Sizing a market that does not exist yet

Core idea#

If your technology does something that wasn't possible before, there's no analyst report with your market's size in it. Founders respond in one of two bad ways: they borrow a huge number from an adjacent category ("the $400 billion chemicals market") or they give up and say the market is unknowable. Investors discount both.

The better answer is that new markets are almost never entirely new. Someone, somewhere, is already spending money or living with a cost to solve the problem, badly. Your job is to find that spending, work out how much of it your technology could capture or unlock, and show how fast adoption might realistically happen.

A good market-size estimate for a new technology has three parts:

1. A bottom-up number: how many customers, times what each would pay a year, built from things you can check.
2. An adoption path: how that number grows over 5 to 15 years, based on how comparable technologies spread.
3. Your assumptions, stated openly, so an investor can change them and see what happens.

Who it's for: founders of companies whose product creates a new category, enables something previously impossible, or replaces an existing method so different that today's market data doesn't describe it.

Four methods, and when to use each#

Use at least two. If they land within the same order of magnitude, you have a number you can defend.

1. Substitution: what does it replace?#

Most new technologies replace an existing way of doing a job, even if the old way is manual, expensive or partial.

• List the current methods customers use (people, equipment, outsourced services, workarounds).
• Estimate what customers spend on them a year.
• Ask what share of that spending your product would capture, at your price.

Best for: technologies that do an existing job better, cheaper or faster (new sensors, new materials, new manufacturing processes).

2. Bottom-up value: what is it worth to each customer?#

• Count the customers who have the problem, by segment.
• For each segment, estimate the economic value your product creates per customer per year (cost saved, revenue gained, risk avoided).
• Assume you can charge 10% to 30% of that value; customers need to keep most of it to switch.
• Multiply: customers × price.

Investors generally prefer bottom-up because the assumptions can be tested; Pear VC, after surveying 30 VCs, advises presenting bottom-up estimates and using top-down only as a sanity check (Pear VC).

Best for: B2B products where you can talk to customers about their costs.

3. Analogy: what happened with something similar?#

Find a past technology with a similar buyer, price point and switching difficulty, and look at how its market grew: LED lighting, industrial robots, genetic sequencing, solar panels, cloud computing.

Best for: sanity-checking your adoption speed and your eventual market share.

4. Adoption curve: how fast will it spread?#

Markets for new technology tend to follow an S-curve: slow at first, then faster as adopters influence each other, then levelling off. The Bass diffusion model describes this with two parameters, one for innovators who adopt on their own and one for imitators who follow others, plus a total market potential (Wikipedia). The parameters are usually borrowed from an analogous product, and published values exist for many technologies (Forecast Pro).

You don't need the full maths at pre-seed. You do need to show you understand that yearone revenue is a small fraction of the eventual market, and roughly when the steep part of the curve might start.

Best for: showing investors the timing, which matters as much as the size in deeptech.

Worked example (illustrative)#

A team has developed a low-cost wireless sensor that continuously monitors corrosion inside industrial pipes. Today, plants inspect pipes manually with handheld ultrasonic tools. All numbers below are invented to show the method; replace them with figures from your own customer interviews.

Step 1: Define the first segment precisely#

Refineries and large chemical plants in the US, Europe and India: about 1,800 sites (from industry directories and interviews).

Step 2: Find what they spend now (substitution)#

From 15 interviews with plant inspection managers:

• A typical site has about 2,000 monitoring points.
• Each point is inspected twice a year at about $150 per inspection.
• Manual inspection spend per site: 2,000 × 2 × $150 = $600,000 a year .
• Across 1,800 sites: about $1.08 billion a year .

Step 3: Estimate value per customer (bottom-up)#

• The sensor could replace about 40% of manual points: saving 0.4 × $600,000 = $240,000 a year .
• Earlier warning reduces the expected cost of unplanned shutdowns by an estimated $200,000 a year (managers' own estimate).
• Total value: about $440,000 per site per year .
• Price at about 20% of value: $90,000 per site per year (hardware plus monitoring subscription).

