Syntha Lab RU

Market research: TAM/SAM/SOM, CustDev, UX tests and benchmarking

The document describes how to estimate market size and the company share, talk to customers, test a website and collect competitor prices so that the conclusions can be rechecked. It decides whether to enter a segment, which sales plan to set and in which price corridor to go to market.

Brand, customer, organisation Brand and retail Verified: 2026-11-14

Open interactively →

Why this document

The decision to enter a segment or raise a price is often made on a feeling, and a mistake costs a whole collection. The owner, the development director and the marketer read this document before launching a new direction and when setting the sales plan. It shows how to estimate market capacity bottom-up and top-down, how to test a hypothesis in interviews and how to collect prices so that the conclusion can be repeated. That means the plan figure rests on assumptions that are visible and checkable, not on the confidence of the author of the presentation.

At a glance

What is this document and why is it needed?
The document describes how to estimate market size and the company share, talk to customers, test a website and collect competitor prices so that the conclusions can be rechecked. It decides whether to enter a segment, which sales plan to set and in which price corridor to go to market.
Who owns the document and how often is it rebuilt?
Brand: strategist, product director or analyst, before launching a line, entering a new channel or category and once a year for the plan; the decision is with the commercial director and the owner. Retailer: development director and category analytics, assessment of locations and formats.
What mistakes are the most common?
TAM is counted as the whole national clothing market → a huge and useless number → limit by audience, region, price segment and channels (SAM). SOM is set by wish (“1% of the market”) → the plan is unrelated to capacity → compute SOM from stock, outlets and budget. Quantitative conclusions are drawn from CustDev → 10 interviews pass for statistics → interviews give hypotheses, test them by survey or data.

For whom

Brand
A brand needs market size by price segment and channel to set the plan and understand whom it displaces. The bottom-up method (buyers × frequency × basket) and top-down (market estimate, segment share) are reconciled and compared. CustDev gives hypotheses about motives, and a quantitative survey and sales data test them.
Retail
A retailer cares more about the capacity of a district and format: population, income, footfall, competitors in the zone, churn share. For a new outlet the market is counted by catchment area, and competitive benchmarking includes assortment, prices and opening hours. Price monitoring at a retailer is regular, by a list of comparable items.

What it consists of

Market size

  • Brand: SAM is limited by the price tier and the channels where the brand is really present (wholesale, D2C, universal marketplaces); SOM is tied to production capacity, MOQ and the number of partner points.
  • Retailer: the market is counted by the store catchment (travel time) and format; SAM is limited by the price tier and the range, SOM is tied to space, traffic and conversion.
  • TAM: all solvent demand for the category in the region; SAM: the part available by price segment and channels; SOM: the part attainable in the period given the company strength.
  • Brand: share is computed at consumer retail (from the estimate of partner sales), not by shipments: otherwise wholesale inflates or deflates share by the size of stock.
  • Retailer: share is computed in the catchment and category (the retailer share in city demand for the category); a total chain share says little, so state the zone and format.
  • Market share = Company revenue ÷ TAM (or ÷ SAM if the company works only in the segment); state what the share refers to.
  • Brand: bottom-up is built from shipments by channel (wholesale, D2C, universal marketplaces) and the number of active customers; wholesale shipments do not equal consumer sales, so they are adjusted for partner stock.
  • Retailer: bottom-up is built from receipts and traffic of stores and online: visitors × conversion × average basket; the check is made against revenue per square meter and share in the catchment.
  • A bottom-up method from your own data and a top-down method from estimates; a gap over 30% means the assumptions should be revised.

Customer research

  • Brand: interviews run with both end customers and wholesale buyers: why they take or do not take the brand, what stopped them, which terms get in the way; two scripts, two reports.
  • Retailer: interviews run with store shoppers and online customers; the script is about choosing the place: why they came here, what they compared with, what they looked for and did not find.
  • CustDev: 8-15 in-depth interviews per segment, a script on the buying situation (how they chose, what they compared with, what stopped them); the result is hypotheses.
  • Brand: the sample is taken from the D2C base and followers plus an external panel so the market is not judged by loyal customers alone; the sample bias is recorded.
  • Retailer: the sample is drawn from loyalty-card shoppers and store visitors, with lapsed customers separately; answers are linked to the receipt and the purchase category.
  • Quantitative survey: sample, quotas, questions, error; with a simple random sample of 400 the error of a share is about ±5 pp.
  • Brand: the brand site and partner cards are tested: whether the size chart and composition are clear, whether the brand difference is visible in the partner platform listing.
  • Retailer: the chain site and app are tested: find a size, pay, process a return in store, use the loyalty card; scenarios are the same for all brands on the shelf.
  • UX test: 5-8 users per scenario (find, choose a size, check out), recordings and a barrier list; then an A/B test on the changed version.

