What the term actually means
a16z put it more usefully than anyone else has: new media is best understood as a type of packaging designed to help you be interesting, rather than as a category of content.
That definition does a lot of work, because it moves the question away from which platform to post on and towards what is worth saying at all.
The jobs underneath it are old. Writing, video production, running social accounts. What changed is who does them and how much responsibility comes attached.
The roles now attract subject-matter experts who happen to be internet-native, rather than communications staff who happen to have been assigned a channel.
The rule that inverted
Old media rewarded staying on message and avoiding trouble. A founder went through media training, learned to bridge back to three approved talking points, and succeeded by being quotable without being memorable. The worst outcome was saying something that could not be taken back.
Ben Horowitz described the consequence bluntly: the one rule of old media is do not be interesting, and it is the worst thing you can do. That was rational when a handful of publications decided who got heard and a single bad quote followed you for years.
New media inverted it. Unlimited formats, unlimited channels, and a feed that ranks on whether anyone reacted. Being uninteresting is now the failure mode with the highest cost, because it produces silence rather than damage, and silence is invisible on a dashboard.
Packaging is the job
If new media is packaging, then the work is taking something true about your company and putting it in a shape somebody would choose to consume. The truth comes first and the shape comes second, and most companies get the order backwards by choosing a format and then looking for something to fill it.
The packaging question has a reliable test. Would a person who does not work for you, and has no reason to care about your company, finish this? If the honest answer is no, changing the platform will not help.
What helps is finding the version of the story that is about something happening in the world rather than about your company, which is covered in the bigger story your company sits inside.
The four things a company runs
A company that is not in the media business still ends up running a small number of things, and the list is shorter than the discourse suggests.
- A founder channel. One person, posting in their own voice, who will still be at the company in five years. This is the only one that cannot be delegated without becoming obvious.
- A longform surface. Somewhere arguments live at full length, on a domain you own. This is what gets cited, by people and by the systems answering questions about your category.
- A clip habit. Short extracts pulled from the longform, which do the reaching. The two ends together are the barbell.
- Launch amplification. Episodic, concentrated, and the one most companies staff badly because it is needed six times a year rather than continuously.
Everything else is optional at most sizes. A podcast, a newsletter, an events programme and a documentary team are all real assets and all of them assume you have the first four working.
Who actually does this
The hiring answer is unusually clear and unusually inconvenient. The bar a16z screens against is whether the candidate has done it: built some brand on something, built an audience, and can deliver end to end.
Heavy experience in traditional communications counts against a candidate rather than for them, because the two disciplines run on opposite rules.
The reason is storytelling. It is a real skill, the elite level of it sits far above the average level, and people from a conventional marketing background often never developed it because they always had a principal to supply the story. Hiring someone to package a story they cannot construct produces exactly the output you would expect.
The same bar is why this is hard to staff and easy to underestimate. a16z accepted 65 people from 2,000 applications into its media fellowship, at a firm that can pay whatever it likes.
The order it has to happen in
a16z describes four motions running in parallel, which is what a mature operation looks like from the outside. A company starting from nothing has to sequence them, and running them in the wrong order is the most common way this stalls before it compounds.
Longform first. You cannot cut a clip from something that does not exist. Teams that start at the short end produce a feed of confident one-minute takes with nothing underneath, and a viewer who wants to check whether you actually know the subject finds nothing to check.
The founder channel second, and early. Reach is measured against your own recent baseline, so an account silent for two months is capped before a launch begins. The channel has to be warm before you need it, which means starting it in a quarter when nothing is shipping and there is no reason to.
Clips third. Once an argument is on the record, extracting from it is the cheapest content you will ever make, because the expensive part was having the argument.
Launch amplification last. It is the only one of the four that is episodic, and the only one worth renting. Everything above it has to exist first, because a launch amplifies whatever is already there.
The sequence matters because each step is the input to the next one. A company that runs the launch first is buying reach for an empty channel, and the reach arrives, converts nothing, and is recorded as evidence that distribution does not work.
Where it breaks at a technical company
The strategic mistakes above are the ones people write about. These four are the ones we actually watch happen, and none of them is a strategy problem.
Approval removes the reason to read
The draft goes to legal, to product marketing and to two founders. Every pass removes the specific number, the named constraint or the sharp claim, because each of those carries a small risk and none of them is anyone's job to defend.
What ships is accurate and unreadable. The fix is a named owner with authority to publish rather than a better review process.
The longform assumes context the reader does not have
Technical teams write for peers who share their assumptions. The reader who could become a buyer is two steps behind, hits an unexplained term in the second paragraph and leaves.
This is the most common reason a genuinely good technical essay reaches forty people, and it is invisible internally because everyone who reviewed it already knew the missing piece.
The channel exists only around releases
Content appears for six weeks around a launch and then stops for a quarter. Each cycle starts from a cold account, which caps the launch it was built for, and the compounding never begins.
Consistency in a quiet month is worth more than volume in a launch week, which is the argument in weeks with nothing to announce.
The company account is asked to do the founder's job
Posting from the logo feels safer and is discounted automatically by everyone reading it. A brand account describing its own product carries the least credible voice available to you, and no amount of budget changes that. The mechanism is in the credibility ceiling on owned media.
What changes with size
Under twenty people this is one person, usually the founder, and the honest advice is to do less of it better. A single channel posted consistently beats four channels posted occasionally, and the compounding only starts once the cadence is real.
Between twenty and a hundred, the founder channel stays with the founder and everything else needs an owner. This is the range where distribution becomes nobody's job without anyone deciding it should, covered in who owns launch distribution.
Past a hundred it becomes a function with a headcount, and the build-or-rent question gets sharper. The framework for that, and what a16z productised for the companies it backs, is in launch as a service.
Questions we get asked
What does new media actually mean for a company?
Packaging rather than a category of content. The underlying jobs, writing, video, running accounts, are old. What changed is that the work is now done by subject-matter experts who are internet-native rather than by communications staff assigned a channel, and that the goal shifted from staying on message to being worth reading.
Why is traditional PR experience a disadvantage in new media?
Because the two disciplines reward opposite behaviour. Old media rewarded staying on message and avoiding trouble, since a handful of publications decided who got heard and a bad quote followed you for years. New media ranks on whether anyone reacted, so being uninteresting produces silence, which is the more expensive failure and the harder one to notice.
What should a company actually run if it is not a media company?
Four things. A founder channel in one person's own voice, a longform surface on a domain you own, a clip habit that extracts short pieces from that longform, and launch amplification for the six or so moments a year that concentrate attention. A podcast, newsletter or events programme assumes the first four already work.
In what order should a company build its media function?
Longform first, because you cannot extract a clip from something that does not exist. The founder channel second and early, because reach is measured against your own recent baseline and a silent account is capped before a launch begins.
Clips third, once there is an argument on the record to cut from. Launch amplification last, because it is the only episodic one and the only one worth renting. A company that runs the launch first is buying reach for an empty channel.