Start with a question you can actually answer
You read that a company is about to transform its industry. The chart looks encouraging, the headline sounds convincing, and the comments are full of people who have already bought shares. Before opening your brokerage account, ask a more useful question than “How far could this go?”: what do I actually know, and where did that information come from?
You do not need to turn every idea into a research dissertation. You do need to separate documented facts from somebody’s expectations. Good information can help you understand an investment that later loses money. Bad information can lead to a lucky outcome. The result of one trade does not tell you whether your process made sense.
Write down the claim you want to investigate. “This is a great company” gives you little to work with. “Revenue from its main business has grown, and I want to understand why” gives you something concrete to research. A specific question makes it easier to avoid opening ten tabs that all repeat the same story.
Find the document behind the headline
A news article can be a useful starting point. Next, look for the material it draws on: company results, annual accounts, a regulatory filing or the terms of the investment product. For US public companies, EDGAR provides access to filings submitted to the SEC. A company’s investor relations website can also help you locate the relevant documents.
A primary source is not automatically impartial. Management chooses what to emphasise in a presentation. Compare the headline with the accounts, accompanying notes and disclosed risks. The fact that information appears in an official filing does not mean the regulator recommends buying the company.
Try a simple check: can you open the original document and find the number yourself? If five websites quote the same commentator, you do not have five independent confirmations. You have one claim repeated five times. Save the link to the underlying source, not just the article that led you there.
The date matters as much as the number
Keep three dates separate: when something happened, the period a figure covers, and when the information was published. An article updated today may discuss last quarter’s results. A screenshot shared this morning may show a price from months ago.
Record the unit as well: dollars or euros, thousands or millions, a total amount or a per-share figure. Check whether a percentage compares with the previous quarter or the same quarter a year earlier. Those comparisons answer different questions. Do not mix an annual figure with a quarterly one without explaining the conversion.
Here is a hypothetical example. A company’s revenue rises from 100 to 120, a 20% increase. You still do not know whether it acquired another business, benefited from currency movements or saw its margins deteriorate. The number can be correct while your interpretation remains incomplete. Your next task is to find the explanation rather than supply one yourself.
Separate facts, interpretations and forecasts
All three can appear in a single sentence: “The company increased sales, its strategy is working, and it will dominate the market.” The first part may be verifiable in the accounts. The second needs an argument. The third concerns an uncertain future.
Split what you have read into three lines. Fact: what happened and where it is documented. Interpretation: the explanation the author proposes. Forecast: what would have to happen for the scenario to play out. Then add a fourth line: what evidence would weaken that explanation?
That last question is especially useful when you like the idea. Looking only for supporting arguments can make research feel stronger than it is. Spend some of your review looking for risks and alternative explanations. The aim is to understand the uncertainty you would be accepting, rather than to reject every possible investment.
Take this question further: How do I read a company press release without absorbing its spin? Then read How to review an investment idea in 20 minutes without rushing a decision.
Ask who is speaking and what they might gain
Being a persuasive communicator does not establish expertise or remove conflicts of interest. Check whether the author holds the asset, receives payment to promote it, or earns money when you open an account through a link. An incentive does not automatically make the content false. It is still relevant context.
The CNMV’s guide to financial content on social media explores risks associated with following finfluencers. A practical lesson for your own process is not to substitute popularity for checking. Followers, comments and confident delivery do not establish that an idea suits your circumstances.
Be particularly cautious when someone pressures you to decide before reading the terms, promises outcomes without explaining risks, or shows only winning trades. If a firm is offering financial services, consult the relevant regulator’s register and warnings. Check the website address and contact details carefully: a similar name may belong to a different organisation.
Use a six-point research note
Before spending more time on an idea, complete this short checklist. A plain note is enough; you do not need to pay for a new tool. The purpose is to make the gaps visible, not to collect documents for their own sake.
- Claim: what am I trying to establish, without promotional adjectives?
- Source: which original document supports it, and where can I find it again?
- Date and unit: what period, currency and comparison does the figure use?
- Explanation: what might explain the observation, and what alternatives exist?
- Incentives: what connection does the author have to the company, product or service?
- Open question: what still needs checking before I consider a decision?
If you cannot complete a line, say so. “I do not know yet” is a useful research outcome. Filling a gap with an assumption and later treating it as a fact is what makes a process unreliable.
Test the exact claim, not a nearby claim
A useful verification exercise starts by deciding what would count as an answer. Suppose a hypothetical article says a manufacturer is becoming more profitable because its latest product is popular. That sentence contains three separate questions: whether profitability increased, whether demand for that product increased, and whether the second change explains the first. Evidence for one does not settle the others.
Imagine reported revenue increases from 100 to 120 while operating profit increases from 10 to 11. Profit is higher in absolute terms, but operating margin falls from 10% to about 9.2%. Calling this improved profitability without naming the measure leaves readers with an ambiguous conclusion. Neither number tells you which product drove the change.
Write a verification target before opening another document: I need comparable product revenue and an explanation of the operating profit movement. If the company discloses only group totals, narrow the conclusion to those totals. Avoid using customer reviews as a substitute for undisclosed product margins.
This protects against answering an easier question because its data happen to be available. You can confirm that customers like a product and still leave its contribution to company earnings unresolved. Useful research makes that boundary visible instead of hiding it behind a broad positive adjective.