Competitive benchmarking and prices

  • Brand: the set consists of brands of the same segment and channel; own prices are compared with partners retail prices for the same brand, that is parity across channels.
  • Retailer: the set consists of chains and universal marketplaces with overlapping range; prices for the same brands and for the private label against analogs are compared.
  • Competitor sample: 5-8 direct and 2-3 indirect; selection criteria recorded.
  • Comparable items: type, material, segment; the non-promo price, the promo price separately; measurement date and source (site, store).
  • Brand: partner price monitoring serves parity control and the recommended price; a mandatory resale price is not imposed, a recommendation is recorded as a recommendation.
  • Retailer: competitor price monitoring runs on a list of key value items (KVI) and category roles; the reaction to a competitor price is decided by category role, not automatically.
  • Price collection: manually by list, through platform exports or parsing of public pages respecting site rules and the personal data law; storage: a table with date, store, SKU, price, promo columns.

How to read the result and which decisions follow

  • TAM of RUB 36,300 m narrows to SAM of RUB 7,115 m and SOM of RUB 177.9 m; company revenue of RUB 38 m is 0.53% of SAM. Decision: TAM is a frame, and the sales plan is built from SAM and SOM, otherwise a figure like a market of 36 billion decides nothing.
  • Brand: SOM depends on three assumptions, and the sensitivity table shows the range: at -20% on both SOM is RUB 113.8 m, at +20% RUB 256.1 m. Decision: plan on the lower part of the range and treat the upper part as a target to be confirmed by research.
  • Retailer: your SAM is limited by channels and cities of presence, not the whole market; a survey with n = 400 gives an error of ±4.9 pp, and ±5 pp needs 385 respondents. Decision: do not compare segments of fewer than several dozen answers.
  • Brand: our price of RUB 4,490 at a median of RUB 4,440 gives an index of 1.01; CustDev and UX tests explain what people pay above the median for, while the index only shows the position.
  • Retailer: benchmark competitors in the same price groups and channels as your shelf; a comparison with a brand that sells only on its own site gives a false index.

Formula derivation and sensitivity tables

  • Derivation of the sampling error. A share p is estimated from n independent answers, the variance of the estimate is p(1 - p)/n, the standard deviation √(p(1 - p)/n). A 95% interval gives e = 1.96 x √(p(1 - p)/n); the maximum at p = 0.5 is e = 0.98 / √n.
  • Derivation of the sample size: from e = z x √(p(1 - p)/n) it follows that n = z² x p(1 - p)/e². For e = 5 pp 385 are needed, for 3 pp 1068, for 10 pp 97 respondents (p = 0.5, code).
  • Derivation of market size: TAM as a product of independent assumptions gives equal relative sensitivity: a change of any factor by x% changes SOM by x%; so what matters is not the factor but how unconfirmed it is.
  • Numeric SOM sensitivity table, RUB m (rows segment share -20..+20%, columns reachable share -20..+20%): -20%: 113.8 / 128.1 / 142.3 / 156.5 / 170.8; -10%: 128.1 / 144.1 / 160.1 / 176.1 / 192.1; 0%: 142.3 / 160.1 / 177.9 / 195.7 / 213.4; +10%: 156.5 / 176.1 / 195.7 / 215.2 / 234.8; +20%: 170.8 / 192.1 / 213.4 / 234.8 / 256.1.
  • Brand: check the reachable share of SOM against channel capacity and production capacity: at SOM of RUB 177.9 m and an average ticket of RUB 4,100 about 43383 orders a year are needed.

