Overview

FMI Research Methodology: Grounded in Primary Data and Multi-source Triangulation

We combine direct industry inputs with disciplined validation across multiple sources. Since 2014, this has helped us build market intelligence that is dependable, consistent, and accurate.

250+

primary interviews per report

500+

non-market-research sources cross-checked

100%

independent QC before publication

Market Sizing Built through Triangulation

We size each market using three separate approaches and reconcile the results before publication. If the estimates differ beyond the stated threshold, we revisit the model instead of basing our results on averages.

1. Supply-side estimation

  • Full mapping of the relevant participant base across global, regional, and local tiers
  • Cross-checks on operating scale, capacity, or delivery footprint where relevant
  • Price ranges structured by offering type, customer segment, and geography
  • Smaller and private participants estimated through peer benchmarking
  • Revenue built from unit, volume, or contract-value estimates using weighted pricing assumptions

2. Demand-side estimation

  • End-market mapping specific to each product or service category
  • Buyer-side interviews and procurement checks to validate volume patterns
  • Application-level review of usage, replacement, or renewal cycles where relevant
  • Regional conversion assumptions documented by customer type, use case, or market context
  • Supply-demand reconciliation completed before any numbers are finalized

3. Top-down validation

  • Trade flow analysis using HS code import and export data
  • Benchmarking with parent market sizing and established industry benchmarks
  • Cross-checks against capital expenditure patterns and automation trends where relevant
  • Domestic sales reviewed against production, imports, exports, and inventory changes
  • Scenario testing across base case, optimistic case, and conservative case

B2B (All Industries)

What is collected

Primary interviews and surveys form the starting point for every dataset. On top of that, regulator and institutional records are added alongside trade series, production data, logistics records, and tender filings. Developer and app telemetry is included where it is relevant to the market being sized. Company filings and plant registries are part of the standard input set. We do not scrape e-commerce sources unless the data comes through a licensed feed.

How data is collected

Primary research is fielded in structured waves. Samples are stratified by region, company size, channel, and product family before fieldwork begins. Institutional datasets come in through public APIs or licensed feeds, and each one is logged with its date and license terms.

Ecosystem signals from B2B marketplaces, distributor portals, and tender boards are captured under applicable terms of service. Company masters and product masters are maintained throughout the process. Where records need to be linked, deterministic keys are applied first. Fuzzy matches that remain are reviewed and adjudicated by analysts.

When reports are updated

Prices, shipping patterns, and outage data are refreshed weekly. Trade flows, production activity, and category mixes follow a monthly cycle. Capacity figures, installed base data, and baseline metrics are reviewed each quarter. If a market shock occurs, an ad hoc update is triggered outside the standard schedule.

Domain lenses

Each sector gets a focused lens applied to its data. The full detail sits in the reports. The short version is in the table below.

Interactive: Price–Volume scenario

Reduces ASP by intensity × 0.25%

Scaled: −(value / 100). Higher = more sensitive

Value = 56,745,000
0 25% 50% 75% 100% Promo → Value Index
Domain Primary signals Institutional / ecosystem Cadence
Automotive & EV Production, registrations, Tier-1/2 shipments, charging sessions, battery MWh OICA; ACEA/JAMA/SIAM; IEA EV; NHTSA/KBA; UN Comtrade HS 84–87 Monthly–Quarterly; EV weekly when available
Chemicals & Materials Capacity, utilization, spreads, HS flows, turnarounds IEA/EIA; USGS; Eurostat/PRODCOM; UN Comtrade HS 25–40; CEFIC; IFA Weekly–Monthly–Quarterly
Healthcare Approvals, UDI, trials, procedure volumes, payer schedules FDA/EMA/EUDAMED/PMDA/CDSCO; ClinicalTrials.gov; WHO ICTRP; CMS DRG Weekly–Monthly–Quarterly
Technology & Semis Filings, deployment telemetry, app installs, fab/node EDGAR; SEMI; HS 8542; GSMA/3GPP; spectrum auctions Quarterly + monthly capex
Packaging Substrate flows, converter capacity, OEE, format adoption UN Comtrade HS 39–48, 76; FAOSTAT; PRODCOM/NAICS Monthly–Quarterly
Food & Beverage Farm-to-fork balance, price/mix, claim shares FAOSTAT; USDA WASDE/ERS; HS 02–24; FAO FishStat Monthly (+ weekly for volatile)
Oil & Gas Capacity, production, storage, trade, emissions IEA/EIA; OPEC; HS 27; Baker Hughes Weekly–Monthly
Industrial Automation Shipments, installed base, retrofit cycles, tenders PRODCOM/NAICS; EU TED; SAM.gov; HS 84–85 Monthly–Quarterly

Lineage & confidence (B2B)

