Stop Strategizing. Start Selling. The Lighthouse vs. Landgrab Framework That Actually Works.

The most expensive mistake in AI right now isn’t picking the wrong sales strategy. It’s spending 99% of your time trying to pick one. You can feel the paralysis in every founder call: Should we chase the big whales or carpet-bomb the mid-market? Do we need a lighthouse or a landgrab? Meanwhile, the market is moving, the AI budget window is closing, and your competitors are taking your customers while you’re building PowerPoint frameworks.

This isn’t a theoretical debate. It’s the single most practical decision an AI startup founder will make this year. And according to Andy McCall, the a16z partner who built the sales machines behind two billion-dollar companies, the answer isn’t a calculation. It’s a conversation.

McCall’s track record speaks for itself. He joined Meraki in 2009 when Cisco and HP owned the enterprise networking market. A no-name startup couldn’t get within 100 yards of a Fortune 500 CIO. So they didn’t try. They sold to the mid-market, gave away free access points, and let the simplicity of their cloud-managed gear do the talking. Cisco acquired them three years later.

Then came Samsara in 2017. A two-year-old company selling connected sensors and vehicle telematics. The ELD mandate was about to force every trucking company in America to adopt electronic logging devices. Suddenly, there was a massive, unallocated budget pool. McCall didn’t send his team after the biggest carriers. He went after the mid-market where decisions were fast and feedback was faster. Within six years, Samsara went from millions to a billion in revenue and went public.

Neutrality in sales strategy is death. You’re not choosing between good options. You’re choosing between speed and proof.

So what’s the actual framework? Joe Schmidt, McCall’s partner at a16z, has been studying this exact problem. He calls it the Lighthouse vs. Landgrab matrix. It’s deceptively simple, but it cuts through the noise.

Ask yourself two questions. First: What is the perceived risk for the buyer? If they bet on you and you fail, do they lose money, their job, or their compliance standing? Second: How quickly does social proof travel in this industry? Does one marquee win convince a whole sector, or do you have to knock on every door yourself?

If you answered “high risk” and “strong proof spread,” you’re in a Lighthouse market. Think Harvey for legal AI. They didn’t try to sell to every law firm. They went after the most prestigious, most conservative, most risk-averse firms on the planet. Why? Because once the top-five international firms are using you, every other GC in the world says: “If it’s safe for them, it’s safe for me.” That’s the power of an immovable lighthouse.

If you answered “low risk” and “weak proof spread,” you’re in a Landgrab market. The budget already exists. The buyer doesn’t need a vision story. They need a mathematical proof: your AI tool is cheaper, faster, and more effective than the current software or the team of humans they’re paying for it. Stuut, an AI accounts receivable company, is the textbook case. They walk in with a spreadsheet, show you the ROI, and close the deal. No missionary work required.

But here’s the twist that most founders miss: this isn’t about choosing your identity. You can be a lighthouse company that also executes a landgrab. Samsara started with a landgrab in the mid-market and then pivoted to lighthouse mode when it wanted to conquer public sector or enterprise verticals. Harper did the opposite with insurance. The point isn’t which quadrants you can occupy. It’s that you need to stop deliberating and start discovering which game you’re actually in.

The only way to know which game you’re playing is to enter the game. Strategy is revealed through execution, not contemplation.

Ten fast closings in the mid-market will teach you more about your product than a year of trying to close one JP Morgan. Mid-sized customers don’t require social proof. They don’t have endless procurement cycles. They just want to know: does it work? Can I deploy it fast? Will it save me money? They are your fastest feedback loop, your best product R&D, and your quickest path to revenue. As McCall bluntly puts it: “The market will tell you which game you’re playing. If they’re willing to pay you, you’re playing landgrab. If they’re not, it’s time to go find a lighthouse.”

How do you know which one you’re in? Follow the budget. Can your customers transfer existing spend from a current vendor to you? That’s landgrab territory. If they have to create a new category in their minds, get internal alignment, and educate the org just to write the check, you’re in lighthouse territory. And extremely few startups can survive a year of missionary work.

The uncomfortable truth is that most founders want to be in lighthouse mode. “Selling to JP Morgan” sounds incredible for your ego and your cap table. It feels like validation. But as Joe Schmidt points out, ten customers who close in six weeks are more valuable than one customer who takes eight months. The ACV (annual contract value) curve will naturally rise as you mature. You don’t have to force it. You just have to build a repeatable engine.

Which brings us to the real problem. It’s not that founders can’t figure out whether they’re lighthouse or landgrab. The problem is that they’re spending 99% of their time in the analysis, and 1% in front of customers. McCall’s advice is violently anti-analyst: “Don’t spend too much time in analysis paralysis. Go talk to your clients. Find out who will buy what you currently have, and then walk down that path. You can re-evaluate after the first year.”

You don’t get extra credit for hard-to-earn revenue. You don’t get a higher revenue multiple for landing the famous logo on your homepage. Just go close the customers who are willing to pay you.

This also requires a brutal audit of your POC pipeline. AI products are changing faster than enterprise procurement cycles. If your concept phase has no deadline, no defined scope, and no success criteria, you’re not selling. You’re running a science project. McCall suggests a 30-45 day window with a clear “what we will prove” statement. If the software achieves it, they buy. If not, it was a failed experiment. This is how you separate the tire-kickers from the budget-holders.

And for the love of God, stop with the Product-Led Growth delusion. This AI moment is not like the cloud-to-cloud software era. That was a one-for-one button swap. Nobody switches CRMs because your button is blue and the old one is green. But now, enterprises are looking at fundamentally different business processes. They’re not trying to digitize the old workflow. They’re trying to invent a new one where AI does the work. This is not a self-serve product question. This is a “salesperson who can articulate a future state” question. It’s time to build a door-knocking machine, not a download page.

The era of credible PLG for enterprise software is over. The new era is AI’s landgrab, and it is being won by people who are unafraid to shake hands, hop flights, and build a sales org with the hunger of a startup.

You don’t persuade an organization to buy your future. You’re either proving someone right, or you’re proving the math. Pick the one that gets you a check this quarter.

So where do you start? Not with a whiteboard. With a road trip. Find the customer in the mid-market who has a problem and a budget. Offer them a trial. Prove your value. And let that momentum carry you to the list of names.