Your Marketing Is Buying Logos. The Buyer Just Wants the Math.

a16z partners Joe Schmidt and Julian Marx published an essay on July 27 arguing that AI companies selling into the enterprise face one expensive early decision, and that most of them get it wrong by reflex. Schmidt taped a podcast episode on it on August 13, which is where the argument got blunt enough to be useful.

The two options, in their words. Lighthouse: “win a few marquee customers, build social proof, and reassure the buyer who’s afraid of making the wrong call.” Landgrab: “win on math, move fast, and sign the largest number of customers possible, logo be damned.”

Founders default to Lighthouse. They assume the market needs educating, so they chase the recognizable name that will make everyone else feel safe.

Schmidt’s counter, at 00:23 in the episode, is a story about selling enterprise networking in 2009:

“If you think about sort of the enterprise networking world in 2009, people thought we were crazy. Like we had no chance of getting into the largest corporations in the world, Lighthouse, because Cisco and HP had them all tied up. But what we could do is we could say, listen, we can configure, we can deploy faster, we’re simpler to use. And that was very much a land grab strategy.”

At 00:41 he names what actually drives the default:

“Too few people are willing to pick up the phone and willing to get on the plane and willing to get, you know, in front of those customers right now”

This is a marketing budget question wearing a sales costume

The essay is written for founders picking a sales motion, and it lands on the marketing plan first, because the two strategies want opposite things from the same budget.

Lighthouse marketing buys reassurance. Case studies with a name on them, analyst briefings, the conference booth next to companies your buyer has heard of, a website that leads with logos. It’s slow, it’s expensive per account, and it works when the buyer’s real fear is being blamed for the decision.

Landgrab marketing buys proof and reach. A calculator, a trial that proves the number in a week, comparison pages, paid search on the problem rather than the category, and self-serve everything. It works when the buyer already knows what the problem costs and only needs the numbers to clear.

Run the wrong one and nothing breaks loudly. Pipeline just stays thin while everybody agrees the brand is looking stronger.

The two questions that pick for you

Schmidt and Marx reduce it to two. The first is buyer exposure: how much personal career risk does the purchase create for the person signing? Replacing a system of record at a regulated bank is high exposure, and high exposure is Lighthouse territory, because the buyer isn’t shopping for ROI, they’re shopping for cover.

The second is whether social proof travels. In concentrated, status-driven markets like law and finance, a logo carries. In fragmented markets it doesn’t, and every marquee customer you win buys close to nothing with the next prospect.

Most AI tooling aimed at marketing teams fails both tests. The buyer is a VP with a budget and no board resolution required, and the market is thousands of companies who have never heard of each other. That’s a Landgrab market, and a great deal of the AI marketing category is running Lighthouse plays into it.

The pitfalls list is where the essay stops being theory. On the Lighthouse side: hostage-to-logo dynamics, prestige without payback, pilot purgatory, and building a product for a single customer. Anyone who has watched a marketing team spend two quarters producing collateral for one named account will recognize all four. On the Landgrab side: weak qualification, scaling before the product is ready, and mistaking geographic coverage for market opportunity.

Why AI made the choice sharper

Two things changed at once. Buyers got faster at understanding what these tools do, because they’ve been using consumer versions of them personally for three years. And the products got easier to evaluate in a week, because most of them will show you output on your own data before you sign anything.

Both push toward Landgrab. When the buyer already understands the problem and can test the answer on Tuesday, the marquee logo is doing less work than it did when nobody could evaluate a category without an analyst’s help. That’s the same shift underneath B2B deals being decided before the pitch, and it’s why the old blog-and-rank engine stopped feeding small-team pipeline.

The counterweight is that AI also raised buyer exposure in some rooms. Anyone signing off on a system that writes to a customer-facing surface has more career risk than they had buying a reporting tool, which pushes those specific purchases back toward Lighthouse. The framework holds. The answer just differs by product.

There’s an honest limit worth naming. This is a venture firm’s framework, published by people whose portfolio benefits from founders moving faster and signing more customers, and the essay is a year zero document rather than measured evidence. It offers no data on how the two motions performed across a cohort.

Take the diagnostic and leave the advocacy. Ask what happens to the person who signs your contract if it goes badly, and ask whether your best customer’s name means anything to your next prospect. If the answers are “not much” and “no,” every dollar of marketing spent manufacturing reassurance is buying a product your buyer isn’t shopping for.