Gains:
- Ability to construct and cross-check the size of a market with artificial intelligence support using top-down and bottom-up methods
- Ability to clearly write the TAM/SAM/SOM distinction and the chain of assumptions and test each assumption with real data
- Ability to detect and correct pitfalls in market size calculation (fabricated base number, double counting, unrealistic share)
“How big is this market?” question is at the heart of almost every strategy, investment and market entry project. If you get the answer wrong, the entire business plan built upon it will collapse. Market size forecasting is the most classic and most tested skill in consulting. Artificial intelligence helps in this task in two ways: it quickly builds the skeleton of the calculation and lists the assumptions. But here is the most dangerous mistake; The model can fit a base number (e.g. number of digits) and the whole calculation builds on this flimsy foundation. In this unit, we will learn how to make safe predictions using two independent methods.
TAM, SAM, SOM: three rings
Market size is not a single number; It consists of three interlocking rings.
- TAM (Total Addressable Market): All demand that the product/service theoretically addresses. "Everyone who drinks coffee in Türkiye."
- SAM (Serviceable Available Market): The part of TAM that your business model can actually reach. “Third generation coffee drinking, urban, premium segment.”
- SOM (Serviceable Obtainable Market): The share of SAM that you can realistically earn in the early years. "The share we will get in 4 provinces in the first 3 years."
The classic mistake in consulting is to mistake TAM for SOM. "50 billion market" sounds great, but the company's realistic goal for the first year is perhaps 200 million. Asking the artificial intelligence for all three rings separately and with their assumptions avoids this confusion.
Your role: market size analyst.Task: define and calculate TAM, SAM and SOM separately for [product/service].- TAM: theoretical total demand (definition + calculation).- SAM: the part that our business model can actually reach (what constraints narrowed it?).- SOM: realistic achievable share in the first 3 years (to what share, on what grounds?).Rules: write clearly under what assumption each ring narrows; I gave the base numbers, they are fictitious; Give each ring as spacing.
Two methods: top-down and bottom-up
The secret to a safe prediction is cross-checking with two independent methods.
- Top-down: You start with a larger total and allocate shares. "Türkiye's retail turnover is X; Y% of it is food; Z% of it is snacks."
- Bottom-up: You build from the bottom up. "How many households × how many people × number of annual consumption × average price."
If the results of two methods are close to each other, confidence increases. If it differs greatly, an assumption is faulty and you try to find it. Never fully trust the number obtained by a single method.
Your role: market size analyst. Task: Calculate the bottom-up of the Turkish [product] market. Rules:- Set up the calculation step by step: a single multiplier at each step.- Write ON the assumption for each multiplier and state its source; if I don't have a source, put a label "[ASSUMPTION - must be verified]". - Give the result as a low-high RANGE, not a single exact number. - Don't use a made-up base number; I will give the base data: [household=..., population=...]Format: "step = multiplier (assumption/source)" on each line.
Set up the same market as a top-down and compare the two results:
Your role: market size analyst. Task: Calculate the top-down of the [product] market.- Start from a large known total at the top (I gave the source: [total]).- Allocate sequentially: a single percentage at each step and the source/assumption of that percentage.- Check and flag the risk of double counting at each step.- Range the result and compare it with my bottom-up result (below); If the difference is greater than 20%, investigate which assumption might be responsible. My bottom-up result: [value]
Tip: You give the base numbers (number of households, population, official turnover) to the model. These come from official statistics and are the most critical basis of calculation. If you let the model "remember" these numbers, the whole calculation falls on a fictitious basis.
Writing the chain of assumptions clearly
A market size calculation is a chain of assumptions. Every link in the chain must be negotiable. A good advisor defends not the number, but the assumptions that lead to the number. Therefore, it is essential to write each assumption on a separate line, with its source or with the "must be verified" tag.
step
Value
Source / Assumption
Number of digits
26 million
TURKSTAT (source)
Rate of households using the product
35%
Industry survey (source)
Annual purchase quantity
4
Assumption - must be verified
Average unit price
180 TL
Retail observation (source)
Bottom-up FULL
≈ 6.5 billion TL
(26M×0.35×4×180)
Anyone who looks at this table moves the discussion to the right place: "Is 4 annual units realistic?" It is not the number that is discussed, but the assumption. For sensitivity analysis, you can derive lower and upper bounds by making this assumption 3 and 5.
Sensitivity analysis: think range
An odd number is misleading because it gives false precision. Instead, run the two or three most uncertain assumptions with their low and high ends to produce a range. The sentence "The market is in the range of 5.2–8.1 billion TL, the most likely value is ~6.5 billion" is both more honest and more useful for the decision.