Sources UN Comtrade; ITC Trade Map; Eurostat/PRODCOM; US Census; BEA; BLS; OECD; IEA/EIA; OPEC; USGS; FAOSTAT; USDA; FAO FishStat; FDA/EMA/EUDAMED/PMDA/CDSCO; ClinicalTrials.gov; WHO ICTRP; OICA; ACEA; SIAM; national statistics; tender boards.
Access Public / Licensed / Consent (logged with pull dates and licenses)
Transforms Unit conversions; currency basis; geo normalization; seasonality method
Confidence A/B/C by metric, with reasons
Caveats Re-exports and intra-company transfers are handled conservatively where final use is uncertain

B2C / Retail

What is collected

Public filings and trading updates form the base. Store-locator counts and opening or closure announcements are tracked alongside retailer circulars and newsletters. App store pages and reviews are included. Public social content from Instagram, TikTok, YouTube, and X is pulled in, as are search-interest signals. Supplier and distributor anecdotes are used as guardrails, not as primary inputs. No private or proprietary data is accessed.

Retail proxy sandbox

SRAP=49 | PII=21 | RHI=55

Weights are illustrative. Proxies are labeled with confidence bands.

t0t1t2 t3t4t5 t6t7t8 RHI Index

What is not done

At FMI, EPOS and panel datasets are not used. Shopping cart scraping is avoided. Private pages are never accessed, and PII is not collected. No surprises (or shocks) here.

Lineage & confidence (Retail)

Sources Data is drawn from retail filings and trading updates, store-locator pages, retailer circulars and newsletters, public social posts, app store public pages, and national retail indices.
How the data is processed Weekly data is rolled up to monthly before analysis. Engagement figures are normalized across platforms so they can be compared on the same scale. Duplicate records are filtered out. All currency figures are converted to constant USD
Confidence levels Metrics taken directly from company filings are rated Tier A. Estimates built from two or more independent signals are rated Tier B. A reading that rests on a single signal is rated Tier C and treated with caution
What to watch for Viral content can push short-term momentum figures higher than the underlying trend justifies. Some retail chains also post less promotional content publicly than others, which means their social footprint will look smaller than it is.

Mathematical Modelling

Interactive: quick calculators

Units × ASP

Value = $270,000,000

Apparent capacity

Apparent capacity ≈ 941,176

Equations

Value = Units × ASP InstalledBaseₜ = InstalledBaseₜ₋₁ + Shipmentsₜ − Retirementsₜ
OtherVendors = MeanRevenue(tier) × VendorCount(tier) ApparentCapacity ≈ Production / CapacityFactor

Lineage & confidence (Modelling)

Transforms Winsorized ASP; state-space elasticities; survival-curve retirements; constant-FX forecasts
Confidence Tier A for audited + multi-source; Tier B when partial; Tier C when proxy-heavy
Caveats Re-export uncertainty; EPOS gaps where proxies are used; vendor fiscal shifts around year-end
Lane Examples
Trade & Production UN Comtrade; ITC Trade Map; Eurostat/PRODCOM; US Census; BEA; BLS; OECD
Energy & Commodities IEA; EIA; OPEC; USGS; UNCTADstat
Agri & Food FAOSTAT; USDA WASDE/ERS; FAO FishStat
Healthcare FDA (Drugs@FDA, PMA/510k, GUDID); EMA; EUDAMED; PMDA; CDSCO; ClinicalTrials.gov; WHO ICTRP; CMS DRG
Automotive & Mobility OICA; ACEA; JAMA; SIAM; NHTSA; KBA; IEA Global EV
Technology & Telecom EDGAR; SEMI; GSMA; 3GPP; spectrum auctions; HS 8542
Procurement & Tenders EU TED; US SAM.gov; UK Find a Tender; India GeM
Retail Public Filings; store-locator pages; circulars/newsletters; app-store pages; public social posts

From Scope to Deliverable

From scope to deliverable, FMI follows a multi-step approach to ensure our findings are highly accurate. Depth is adjusted to the complexity of the market and the strength of the available evidence.

Step 1. Market definition and scope alignment

The market is defined before analysis begins, including product boundaries, geographic coverage, and the time period under review. This creates a clear base for the analysis that follows.

Step 2. Secondary research foundation

The research base is built from a wide range of public, proprietary, and company-level sources. This includes industry publications, trade data, company disclosures, and subscription databases.

Step 3. Primary research validation

Each report is supported by an interview program with industry experts and value-chain participants. The primary research is used to test assumptions and close information gaps.

Step 4. Bottom-up market modelling

The model is built company by company, with each revenue estimate linked to that participant's activity. The final figure emerges from the underlying estimates built across the market.

Step 5. Top-down reconciliation

The bottom-up estimate is reviewed against the wider market context, including trade flows and adjacent market indicators. If the gap falls outside the established tolerance range, the model is refined before the final figure is confirmed.

Step 6. Triangulation and quality checks

Key assumptions are tested against a minimum of two independent inputs. Where direct evidence is hard to get, estimates are kept conservative, with assumptions documented clearly as part of the model record.

Step 7. Scenario and sensitivity analysis

Three scenarios are built using clearly defined assumptions and documented market drivers. The assumption base is structured so clients can test how changes in key inputs affect the final view.