Control checks

  • Product check: TAM of 36300 equals 4.2 m x 62% x 3.4 x RUB 4,100 (the code matches), SOM equals the central cell of the sensitivity table.
  • Brand: write all TAM, SAM, SOM assumptions together with the basis of estimate (own data, survey, expert opinion) and date; without it the figure cannot be reproduced.
  • Retailer: compute the sample size before the survey; a survey with n below the needed value gives an error that analysis cannot fix.
  • Brand: process interview and survey data under the personal data law; consent and anonymization are needed before analysis.
  • Retailer: p = 0.5 gives the maximum error and works as a conservative estimate; if the share is known, compute with it.

Where the data comes from

Data sourceFields
Own sales and customersRevenue, basket, frequency, repeat purchases, geography, channel, segment share.
Open statistics and market estimatesPopulation, income, household spend on clothing and footwear; use as assumptions and record the date and definition.
Interviews, surveys, testsRecordings, transcripts, questionnaires, scenario results; participant consents.
Competitor price tableDate, competitor, style, non-promo price, promo price, size availability.

How to build it step by step

  1. Step 1. State the decision the research must support (enter a segment, price, plan) and the threshold at which the decision changes.
  2. Step 2. Define the target audience and geography: number of people, share buying the category, frequency, average basket excluding VAT.
  3. Step 3. Compute TAM, SAM and SOM by the steps of the document “Target audience segmentation and segment size” (the “Segment size” calculation).
  4. Step 4. Reconcile with a top-down estimate (market and segment share); if the gap exceeds 30%, revise the assumptions.
  5. Step 5. Run CustDev: write the script, recruit respondents by segment, record and transcribe; extract recurring motives and barriers.
  6. Step 6. Test hypotheses by survey or sales data; set the sample size by the allowed error.
  7. Step 7. Run a UX test of key site and app scenarios, fix barriers, then run an A/B test.
  8. Step 8. Collect competitor prices for the list of comparable styles, compute the median and the company price index.
  9. Step 9. Put the conclusions in a “conclusion, evidence, reliability, decision” table; mark what is checked and what is not.
  10. Step 10. Set the review date: market size annually, prices monthly, interviews as needed.

Formulas

MeasureFormulasNote
Market size: where the formulas areThe formulas of TAM, SAM and SOM and the “top-down” and “bottom-up” paths are written once: in the document “Target audience segmentation and segment size” and in the “Segment size” calculationHere the market size serves the choice of a research method; still treat the multipliers as dated assumptions.
Market shareShare = Company revenue ÷ TAM × 100 (or ÷ SAM)Revenue and TAM in the same prices excluding VAT and for the same period.
Survey share errorError = 1.96 × √(p(1−p) ÷ n); maximum at p = 0.5Valid for a random sample; for quota and online panels the real error is higher.

Formulas in notation

TAM
TAM = Pop × s_buy × f × P
Pop is the audience, s_buy the share of buyers, f purchases per year, P the average ticket excl. VAT.
SAM and SOM
SAM = TAM × s_seg × s_ch, SOM = SAM × s_ach
s_seg is the price segment share, s_ch the share of available channels, s_ach the reachable share.
Company share
Share = Revenue / TAM
Also computed to SAM; the share of TAM is almost always small.
Sampling error
e = z × √(p(1 − p) / n)
z = 1.96 at 95% confidence, p is the sample share, n the sample size.
Required sample size
n = z² × p(1 − p) / e²
For p = 0.5 and e = 5 pp this gives about 385 respondents.
Index to the median
I = P_own / median(P_comp)
Comparison of comparable items on one date.

Worked example

Exercise: market size and company share (conditional assumptions)

LayerHow computedValue, RUB m
TAM4.2 m people × 62% × 3.4 purchases × RUB 4,10036,300
SAMTAM × 28% (price segment) × 70% (channels)7,115
SOMSAM × 2.5% (attainable share)177.9
Company revenuefrom accounts38

The company share is 0.1% of TAM and 0.5% of SAM: against SAM the share is more visible because the company works in its segment. A survey of 400 gives a share error of about ±4.9 pp. Price benchmarking: the median of six comparable styles is RUB 4440, the company price RUB 4490, index 1.01. All assumptions are conditional and should be replaced with your own data.