In my bottom-up calculation below, 3 assumptions are unclear: usage rate, annual units, price. Task: Run each of the 3 low/medium/high scenarios and produce a low-medium-high range for the total market. Show the account step by step; I will check manually.
three mini cases
Case 1 — Rotten sole. A consultant asks "How many SMEs are there in Türkiye?" without providing resources to the model. he asks; the model says "about 5 million." The actual official figure is ~3.2 million. The entire market account swells by 56%; The investment board catches the error in the presentation. The correct way was to give the base number from the official source and let the model do only the multiplication.
Case 2 — Two methods conflicted, error found. For a beverage market, top-down is 9 billion and bottom-up is 4 billion. The difference is big; The consultant stops and compares the assumptions. In the top-down, the "snack" category also includes beverages, so there's double counting. When corrected, the two methods meet around 4.5 billion. Cross-checking prevents the wrong number from entering the presentation.
Case 3 — Mistaking TAM for SOM. A startup presents it as "our target market is 50 billion TL". The consultant makes the TAM/SAM/SOM distinction: TAM 50 billion, but the real business model only reaches the premium segment in 3 provinces (SAM ~3 billion) and the realistic share in the first 3 years is ~250 million (SOM). When the right size is presented to the investor, realistic expectations are established and the company gains confidence in the next round.
Weak prompt / Strong prompt
Weak prompt:
How big is the Turkish coffee market? Give a number.
It generates a single, unsourced, and unverifiable number from model memory.
Powerful prompt:
Your role: market size analyst. Task: Calculate the Turkish instant coffee market using TWO methods: top-down and bottom-up. Base data (I have the source): population=85M, households=26M, [other...]. Rules:- Calculate TAM, SAM, SOM separately and write their definitions.- Give each assumption on a separate line, with source or "must be verified" tag.- Present the result as a range; If the two methods are different, investigate why. - Check the risk of double counting. Format: two tables (top-down, bottom-up) + comparison note.
Logic filter: is the number reasonable?
Put each market size result through a logical filter before trusting the calculation. The simplest test is to compare the result with a known quantity: is the market you have calculated a reasonable proportion of the country's total retail turnover or the known total of the relevant sector? Another powerful check is the “per capita” or “per household” discount: If a market of 6.5 billion people divided by 26 million households yields ~$250 per household per year, does this figure match your actual spending habits? An assumption is incorrect if it does not fit. You can also have the AI do this reverse check: “Express this result per household and as a percentage of the total economy; does it seem reasonable, which assumption is most fragile?” The experienced consultant "sniffs" a number from two or three different angles before presenting it; Any counterintuitive conclusion is either a mistake or a truly important insight, both of which deserve investigation.
Common mistakes
- Making the base number fit the model. Basic data such as population, households, official turnover should come from official sources; Relying on the model's memory disproves the entire calculation.
- Being satisfied with one method. Do not trust the resulting number without top-down and bottom-up cross-checking.
- Mistaking TAM for SOM. Confusing the theoretical total market with the realistic available share misleads the investor.
- Double counting. When two methods conflict, an item is often counted twice; chase the difference.
- Presenting an odd number. Give a range and sensitivity analysis that reflects uncertainty rather than false precision.
- Hiding assumptions. The assumption, not the number, is discussed; Write each assumption clearly.
Caution: If a market size number is to be the basis of an investment decision, each assumption in the calculation should be independently checked and, where necessary, approved by a financial advisor or industry expert. Even if the model calculates correctly, an assumption error will mislead the entire result; The ultimate responsibility lies with the consultant.
In summary
Market size estimation is the most critical calculation in consulting and the most easily made incorrectly. AI quickly establishes the skeleton of the calculation and the list of assumptions; But you should give the base numbers from the official source, write each assumption clearly, and cross-check the top-down and bottom-up methods. The TAM/SAM/SOM distinction keeps expectations realistic; Range and sensitivity analysis prevents spurious precision. It is not the number that is defended, but the assumptions that lead to the number.
Application task
Choose a product or service market. Collect baseline data (such as population, household, official turnover) from real/representative sources. Calculate the market both top-down and bottom-up using the powerful prompt; Separate TAM/SAM/SOM. If the two methods differ, find out why (e.g. double counting). Perform a sensitivity analysis with the two most uncertain assumptions and produce a low-high range. Next to each assumption, write its source or “must be verified” label.
checklist
- [ ] I gave the base numbers from the official/real source, I did not adapt them to the model.
- [ ] I calculated and cross-checked the market using two independent methods.
- [ ] I gave TAM, SAM, SOM separately and defined.
- [ ] I wrote each assumption clearly and labeled it as source/verified.
- [ ] When two methods conflicted, I investigated the reason for the difference.
- [ ] I presented range and sensitivity analysis instead of single numbers.
- [ ] I planned expert/financial advisor approval for the critical account.