feed from a built-in web server — local storage, no account, no cloud. Free on Google Play. More at superfunicular.com.
Related: the Week-4 channel-pivot verdict · turning an old Android phone into a free security camera · why a privacy investigation into Meta's AI glasses raises a question for every camera app.
Top comments (0)