Step 8. Market share estimation

Market shares are estimated by comparing company-level activity with the overall market size. Each participant is assessed through a common framework, and shares are expressed as ranges where the evidence supports a range.

95.5%

accuracy rate (client feedback)

±5%

variance threshold, mature markets

3

scenario pathways in every forecast

What Every FMI Report Covers

The framework is applied across all industries FMI caters to. The scope and depth is adjusted to market complexity, while the underlying structure remains consistent.

Coverage Area What Is Delivered Validation Method
Market sizing Historical five-year and forecast ten-year data in value and volume Supply-side estimation, demand-side estimation, trade data
Segmentation Product type, end-use, technology, material, distribution channel, nature, demographics, region, and more Buyer interviews, OEM catalogs
Regional breakdown Country-level data for key markets; state-level where commercially relevant Trade flows, production data
Competitive landscape Twenty or more company profiles, market share ranges, SWOT analysis, production capabilities, global and regional footprint Relative benchmarking, pricing validation
Pricing analysis Price ranges by tier, region, and channel; ex-works and delivered; contract and spot pricing Distributor quotations, buyer procurement interviews
Demand drivers Policy-demand linkages, technology shift timelines, driver impact rankings Regulatory review, primary research
Value chain Bargaining power mapping and operating pain points across participants Primary interviews

The FMI Quality Assurance Commitment

Each market model is reviewed through four defined cross-checks before publication.

1. Supply-demand reconciliation

Buyer procurement volumes are compared with supply-side totals. Gaps are addressed through follow-up research or revision of utilization and pricing assumptions.

2. Trade flow cross-check

Import and export data is used to review domestic sales against local production, imports, exports, and inventory changes.

3. Installed capacity ceiling

The final estimate is tested against the output that the installed production base can support at stated utilization levels.

4. Parent market proportionality

Final totals are reviewed against capital expenditure trends, output growth in consuming industries, and historical demand relationships.

"The assumption tables and variance thresholds were what made the difference for us. We could show our finance team exactly where the numbers came from and how confident we should be in each one."

Strategy Director, Global Packaging Manufacturer, North America

Cadence & Deliverables

What is Included in Every Engagement

Standard depth across industries. The scope and depth is adjusted to market complexity, while the underlying structure remains consistent.

Pricing analysis

Average selling prices mapped across products, customer groups, channels, or regions, with clear assumptions on how pricing moves over time.

Cross-segment analysis

The model can be cut across multiple dimensions such as product, end use, channel, customer type, or geography to show where growth is coming from.

Historical and forecast market model

Fully populated historical and forecast data for 10 years, with no placeholders or undisclosed values.

Scenario-based forecasts

Optimistic, base case, and conservative paths, each supported by stated assumptions and documented market drivers.

Assumption tables

Conversion ratios, penetration rates, and utilization assumptions are listed with their stated basis.

Supply-demand view

Where relevant, the analysis brings together supply-side capacity and demand-side consumption to test whether the market structure is realistic.

Country-level segmentation

Regional analysis is carried down to country level for commercially significant markets.

White-space identification

FMI research methodology highlights underpenetrated segments and revenue pockets that may not be obvious in a top-line market view.

Competitive share estimates

Twenty or more company profiles mapped to the same framework used in the market model. Shares are expressed as ranges where appropriate.

Interactive dashboards

Outputs delivered through interactive dashboards that allow users to filter, compare, and review market cuts without rebuilding the model manually.

Editable Excel model

Provided in Excel format with linked tabs, transparent calculations, and structured tables that teams can use internally.

Executive summary pack

A presentation-ready summary with strategic takeaways tailored for leadership review.

FAQs

Who are the key opinion leaders interviewed for each report?

The mix varies by market. It typically covers manufacturers, distributors, buyers, and sector specialists. Participants are sourced through FMI's research panel and direct outreach. Names are kept confidential, but the split by role and region is disclosed in the report.

What happens if the three sizing approaches do not reconcile?

We do not average conflicting estimates. The variance threshold is stated upfront, and if any two approaches fall outside it, the assumption that is driving the gap gets investigated before the number is published.

Market share is shown as a range. Why not a single number?

A single number implies a precision the data rarely supports, especially for private companies. A range is more honest. Where the evidence is strong enough to narrow it, we narrow it. Where it is not, we say so.

How are smaller private companies estimated when there is no public data?

Through peer benchmarking. We compare operating footprint, capacity, and pricing against similar participants where data does exist, then apply weighted assumptions. The basis for each estimate is documented in the model.

The accuracy rate is 95.5%. How is that measured?

It comes from client feedback across purchased reports, not from internal validation. Clients who have used the data operationally and reported back on fit. It is a stated rate, not a modelled one.

Can I request a scenario that is not in the standard three?

Yes, through a custom engagement. The standard report includes base case, optimistic, and conservative paths. If your decision depends on a specific input moving differently, that can be modelled separately.

What does analyst access mean in practice?

It means direct contact with the research team after delivery. The analyst who built the model can walk you through the assumptions or answer follow-up questions on the data.