Step 4: Market potential#

• Bottom-up: 1,800 sites × $90,000 = $162 million a year .
• Substitution check: $162 million is 15% of current manual inspection spend of $1.08 billion. Plausible.
• Adjacent segments for later (pipelines, power plants, offshore): note them, but don't add them to the headline number yet.

Step 5: Adoption path#

Industrial monitoring technologies typically spread slowly at first, because plants trial a few units, then roll out site by site. Assume:

Step 6: What the investor sees#

• A first market of about $160 million a year, growing if adjacent segments open up.
• A credible path to $30 million or more in revenue by year 8.
• Assumptions they can test: points per site, cost per inspection, share replaceable, price as a share of value, adoption speed.

That's a more convincing story than "1% of the $20 billion asset integrity market", even though the number is smaller.

What investors look for#

• A number they can rebuild. If a partner can't follow the arithmetic in 30 seconds, it won't help you.
• A first market you can win, not just a large total. Venture investors need to believe the company could reach a meaningful revenue (often $100 million a year or more for a venture-scale outcome), but they also want to see where you start.
• Explicit assumptions and exclusions. Saying who is not in your market builds trust. Timing. In deeptech, the question "when does the steep part of the curve start?" decides whether the investment fits a fund's life. Show what has to happen first: regulatory approval, cost reduction, a standard, a pilot.
• A sense of upside. Once the first market is credible, it's fine to show where the technology could go next, labelled as upside.

Deeptech specifics#

• Cost curves change the market. If your cost falls by half every few years (as solar, batteries and sequencing did), new segments open. Show the market at today's cost and at your target cost.
• Platform technologies have many markets. Pick one beachhead and size it properly; list the others as options, not as additions to the headline number.
• Regulated markets (medical, aviation, energy) have long lead times. Put the approval timeline into the adoption curve.
• Willingness to pay is the weakest assumption. Test it in discovery interviews (playbook 1.19) and with paid pilots, and say how you tested it.

Data sources by country#

Analyst reports are useful for top-down sanity checks, but read their methodology notes; many are extrapolations from small surveys, and definitions vary between publishers.

Scripts#

Asking customers about current spending#

"Roughly how much does your site spend a year on [the current method], including people, contractors and equipment? And when it goes wrong, what does that cost you?"

Ask about the last real example, not hypotheticals. People are much more accurate about what happened than about what they'd do.

Presenting the slide#

"We start with refineries and chemical plants: about 1,800 sites that spend around $1 billion a year on manual inspection. We think we can charge about $90,000 a site, which is 20% of the value we create. That's a $160 million first market. If we reach 20% of sites by year eight, that's over $30 million in revenue, before power plants and pipelines."

When an investor says "that's too small"#

"That's our first market, sized conservatively from customer data. The same sensor works in [adjacent markets]; if half of those materialise, the market roughly triples. We'd rather show you the number we can defend and the options on top."

Common mistakes#

• The "1% of a huge market" slide. It shows you haven't worked out who the customer is.
• Borrowing a category that's too broad. "Advanced materials" or "healthcare" isn't your market.
• Assuming customers pay full value. They need to keep most of it to bother switching. Ignoring the adoption curve. A $1 billion market reached in 25 years is not the same as one reached in 8.
• Adding every possible application to the headline number. Keep upside separate. Hiding the assumptions. Investors will find them anyway. Show them and let the investor change them.
• Never talking to customers about money. The price is the hardest number; ask about it early.

Related playbooks#

• 1.07 Tech-push vs market-pull
• 1.09 Founder-market fit
• 1.19 Customer discovery from inside the lab
• 1.20 Writing a one-page thesis

Sources#

• Pear VC, Market sizing guide and survey of 30 VCs on market sizing.
• Bass diffusion model: overview; Forecast Pro example using analogous coefficients; Bass, "A New Product Growth Model for Consumer Durables", Management Science 15, 1969.

Worked example is illustrative; all figures are invented to show the method. Last checked November 2026.