Figures

From TAM to SOM: what is cut at each step
Waterfall: TAM 36,300, Outside the price segment −26,136, Outside available channels −3,049, SAM 7,115, Unreachable share −6,937, SOM 178010,00020,00030,00040,000TAMTAMTAM: 36,30036,300Outside the price segmentOutside the pricesegmentOutside the price segment: −26,136−26,136Outside available channelsOutside availablechannelsOutside available channels: −3,049−3,049SAMSAMSAM: 7,1157,115Unreachable shareUnreachable shareUnreachable share: −6,937−6,937SOMSOMSOM: 178178RUB m
  • positive
  • negative
  • total
Data as a table
StepChangeLevel
TAM—36,300
Outside the price segment−26,13610,164
Outside available channels−3,0497,115
SAM—7,115
Unreachable share−6,937178
SOM—178

From TAM to SOM: what is cut at each step. The market in RUB m: first outside the price segment, then outside available channels, then the unreachable share.

Our price and competitor prices
Bar chart: Competitor 1 3,590, Competitor 2 3,990, Competitor 3 4,290, Competitor 4 4,590, Competitor 5 4,890, Competitor 6 5,29001,0002,0003,0004,0005,0006,000Competitor 1Competitor 1Price without promotion: 3,5903,590Competitor 2Competitor 2Price without promotion: 3,9903,990Competitor 3Competitor 3Price without promotion: 4,2904,290Competitor 4Competitor 4Price without promotion: 4,5904,590Competitor 5Competitor 5Price without promotion: 4,8904,890Competitor 6Competitor 6Price without promotion: 5,2905,290Our priceOur pricePrice without promotion: 4,4904,490RUB
Data as a table
MeasurePrice without promotion
Competitor 13,590
Competitor 23,990
Competitor 34,290
Competitor 44,590
Competitor 54,890
Competitor 65,290
Our price4,490

Our price and competitor prices. Comparable items: our price against six reference ones; the index to the median shows the position.

Survey sampling error and sample size
Line chart: Error, ± pp: 13.9 → 2.5051015501002004008001600our surveyError, ± pp, 50: 13.9Error, ± pp, 100: 9.8Error, ± pp, 200: 6.9Error, ± pp, 400: 4.9Error, ± pp, 800: 3.5Error, ± pp, 1600: 2.5%
Data as a table
MeasureError, ± pp
5013.9
1009.8
2006.9
4004.9
8003.5
16002.5

Survey sampling error and sample size. The margin of error of a share at p = 0.5 and 95% confidence: falls four times when the sample grows 16 times.

SOM as segment share and reachable share change
Heat map: 5×5; min 114, max 256Reachable -20%Reachab…-20%Reachable -10%Reachab…-10%Reachable 0%Reachab…0%Reachable +10%Reachab…+10%Reachable +20%Reachab…+20%Segment -20%Segment -20%114Segment -20% / Reachable -20%: 114128Segment -20% / Reachable -10%: 128142Segment -20% / Reachable 0%: 142157Segment -20% / Reachable +10%: 157171Segment -20% / Reachable +20%: 171Segment -10%Segment -10%128Segment -10% / Reachable -20%: 128144Segment -10% / Reachable -10%: 144160Segment -10% / Reachable 0%: 160176Segment -10% / Reachable +10%: 176192Segment -10% / Reachable +20%: 192Segment 0%Segment 0%142Segment 0% / Reachable -20%: 142160Segment 0% / Reachable -10%: 160178Segment 0% / Reachable 0%: 178196Segment 0% / Reachable +10%: 196213Segment 0% / Reachable +20%: 213Segment +10%Segment +10%157Segment +10% / Reachable -20%: 157176Segment +10% / Reachable -10%: 176196Segment +10% / Reachable 0%: 196215Segment +10% / Reachable +10%: 215235Segment +10% / Reachable +20%: 235Segment +20%Segment +20%171Segment +20% / Reachable -20%: 171192Segment +20% / Reachable -10%: 192213Segment +20% / Reachable 0%: 213235Segment +20% / Reachable +10%: 235256Segment +20% / Reachable +20%: 256
  • 114 … 256 RUB m
Data as a table
Reachable -20%Reachable -10%Reachable 0%Reachable +10%Reachable +20%
Segment -20%114128142157171
Segment -10%128144160176192
Segment 0%142160178196213
Segment +10%157176196215235
Segment +20%171192213235256

SOM as segment share and reachable share change. SOM in RUB m when two assumptions shift by ±10% and ±20%: rows segment share, columns reachable share.

