Bottom Up Market Analysis: Size the Market You Can Win

Bottom up market analysis builds a revenue estimate from real customer data rather than working backwards from a headline TAM figure. You start with the number of addressable buyers, apply realistic conversion assumptions, and arrive at a market size figure that reflects what your business can plausibly capture, not what the entire category is theoretically worth.

It is the method that holds up in a boardroom when someone asks how you got to your number. Top-down analysis rarely does.

I have sat through dozens of strategy presentations where the market size slide cites a multi-billion dollar TAM from a Gartner or IDC report, followed by a line that says the company is targeting just 1% of that market. It sounds conservative. It is not. It is a way of avoiding the harder question: who specifically will buy this, at what price, through which channel, and why now. Bottom up analysis forces you to answer those questions before you can produce a number at all.

If you are working through a broader research and intelligence programme, the Market Research and Competitive Intel hub covers the full range of methods that sit alongside market sizing, from customer interviews to competitive signal tracking.

What Is the Actual Difference Between Top-Down and Bottom Up?

Top-down analysis takes a large market figure, usually from a published report, and applies a percentage to estimate what a business might capture. The logic runs: the global market for X is $40 billion, we are targeting a niche within it, so our serviceable market is $4 billion, and our realistic share is $400 million.

The problem is that every assumption in that chain is borrowed from someone else’s methodology. You do not know how the analyst defined the market, which geographies they included, whether enterprise and SMB buyers were treated the same way, or what year the underlying data was collected. You are multiplying uncertainty by uncertainty and presenting the result as a market opportunity.

Bottom up works in the opposite direction. You identify the specific population of buyers you can reach, estimate what proportion of them fit your ICP, apply a realistic conversion rate based on your sales model, and multiply by your average contract or transaction value. The resulting figure is smaller, almost always, but it is yours. It is built from assumptions you can defend because they come from your own pipeline, your own win rates, and your own pricing.

Early in my career, I had a similar lesson about the difference between inherited assumptions and first-principles thinking. I needed a new website for the business and was told there was no budget. Rather than accept the constraint, I taught myself to build it. The point was not the website. It was that working from the ground up, with actual materials and real constraints, produces something more honest than working from a wish list. Market sizing is no different.

How Do You Define the Addressable Pool Before You Start?

This is where most bottom up analyses go wrong. Teams inflate the addressable pool at the start, which inflates every subsequent calculation, and the final number ends up being just as misleading as the top-down figure it was supposed to replace.

Defining your addressable pool means identifying buyers who have the problem you solve, the budget to pay for your solution, and the organisational profile that makes them a realistic sales target. That is a much smaller group than “companies in this sector” or “brands spending on digital.”

For B2B businesses, this is where a structured ICP definition becomes essential. A well-built ICP scoring rubric does not just describe your best customer in general terms. It assigns weighted criteria, firmographic filters, and behavioural signals that let you separate genuinely addressable accounts from accounts that are merely in the same industry. Without that precision, your addressable pool is a guess.

For consumer businesses, the equivalent discipline is demographic and behavioural segmentation. Age range and income bracket are a start. Purchase intent, category engagement, and channel reachability are what actually define the pool you can convert.

Forrester has written usefully about how solution type shapes buyer behaviour, which is relevant here because the same product can face very different addressable pools depending on whether it is positioned as a point solution, a platform, or a service. Your ICP definition and your market sizing need to be consistent with each other.

What Data Do You Actually Need to Run the Model?

A credible bottom up model needs four inputs: the size of the addressable buyer population, your conversion rate from prospect to customer, your average revenue per customer, and your retention or repeat purchase rate if the business has recurring revenue dynamics.

Each of these should come from internal data where possible. If you have an existing customer base, you can calculate your real conversion rate from first contact to closed deal. You know your average contract value. You know your churn rate. These numbers are not theoretical. They are observable.

For businesses that do not yet have trading history, the assumptions have to come from somewhere defensible. Comparable businesses in adjacent categories, pilot campaign data, or early-stage pipeline metrics are all more credible than industry averages from a report. When I ran paid search at scale across multiple categories, I saw firsthand how differently conversion rates behaved across sectors. A campaign for a consumer travel product could generate six figures of revenue within a single day from a relatively contained spend. A B2B software campaign in the same week might generate a handful of qualified leads. The unit economics were not just different in magnitude. They were structurally different. Any market sizing model that applies a uniform conversion assumption across different sales motions is not a model. It is a spreadsheet with opinions in it.

Search data is one of the most underused inputs in bottom up analysis. The volume of branded and category-level queries in your target market tells you something real about the size of active demand. Search engine marketing intelligence can surface not just demand volumes but also how that demand is distributed across competitors, which is a more grounded input than a market share estimate from a third-party report.

How Do You Stress-Test the Assumptions in Your Model?

A bottom up model that is not stress-tested is just a top-down model with more steps. The value of the methodology is in the discipline it imposes, and that discipline only holds if you challenge each assumption before you present the output.

The three assumptions most likely to be inflated are the addressable pool, the conversion rate, and the average revenue figure. On the addressable pool, the question to ask is: how many of these accounts have we actually spoken to, and what proportion converted? If you have not spoken to enough accounts to have a real conversion rate, your pool estimate is speculative.

On conversion rate, the question is whether you are using a full-funnel rate or a partial one. A lead-to-MQL rate is not a conversion rate for market sizing purposes. You need prospect-to-revenue, which is almost always a smaller number than teams want to use.

