Ask a developer how they plan to market their product and you usually get a list of channels: post on X, maybe do a Product Hunt launch, try some ads. Ask a marketer the same question and you get a question back — who is this for, and where do they already go when they have this problem?
That difference is the whole thing. What follows is each major channel, what it actually rewards, and the specific mistake that comes from approaching it with an engineer's instincts.
Social platforms
The first error is treating “social media” as one channel. It is five or six unrelated channels that happen to share a phone icon, and they reward completely different behaviour.
X rewards a strong opinion stated compactly, and it rewards frequency. LinkedIn rewards a narrative arc with a lesson at the end — widely mocked, consistently effective for B2B. Reddit rewards being a real participant and punishes promotion faster than anywhere else. TikTok and Instagram reward visible transformation: something that looks different at the end than the start. YouTube rewards demonstrating a thing properly, and pays out for years rather than hours.
The developer's mistake here is usually one of two. Either they pick the platform they personally enjoy rather than the one their buyers use — which is how you end up with a B2B compliance tool being posted about on TikTok — or they post the same announcement to all of them simultaneously, in a format native to none.
The second mistake is more subtle: posting once. A developer announces the launch, gets four likes, and concludes social does not work. A marketer knows that a given post reaches a small fraction of even your own followers, that of those maybe a tenth are paying attention, and that of those a fraction are in a position to care today. So they say the same thing fifteen different ways over three months. To the engineer that feels like spamming. To the marketer, saying it once is the equivalent of shipping a feature and never mentioning it in the changelog.
Building in public
The one developers reach for most, because it is the only channel where the subject matter is something they are already comfortable talking about: the work.
It does work — but it is worth being precise about what it works for. Building in public builds an audience of people interested in building. That is a genuinely useful asset: feedback, encouragement, early testers, and a real defence against the isolation of working alone.
What it very rarely builds is a customer base. The people following your MRR updates are overwhelmingly other founders doing the same thing. They will cheer your launch, retweet it, and not buy it, because they are not the person with the problem — they are a colleague. Founders regularly reach several thousand engaged followers and are baffled that it converts to almost nothing. It converted exactly as it should. The audience was makers.
Use it deliberately: build in public for the support and the feedback loop, and run a separate, unglamorous channel aimed at the people who actually have the problem.
Podcasts
Structurally excellent and almost universally underused. An hour of undivided attention from a self-selected audience, permanently indexed, and small shows are often actively looking for guests.
The mistake is pitching the product. Nobody has ever booked an episode because a piece of software exists. Hosts need an interesting hour, and their unit of interest is a story or an argument, not a feature set.
So pitch a topic you can carry: the technical problem that turned out to be harder than expected, the thing you were confident about and got wrong, a counter-intuitive pattern in your data, the reason your whole category does something backwards. The product comes up naturally in the way hosts always ask, and the mention is worth more for having been earned.
Go small. A show with two thousand committed listeners in your exact niche will outperform a general business podcast with fifty thousand, and it will say yes.
Influencers and creators
The version developers imagine is paying someone with a large following to mention the product. That is the expensive version and usually the worst-performing one.
The version that works is finding the people whose existing content already ranks for the problem you solve — the blogger whose “best tools for X” guide sits at the top of Google, the YouTuber with the definitive tutorial, the newsletter writer whose recommendations get acted on — and giving them something genuinely useful. Free access, your data, a tool built for their audience, a good answer to a question they get asked constantly.
Developers tend to find this distasteful, reading it as buying an endorsement. The reframe that helps: you are not buying a mention, you are giving someone whose job is to be useful to their audience something that makes them more useful. If your product does not do that, the problem is upstream of the channel.
Paid ads
The one where developer instincts do the most financial damage, because ads look like an engineering problem. There is an input, an output, a dashboard, knobs to turn. It feels tractable in a way that posting on LinkedIn does not.
The trap is that ads amplify. They multiply whatever your page already does. If a hundred organic visitors produce zero signups, a thousand paid visitors will produce zero signups and an invoice. Founders reach for ads precisely when nothing is working, which is the exact moment they are guaranteed not to help.
A marketer's sequence is the reverse. Prove that traffic converts. Establish what a customer is worth. Only then buy more of the thing you already know works. The narrow exception — a small fixed sum spent purely to get enough traffic to test whether the page converts at all — is a research cost, and should be budgeted and stopped like one.
The actual difference in thinking
Underneath all of this sits one distinction, and it explains most of the individual mistakes.
The developer's model of promotion is announcement. The product is finished; the remaining task is to inform people it exists. Under that model, saying it once is sufficient, saying it repeatedly is spam, and choosing a channel is a question of reach.
The marketer's model is demand. Almost nobody is looking for your product, because they do not know it is a thing that could exist. They are looking for relief from a problem, in a context, at a moment. The job is to be present in that context when the moment arrives — repeatedly, in their language, wherever they already are.
That is why the marketer asks “who is this for and where do they go” before naming a single channel, and why the developer's channel list, produced before that question is answered, is usually a list of places their users are not.
None of this is hard to learn. It is just a different thing to be good at, and being excellent at building actively obscures it.