Two people, one platform
A six-figure platform purchase at a large company involves a committee, a security review, a procurement process and a signature from someone senior. Marketing organisations look at that chain, identify the signature, and aim everything at the person holding the pen.
That aim is half right. Executives at this level are reachable, and in AI they are often reachable on exactly the same feed as everyone else.
Plenty of CTOs at large companies read the technical accounts, post in public and form views about vendors long before procurement opens a file. Writing them off as unreachable is how enterprise marketing ends up buying trade-press placements that nobody in the decision reads.
What the aim misses is timing. By the time a proposal reaches a signature the field has already been narrowed to two or three vendors, and the executive is choosing between options somebody else assembled.
Winning that meeting matters. Getting into the set of three happens months earlier and is decided by a different person, and that is the part distribution solves.
Our own thesis says this plainly: if you sell to procurement rather than to practitioners, you need the credibility artifacts that PR still produces, and distribution complements those rather than replacing them.
The full version, including who it does not apply to. This page is about the enterprise case specifically, where both are true at once and the sequencing is what people get wrong.
Who writes the shortlist
Ask any enterprise AE where a deal actually started and the answer is rarely the executive. It is a staff engineer who had been following the space, a platform lead who ran a weekend evaluation, or a newly hired head of AI who arrived with three names already in their head.
Those people have three properties that matter enormously.
- They are on X daily, and the technical AI conversation happens there rather than anywhere else. This is the same audience concentration the rest of this site is about.
- They are the internal champion, which means they have to sell you upward. A tool their peers have heard of is dramatically easier to defend in a room than one nobody recognises, and the phrase "I've seen a lot about them" does more work in that meeting than any deck you supply.
- They are unreachable by advertising, for the structural reason that most of them pay monthly to remove it. The paid route to this buyer is weak by construction.
So there are two people to reach and they share a platform. The champion needs something specific enough to file you under a category.
The executive needs to recognise the name when the champion raises it, so the champion spends their credibility arguing for you rather than explaining who you are. One campaign does both, which is the part paid cannot replicate and PR only does for the second.
The artifact you are trying to create sits months upstream of a lead. It is the moment a platform lead reads something specific about how your system handles a hard case, and files your name under the category.
Three jobs distribution does here
Be precise about the mechanism, because the enterprise version is genuinely different from the self-serve one and the difference is where budgets get wasted.
- It gets you into the consideration set before the cycle starts. Nobody evaluates a vendor they have never heard of. The evaluation list is assembled from memory and a few searches, and memory is built months in advance.
- It arms the champion. Your buyer inside the account has to justify you to a committee. Every independent thing written about you is ammunition, and there is a large difference between a champion sending a vendor deck and a champion sending three pieces other people wrote.
- It makes outbound answerable. A cold email from a company the recipient has seen discussed gets a reply at a rate a genuinely cold one does not. This is the least glamorous of the three and often the one with the clearest effect on the sales team's week.
What this buyer's layer actually reads
The instinct at enterprise scale is to make the content more corporate as the deals get larger. It is precisely backwards for the audience that writes the shortlist, because seniority in engineering correlates with a sharper allergy to marketing language rather than a duller one. The audience that detects and publicly rejects marketing covers this in general; the enterprise-specific version is short.
| Lands | Does nothing |
|---|---|
| A benchmark with the methodology attached, including the cases where you lose | "Enterprise-grade performance at scale" |
| An architecture note explaining a decision and its tradeoff | A feature list |
| A post-mortem on an outage, written by the person who fixed it | A status page nobody links to |
| How you handle the hard case in their environment: air-gapped, on-prem, their compliance regime | "Built for the enterprise" |
| A named customer describing a specific problem in their own words | A logo wall with no names attached to any claim |
The consistent property of the left column is that a competitor could not have written it. Anything true of every vendor in your category is filler that happens to be about you.
One more that punches above its weight at this size: the security and compliance page. It gets read by evaluators, searched by name, and quoted in internal write-ups, and most companies bury it.
Treat it as a distribution asset, keep it ungated, and make it specific enough to answer the questions a security review will ask anyway.
Where credibility artifacts still matter
The honest boundary of the argument, which the rest of this site states and is worth restating here.
- Procurement and security review. Once a deal is in the process, the questions become about whether you are a real company that will exist in three years. Case studies, a named leadership team, a security posture and a LinkedIn presence that looks like an organisation all do work here that no amount of reach substitutes for.
- LinkedIn, at the right point. It earns its place once a purchase runs through procurement or a security review, and rarely before that. The mistake is treating it as the awareness channel when it is the reassurance one. What it carries that X does not.
- Analyst placement, in some categories. Where a category has an established analyst framework, appearing in it is a procurement checkbox in certain organisations. Worth knowing whether yours is one of those before deciding it does not matter.
- Reference customers. The single highest-value artifact in enterprise, and the hardest to get. A champion will ask for one, and a good answer closes more of the gap than anything above it.
The sequence that works: reach builds the memory, your own technical content converts the memory into evaluation, and the artifacts survive the process. Doing them out of order produces the common enterprise failure of a company with excellent case studies that nobody has heard of.
What actively hurts at this size
Larger deals raise the cost of every credibility error, because the people evaluating you are paid to find them.
- Claims you cannot support in a technical conversation. A benchmark quoted without methodology gets asked about, and the follow-up question is where the deal is lost.
- Logos on the site without a claim attached. Evaluators read a bare logo wall as a pilot that went nowhere until told otherwise, and increasingly they are right.
- A demo that requires a call. The person writing the shortlist wants to try it on a Saturday. Making that impossible removes you from the list at the cheapest possible moment for your competitor.
- Launching before the product holds up. Distribution accelerates the discovery that something is not ready, and at enterprise scale the people who discover it have long memories and public accounts.
Measuring against a nine-month cycle
Judging a launch on closed revenue inside a nine-month cycle guarantees the wrong conclusion, since nothing that the launch influenced has resolved yet and everything that closes this quarter was influenced by something else.
Measure the leading indicators instead, all of which move inside weeks.
- Branded search volume, before and after, which usually steps up and stays up.
- Inbound conversations, counted as conversations rather than form fills, and specifically inbound from accounts on your target list.
- Unprompted recognition in discovery calls. Ask the sales team to record how often a prospect already knows who you are. It is the crudest instrument here and the most persuasive one internally.
- Time to shortlist, if your CRM can carry it. Deals sourced from people who already knew you move faster, and that difference is the cleanest financial argument for this whole approach.
The full measurement scheme, including the baseline you need to record first, is in attributing signups to a launch. The enterprise adjustment is to read at 90 days rather than 30, and to accept that the honest report contains a confidence level rather than a return figure.
Questions we get asked
Do CTOs and CIOs buy software because of social posts?
Rarely from a single post, and scepticism about that is fair. What posts do is make a name familiar before a purchase starts.
Senior technical leaders at AI companies read X, and so do the staff engineers, platform leads and heads of AI who draft the shortlist they choose from. Reaching both means your internal champion is defending an argument rather than introducing a stranger.
Should enterprise AI companies use LinkedIn instead of X?
Both, at different points. X is where the practitioners who write the shortlist spend their attention. LinkedIn earns its place once a purchase runs through procurement or a security review, where a buyer is checking that the company looks like an organisation rather than discovering it exists.
How do we measure a launch when the sales cycle is nine months?
Not on pipeline, because nothing has resolved yet. Measure the leading indicators instead: branded search, inbound conversations, how often your name appears unprompted in discovery calls, and whether new opportunities arrive already knowing what you do. Judging a launch on closed revenue inside a nine-month cycle guarantees the wrong conclusion.