On average revenue, the question is whether you are using mean or median, and whether large outlier accounts are pulling the figure upward. If your top three accounts represent 40% of revenue, your average contract value is not a reliable input for sizing a new market.

Running scenario analysis, a conservative case, a base case, and an optimistic case, is standard practice. What is less common, but more useful, is documenting the specific assumption that changes between each scenario. That makes the model auditable and makes it easier to update as you accumulate real data.

Qualitative research has a role here too. Focus group methodologies can surface the friction points and objections that quantitative models tend to smooth over. If buyers consistently raise the same barrier during research, that barrier should be reflected in your conversion assumptions, not ignored because it complicates the model.

Where Does Competitive Intelligence Fit Into the Analysis?

Bottom up analysis tells you the size of the opportunity. Competitive intelligence tells you how much of it is already occupied and by whom.

These two inputs belong together. A market that is large by bottom up standards but dominated by two or three entrenched players with high switching costs is a different commercial proposition than a fragmented market of similar size. The sizing exercise is incomplete without the competitive context.

Competitive intelligence for market sizing purposes does not need to be exhaustive. You are not trying to produce a full competitive audit. You are trying to understand what share of the addressable pool is genuinely available to you, which competitors are winning which segments, and whether the dynamics of the market are moving in your favour or against you.

One source of competitive signal that is often overlooked in traditional market sizing is grey market activity. Grey market research can reveal where demand is being served by unofficial channels, parallel imports, or informal alternatives, which in some categories represents a meaningful proportion of total market activity that never appears in official figures.

BCG’s work on supply chain dynamics in retail is a useful reminder that markets are not static pools of demand. The structural shifts in fashion supply chains they documented illustrate how quickly competitive positions can move when the underlying economics of a market change. A bottom up model built on today’s competitive landscape may not reflect where the market is heading in 18 months.

How Does Pain Point Research Sharpen the Output?

Market sizing and pain point research are usually treated as separate workstreams. They should not be. The pain points your buyers experience are what determine whether they are genuinely in-market for a solution like yours, which directly affects the quality of your addressable pool estimate.

A buyer who has the right firmographic profile but is not actively experiencing the problem you solve is not in your addressable market. They are a future prospect at best. Including them in your pool inflates the model without improving the commercial reality.

Structured pain point research helps you understand not just what problems buyers have, but how urgently they feel those problems, how they are currently addressing them, and what would need to change for them to consider a new solution. That kind of insight is directly translatable into conversion rate assumptions. A buyer who is actively frustrated with their current solution and has budget available is a fundamentally different conversion prospect than a buyer who is mildly aware of a problem but not motivated to act.

I have seen this play out in agency pitches more times than I can count. A client presents a market sizing exercise that looks compelling on paper, but when you probe the underlying buyer research, it becomes clear that the assumed conversion rates are based on awareness-level interest rather than purchase intent. The model is technically correct but commercially misleading. Pain point research closes that gap.

How Does Technology Strategy Affect Which Markets Are Addressable?

This is a dimension of bottom up analysis that gets very little attention. The markets you can realistically address are partly a function of your technology infrastructure. A business that cannot support self-serve onboarding cannot efficiently address SMB buyers at scale. A business without multilingual capability cannot address markets where English is not the primary language of commerce. These are not marketing problems. They are technology constraints that directly limit the addressable pool.

A technology strategy and SWOT analysis aligned to your commercial objectives will surface these constraints before they become expensive surprises in a market entry programme. If your bottom up model assumes you can address a market segment that your current technology stack cannot serve, the model is wrong regardless of how good the demand signal looks.

This is particularly relevant in international expansion. BCG’s analysis of operational complexity in travel and transportation highlights how structural constraints, whether regulatory, operational, or technological, can dramatically reduce the share of a theoretical market that is practically accessible. The same logic applies to any market entry where the business has capability gaps relative to what the target segment requires.

When I was growing an agency from a team of 20 to over 100 people, the markets we could credibly address expanded not because the market changed, but because our capability expanded. Bottom up analysis is not a one-time exercise. It should be revisited as the business scales, because the addressable pool changes as your delivery capacity and technology infrastructure evolve.

What Does a Credible Bottom Up Output Actually Look Like?

A credible bottom up market analysis is not a single number. It is a structured argument with a number at the end of it.

The output should document the addressable pool and how it was defined, the conversion assumptions and where they came from, the average revenue figure and how it was calculated, and the competitive context that affects what share of the pool is realistically available. It should include a conservative, base, and optimistic scenario, with the specific assumption that differs across each scenario clearly stated.

It should also state what would need to be true for the optimistic scenario to materialise. That is the discipline that separates a rigorous model from an optimistic one dressed up in spreadsheet format. If the optimistic case requires a conversion rate that is three times your current pipeline rate, that assumption needs to be visible and justified, not buried in a cell reference.

The output should also flag the data gaps. Where did you have to use industry proxies rather than internal data? Which assumptions are most sensitive to small changes? A model that acknowledges its own weaknesses is more credible than one that presents false precision. I spent time as an Effie Awards judge, and one of the consistent differences between entries that held up under scrutiny and those that did not was the willingness to show the reasoning, not just the result. The same applies to market sizing.

For a broader view of the research methods that feed into commercial planning, the Market Research and Competitive Intel hub covers the full landscape, from primary research through to competitive signal analysis and demand forecasting.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions