A sound financial research process turns a broad question into a reproducible, reviewable conclusion. It separates facts from assumptions, favors original evidence and preserves enough context for another person to retrace the work. That discipline matters whether the output is an investment memo, budget assumption, market study or client analysis.
Define the decision before searching
Write the decision, audience, date, jurisdiction, entity and time horizon. Then list what would change the conclusion. “Is the company attractive?” is too vague; “Can the company fund its next 18 months without new equity under an adverse revenue case?” produces testable questions.
Create a research log with the query, source, publication or filing date, period covered, access date and notes. Time-sensitive claims should carry an “as of” date.
Use a source hierarchy
| Need | Prefer | Verify carefully |
|---|---|---|
| Company financials | Regulatory filings, audited statements | Aggregators and copied tables |
| Law or tax rule | Statute, regulator or tax authority | Undated summaries |
| Market estimate | Transparent public dataset | Vendor forecasts without methods |
| Management claim | Filing plus operating evidence | Press release alone |
Secondary sources are useful for discovery and interpretation, but important figures should be traced to their origin. Archive the exact table or page used when permitted, because live pages and datasets change.
Normalize before comparing
Check currency, units, accounting period, fiscal year, inflation basis and whether values are nominal, real, annualized or trailing. Reconcile non-GAAP measures to the closest GAAP measure when available. Note changes in definitions and restatements.
For every calculation, retain the formula and inputs. For example, growth rate = (current value ÷ prior value) − 1. A percentage is not meaningful if the two values use different scopes. Our guide to unlevered free cash flow shows why consistent definitions matter in valuation.
Triangulate and challenge the thesis
Look for at least one independent source for a material external claim. More importantly, search for disconfirming evidence: customer concentration, covenant limits, litigation, accounting-policy changes, dilution, cyclicality or data revisions. Label facts, management estimates, analyst assumptions and your own inferences separately.
Review and communicate uncertainty
- Recalculate key figures independently.
- Check citations and dates against the underlying source.
- Test a base, upside and downside case.
- Have a reviewer challenge scope, definitions and omitted risks.
- State confidence, limitations and the next update trigger.
A concise conclusion should say what the evidence supports, what remains unknown and what action is justified—not bury uncertainty beneath false precision. See also our framework for strategic financial decisions.
Frequently asked questions
How many sources are enough?
There is no universal number. Source quality, independence and materiality matter more. A regulatory filing may establish a figure, while a forecast needs methods, assumptions and corroboration.
Can AI summarize financial research?
It can assist with organization, but outputs may omit context or invent facts. Verify every material claim and calculation against the original source, and do not upload confidential data without approval.
What makes research reproducible?
A defined question, dated sources, saved inputs, explicit transformations, formulas, assumptions, version history and a clear separation between evidence and judgment.
Sources reviewed
- SEC EDGAR company filings search
- U.S. GAO Cost Estimating and Assessment Guide
- Investor.gov: Researching investments
Last reviewed: August 15, 2026.