Marketing for engineers
For the person who can build anything and has never sold anything. Not about selling to developers, which is a different job with the same name.
Marketing is the same skill you already have: find out what is true, then tell people clearly. The first week has nothing to do with channels. Talk to twenty people who have the problem and write down the exact words they use, put those words on one page with a real price, and put $100 to $200 behind ads on two or three platforms to see whether anyone clicks. Until someone pays you, marketing is customer research wearing a different hat.
Why does the advice I found not fit?
Because “developer marketing” already means something else. Search it and you get advice on selling software to developers: docs, API tutorials, free tiers, sponsoring newsletters. That is a real discipline and it is not your problem.
Your problem is the opposite direction. You are the engineer, and you need to sell something to whoever your customer is. The other trap is app-install content, which answers “how do I run a mobile install campaign at scale” when the question was “does anyone want this”. Both will have you optimising a funnel before you have anyone in it.
What do I do in the first week?
Seven things, in this order, none of which involve choosing a channel. The whole point of the week is to find out whether anyone wants the thing, using the smallest amount of work that produces a real answer.
Decide what counts as success before you start
Write down the result that would prove people want this, for example ten payments, twenty replies or five people asking when it ships. With no target set in advance every result looks like good news.
Check somebody is already selling this
A competitor is proof that people pay for the problem. If nobody sells it, find out why before assuming you are first.
Talk to twenty people who have the problem
Calls or DMs, not a survey. Write down the exact words they use. Those words are your landing page, and you will not invent better ones.
Put the real price on a page
One sentence, one screenshot, one price, one button. A page people can try to pay for tells you more in a week than a month of building.
Set up the ad accounts today
Google, Reddit, LinkedIn, Instagram, TikTok. Approval takes a day or two, so doing this now costs nothing and not doing it blocks you later.
Buy $100 to $200 of traffic across three of them
Enough to learn whether anyone clicks and which audience is warmest. Once the first data is back, spend a second week trying different targeting and creatives on the platform that responded. It is not enough to learn what an ad costs at scale, which is a later question.
Charge from the first day
Even ten dollars. A payment is the only signal that separates interest from politeness.
Which engineering instincts get in the way?
The ones that assume the problem is technical. Four show up again and again.
Building before asking
Six months of clean architecture for a thing nobody wanted is the most expensive way to do customer research. The landing page test costs an afternoon and answers the same question.
Automating before it works manually
Fifty hand-written messages a week is the job at the start, and it does not scale on purpose. Automating it before it works means scaling something that does not.
Optimising a funnel with nobody in it
Conversion rate on eleven visitors is not a number. Get to a few hundred before you read anything into it.
Treating marketing as beneath the work
The judgement is understandable, most marketing you see is bad. The version that works is closer to debugging: form a hypothesis about why someone would pay, test it cheaply, believe the result.
Which ones actually help?
More than you would expect. Engineers are usually better at the parts of marketing that involve reading primary sources and refusing to guess.
You will read the ad libraries
Meta, TikTok and Google publish every ad running right now, for free. An ad that has been live for months is almost certainly paying for itself. Most marketers never look.
You will instrument it properly
One honest attribution column beats any dashboard. Engineers are the people who actually fill it in.
You can build the boring half
Answer-shaped pages, schema, an llms.txt file, comparison pages. A coding agent does most of this in an afternoon, and it is the part that compounds while you sleep.
You ship faster than you think
A landing page test, three ad variants and a comparison page is a day of work for you and a two week project for most teams.
Good to know
I hate self promotion. Is there a version of this that is not that?
Most of it, actually. Answering a specific question for a specific person who asked it in public is the highest yielding move at the start, and it is closer to helping than to promoting. The broadcast part of marketing, posting into the void hoping for reach, is the part that does not work yet anyway.
Should I learn SEO?
Not first. The foundations that let AI answers and search engines read your pages are worth an afternoon early, because they compound. Writing content for search waits until you know which words your customers use, which is usually around a thousand a month.
Can I just pay someone to do this?
Not at the start. Nobody can outsource finding out whether the idea works, because the answer comes out of conversations you have to be in. Once a channel demonstrably works, hiring a specialist for that channel is the right move.
How much should I budget?
$100 to $200 for the first week, split across two or three platforms, then a second week of targeting and creative changes on the platform that responded. Anything more buys precision you cannot use yet.
This expands the Trailhead camp on the climb, which lays out the stages either side of it. Tools by job are on the route index, the free ones are here, and the other answers are in insights. Updated 21 September 2026.