The part nobody warns you about
I'm a developer. I like the part where you build things. I am extremely good at convincing myself that the next feature is what stands between me and users.
It wasn't. I shipped a thing, posted it on X, got a couple hundred visitors from people who were also shipping things, and that was the entire funnel. The product worked fine. Nobody knew it existed.
The uncomfortable realisation: writing the code was the part I was qualified for, and it was also the part that had stopped mattering. Anyone can ship now. A working product is no longer evidence of anything. Distribution is the constraint, and it's the one thing my twelve years of getting better at programming did nothing to prepare me for.
So I did what developers do when faced with a non-technical problem — I went looking for the part of it that was technical.
Where the users actually are
The people who'd use my product were already watching someone on YouTube. Not on Product Hunt, not on Hacker News — watching a twelve-minute video from someone whose recommendation they actually trust.
That trust is the asset, and it doesn't transfer to your landing page no matter how good your copy is. Getting mentioned in one of those videos beats a banner ad by an embarrassing margin. Nielsen Catalina's work with TapInfluence put influencer content at roughly 11x the ROI of display advertising; Influencer Marketing Hub's benchmark puts average return around $5.78 per $1 spent. Display click-through, for comparison, lives somewhere around 0.05–0.5%.
Mid-size creators are also reachable in a way ad platforms aren't. There's no auction. A lot of them will do a genuine mention for an affiliate cut or a free account, because most of them have never been asked. You're not bidding against a company with a media budget — you're sending an email to a person who is also, usually, a founder.
The catch is volume. At a realistic 5–10% reply rate, you need hundreds of contacts before this is a strategy rather than a hobby. And getting one creator's email by hand takes about fifteen minutes: check the About tab, hit the captcha, give up, follow the Linktree, follow that to a website, find the contact page, scroll to the footer.
Five contacts an hour is not a campaign. It's a rounding error.
The obvious move, and why it doesn't work
Buy a tool.
Then you look at the pricing. Modash starts around $299/month. NinjaOutreach runs about $389/month on an annual commitment. Upfluence, Klear, and HypeAuditor won't show you a number at all — you book a demo, sit through a call, and get quoted something with a comma in it.
These are good products aimed at real customers: brand teams with a media budget and a headcount line for influencer marketing. They ship CRMs, campaign tracking, audience demographics, fraud detection, and twenty other things.
I didn't need twenty other things. I needed email addresses. Paying $299/month to get a few hundred contacts is not a growth strategy at $0 MRR — it's how you go from bootstrapped to broke.
So, the thing developers do instead of paying: I built it.
What I expected to be hard, and what actually was
I assumed the YouTube Data API would do most of the work. It doesn't return email addresses — there's no endpoint for the thing I needed, and the quota model rate-limits you into irrelevance before you've covered a single niche. That was the first afternoon, gone.
I budgeted another afternoon for pulling an address off a page. That turned into the most-rewritten part of the codebase, and it's where I'd point anyone who thinks scraping is a solved problem.
Creators actively defend against this. They write addresses as name [at] domain dot com, in about six variants. They encode them so the rendered page reads fine and the source doesn't. Many don't publish one on YouTube at all and expect you to find their site. Meanwhile the open web is full of things that look like email addresses and aren't — markup fragments, asset filenames, strings lifted out of inline scripts — and every one of those that survives into your output is a bounce.
The lesson that actually transferred to the rest of my work: I spent a week chasing bad matches one shape at a time before admitting the approach was wrong. Enumerating what's invalid means chasing new shapes forever. Defining what's valid and rejecting everything else killed the entire class of bug in a single commit. I now reach for that far earlier than I used to.
None of this is glamorous. All of it is the difference between a list you can send to and a list that bounces 30% and torches your sender reputation — which then quietly costs you the good addresses too, because your domain's reputation isn't scoped per-recipient.
The bit where someone asks if this is okay
Fair question, and I'd rather answer it than have it be the top comment.
These are business inquiry addresses that creators publish, on purpose, so brands can contact them. It's the email equivalent of a "for booking, contact —" line. The extension reads what's publicly rendered, from the user's own browser, at the moment they search. There's no central database of scraped people sitting on a server of mine.
That said: having someone's address is not permission to spam them. GDPR and CAN-SPAM still apply. Blasting 500 identical "I'd love to collaborate!" emails is both ineffective and rude. And if you build outreach on top of this and it doesn't work, the list was probably not the problem.
The thing itself
It's called TubeMailer. A Chrome extension — search a niche, get creators with their public business emails, export to CSV, Excel, or JSON, drop it into whatever you send with. It reads addresses live at search time rather than serving them from a stale database, which is most of why the bounce rate is survivable.
Free tier is 5 emails a day, which is enough to judge whether the lists are any good before you pay anyone. Paid is $49/mo for 1,000/day and $99/mo for 3,000.
What I'd tell past me
The distribution problem does not have a code solution. But it usually contains one component that does, and finding that component is how a developer gets purchase on a problem that otherwise feels like it belongs to someone else. I couldn't will myself into being a marketer. I could remove the fifteen-minute step that was making the marketing impossible.
Start there. The rest is just writing emails, and you can learn that.