A go to market strategy for deeptech is one page. The hard part is what you leave off.

The funnel that never happened

A membrane technology startup once showed me their go to market plan. It was beautiful. Awareness campaigns, an inbound funnel, MQL targets, a CAC payback model, a partner channel in year two. I asked one question: how many potential customers exist in your beachhead? The answer was about two hundred companies, of which maybe forty were realistic. Their plan assumed a funnel fed by thousands of anonymous buyers. Their market was forty named factories with named plant managers who already knew each other.

That mismatch is the norm. Nearly every GTM template in circulation was written for software, where distribution scales with marketing spend and buyers self-serve their education. Deeptech sells a handful of high-stakes decisions to committees of engineers and budget holders. When the market is two hundred companies, you do not need reach. You need a map.

Why the software template breaks

Three assumptions in the standard GTM template fail on contact with industrial reality. First, volume: funnels assume you can afford to convert two percent of a large crowd. With forty realistic buyers, you cannot afford to lose any of them to a badly sequenced first meeting. Second, anonymity: software marketing works on strangers. Industrial buying works on trust, and trust in these markets comes from references, plant visits and a person whose phone number the buyer already has. Third, speed: funnels assume weeks. An industrial sale runs twelve to twenty-four months and most of that time is not your pipeline stage, it is their internal process, which I have written about in the capex approval piece. A strategy that does not model the buyer's internal calendar is a fantasy with good typography.

The one-page structure

The GTM documents that survive contact with reality all fit on one page, and they all answer the same five questions in the same order.

One: the beachhead. One segment, narrow enough that the buyers know each other, with the pain expensive and the budget owner named. Not a list of industries: a named list of the forty accounts, ranked. If you cannot list the accounts, you have not picked a segment yet, and I have written separately about how that choice works.

Two: the buying map. For a typical account in the beachhead, the five roles that decide: who feels the pain, who owns the budget, who can veto on technical grounds, who runs the procurement process, and who signs. With the names of real people at your first ten target accounts filled in, not personas but people.

Three: the proof ladder. The sequence of evidence an account needs before it can say yes: usually a credible reference, a paid pilot with written criteria, and a value case in their own numbers. The strategy is not how do we reach accounts, it is which proof do we build first and which account gives it to us.

Four: the milestones with dates. Not revenue targets pulled from a fundraising model, but commercial events: first paid pilot signed by month X, first reference visit hosted by month Y, second account in procurement by month Z. Dates create the pressure that forces sequencing decisions.

Five: what you will not do. The industries you are declining, the inbound requests you will park, the conferences you will skip. This line is the strategy. Everything above it is planning.

Sequencing is the whole game

When I review a deeptech GTM with a founding team, the discussion is never about channels. It is about order. Which account first, because its reference unlocks the others. Which proof before which account, because one plant manager's site visit is worth forty case studies. Which internal gate at the customer's side determines your calendar, because your pilot is only the third step in their nine-step approval. Teams that sequence well look slow in month three and unstoppable in month eighteen. Teams that optimize for activity look busy the whole way down.

Where marketing fits in all this

Marketing is not absent from this picture, it is just pointed at a different target. In a forty-account market, marketing's job is to make your name familiar before the first meeting and to arm your champion after it: the article their CFO reads, the one-pager their plant manager forwards, the reference story that circulates inside the account without you in the room. Measured not in leads but in how many of the forty buying committees have already heard of you from someone they trust. That is a different discipline from demand generation, and it compounds instead of evaporating when you stop paying.

Common questions

What is a go to market strategy for a deeptech startup?

One page answering five questions: the named beachhead accounts, the buying map of the five deciding roles, the proof ladder of evidence an account needs, dated commercial milestones, and an explicit list of what you will not pursue. Software-style funnel metrics do not apply to markets of forty named buyers.

Does a deeptech startup need marketing?

Yes, but aimed differently: at making your name familiar inside forty buying committees and arming your champions with material that circulates internally. Measured in recognition within the segment, not in lead volume.

How long does a deeptech go to market take?

Plan on twelve to twenty-four months per account, most of it the customer's internal process. A good strategy models the buyer's internal calendar rather than assuming your pipeline stages drive the timeline.

If your GTM is a funnel diagram and your market is forty factories, we should talk. Tell us what you have built.