Specifics by category

Apparel
TAM is computed by item types and price tier, not “for apparel in general”: a dress, outerwear and basic knitwear have different purchase frequency and basket. Seasonality (two to four collections) requires computing the market per season and per year without adding them. Benchmark comparability is strict: T-shirt with T-shirt, same fabric and tier; fashion has no direct analogs, so comparison is by type and price and conclusions are marked as less reliable.
Footwear
TAM is counted in pairs: population × share buying × pairs per year (assumption 2-3, set your own) × price per pair; segmented by purpose (casual, sport, winter, business) and AW/SS season. CustDev asks about width, comfort and the experience of “the size did not fit”; the UX test checks the size hint and the EU, UK, US, Mondopoint conversion. Price benchmarking is by type, upper material and sole: leather and synthetics are not comparable, and winter pairs cost more than summer ones.
Accessories
Subcategories (bags, belts, jewelry, headwear, scarves) differ in frequency, occasion and basket, so TAM is summed from subcategories rather than taken as one share. A large part of demand is gift and seasonal (holidays), so the survey must ask occasion and budget. Precious metals and costume jewelry are separate markets with different regulation (assay supervision) and are not summed into one TAM. Benchmarking compares material and hardware, not only price.
Kids
TAM is built from the number of children by age groups (under TR CU 007/2011 “children” are under 14, “teens” 14-18) × purchases per year × basket; birth-rate dynamics changes the base, taken from official statistics (not verified). The respondent is a parent: parents are surveyed, and interviews with children only with the consent of a legal representative. Size is set by height (6 cm step), so purchase frequency is tied to growth, not only age. The market splits by peaks: school, New Year.
Men
Men buy less often and repeat models, so in CustDev loyalty, reasons for switching brand and expected service life matter more. Benchmarking is more reliable: basics (shirt, jeans, jacket) have exact analogs, the price index is steadier and depends less on season. The size curve is narrower, so the size error in TAM is smaller, but fit matters more in UX and the survey. TAM for men is more often counted through the replacement frequency of basics.
Women
More occasions, silhouettes and subcategories, and purchases are more frequent, so TAM is segmented by occasion (work, evening, casual) and price tier. Benchmarking is harder: fashion has no exact analogs, so comparison is by type, fabric and tier and reliability is marked. Surveys and UX tests must account for vanity sizing: the customer picks a size not only by the chart, so the test checks whether they understand garment measurements. The sample is split by age because occasions and budgets differ.

Common mistakes

  • TAM is counted as the whole national clothing market → a huge and useless number → limit by audience, region, price segment and channels (SAM).
  • SOM is set by wish (“1% of the market”) → the plan is unrelated to capacity → compute SOM from stock, outlets and budget.
  • Quantitative conclusions are drawn from CustDev → 10 interviews pass for statistics → interviews give hypotheses, test them by survey or data.
  • Prices compare models of different quality and promo prices → a false index → comparable models, non-promo price, a fixed date.
  • Parsing is launched ignoring platform rules and personal data → a block or a claim → collect only public prices, do not collect user data.

Russian specifics

  • If surveys and interviews collect answers that identify a person, the personal data law applies: consent, purpose, retention; use anonymization for analytics. Rule: Law 152-FZ of 27.07.2006: Federal Law of 27.07.2006 No. 152-FZ on personal data.
  • Price agreements with competitors and exchange of sensitive price data are not allowed under the competition law; collect public prices on your own rather than exchanging with a competitor. Rule: Law 135-FZ of 26.07.2006: Federal Law of 26.07.2006 No. 135-FZ on protection of competition.
  • Collecting data from sites (parsing) is governed by platform terms and intellectual property law on databases; there is no separate norm in the register, so get legal advice before industrial parsing. Rule: Civil Code part 4: Civil Code, part four (Arts. 1232, 1235, 1237, 1486, 1491, 1515) (rule check not completed).

Open interactively →