Reconcile a plausible headline with an alternative explanation
Consider a second hypothetical business that reports revenue of 150, compared with 120 previously. A commentator calls the 25% increase evidence of stronger demand. Suppose a supplementary table attributes 18 of the increase to an acquired business and 6 to currency translation, using compatible definitions and an additive bridge for this example.
Those specified contributions total 24. Subtracting them from the increase of 30 leaves 6, equivalent to 5% of the original 120. That calculation explains why reported growth and a narrower comparable movement can look different. It does not establish a universally applicable organic growth formula, nor does it prove that all remaining growth reflects additional units sold.
The remaining 6 could combine price changes, volume changes and product mix. Your research note should therefore distinguish the completed calculation from the unanswered economic question. Write that the provided bridge leaves 6 unexplained by acquisitions and currency, then request the next useful breakdown.
If the contributions overlap or use different bases, stop before subtracting them. A company might calculate currency effects after adjusting for acquisitions, while another figure uses reported totals. Numbers that look additive need not be additive. A visible reconciliation gap is preferable to a tidy but unsupported growth estimate.
Decide which uncertainty deserves the next hour
Not every missing detail warrants the same effort. Rank uncertainties by whether a reasonable answer could change your interpretation of the central claim. This is an editorial prioritization rule, not a numerical measure of investment risk.
For a hypothetical subscription business, a missing font attribution on a presentation is unlikely to explain whether customers renew. An unclear definition of paying customer could change the entire revenue argument. Put the second issue first, even if resolving it requires reading a dense note rather than another accessible article.
Use three columns: unresolved issue, competing explanations, and consequence for the claim. For customer growth, the explanations might be new paying accounts, free registrations or acquired accounts. The consequence is whether the published total supports recurring commercial demand. This structure makes the missing definition actionable.
A practical stopping threshold is reached when the remaining uncertainty prevents you from distinguishing explanations that matter. At that point, mark the claim pending. Do not compensate with more confident prose or more secondary links.
Conversely, an incomplete minor detail need not invalidate every verified statement. You might establish the reported customer total and reporting date while withholding a conclusion about retention. Trust is usefully assigned to individual claims, not granted or withdrawn from an entire document at once.
Handle a correction without rewriting your research history
Suppose you save a hypothetical presentation showing 40,000 active accounts. Two days later, the issuer replaces it with a version showing 34,000 and an explanation that duplicate accounts were removed. Keep both versions in your research record and identify which one now supports the current statement.
The decrease between versions is 6,000 accounts, or 15% of the original figure. That is a correction to the published count in this example, not evidence that 6,000 customers left during those two days. Calling it customer churn would invent a business event from a document revision.
Your note needs the original retrieval date, the replacement date if available, the explanation offered, and the conclusions affected. If your revenue model used account counts, revisit that dependency. If it relied on separately reported revenue, the effect may be different. Follow the actual chain of reasoning rather than automatically discarding everything.
If a figure changes without explanation, label the difference unresolved. Ask whether scope, units or definitions changed before assuming misconduct. Retaining a short correction history allows another reader to understand why your conclusion moved. It also prevents a later review from judging your earlier work against information that was not available when you wrote it.
Build a worksheet another reader could reproduce
A reusable worksheet should contain enough detail to reconstruct the key claim without becoming a document archive. Start with the exact sentence being checked. Underneath, record its minimum required evidence: the necessary metric, period, comparison basis and scope.
Next, create an input row for each number. Give the original value, unit, document location and any transformation you applied. In a hypothetical margin calculation, that might be operating profit of 11 divided by revenue of 120, with both figures covering the same period. Preserve the inputs as well as the resulting percentage.
Add a separate assumption row. If you assume a price remains unchanged for a scenario, write that explicitly rather than allowing the assumption to resemble company guidance. Then state one alternative interpretation and what evidence would distinguish it.
Finish with a short supported statement and a narrower unresolved question. For example: reported operating profit increased, but the margin decreased; product level economics remain undisclosed. This is more reusable than a personal confidence rating because someone else can challenge a specific input or inference.
On the next review, update only the changed rows and explain why they changed. That keeps the worksheet useful across reporting periods without turning old assumptions into permanent background facts.
What if the sources disagree?
Disagreement is a reason to inspect definitions before choosing a favorite source. A hypothetical results table might show revenue of 200 while a presentation shows 215. Neither figure can be adopted safely until you know whether they cover the same business, period and accounting basis.
Check whether one includes discontinued operations, presents a scenario or combines periods differently. If the difference is explained, keep both figures with their labels. If it remains unexplained, do not average them. An average would create a third number with no reporting basis at all.
What if the better documented source is older? Use it for the historical question it answers, while identifying the later claim as unresolved. What if a credible author provides no underlying document? Their reputation can justify looking further, but it cannot supply the missing definition or calculation.
Finally, distinguish a false claim from an unverified one. You can reject the statement that revenue of 120 represents 30% growth from 100 because the arithmetic contradicts it. You cannot reject an undisclosed customer forecast solely because you cannot check it. You can decline to rely on it. That difference keeps skepticism precise and avoids replacing promotional certainty with equally unsupported negative certainty.
Know when to put an idea aside
You do not need an opinion on every company. If you cannot explain how it makes money, find the original document or reconcile the numbers, you can leave the idea pending. Your attention is limited too. Spending it on a story you cannot check leaves less time for other research.
A useful habit is to finish each review with one concrete next step: read a note in the accounts, check a claim, or return after the next official release. “Keep watching the price” does not, by itself, answer a question about the business.
Momentu is designed to help organise where to look. The checking comes afterwards. The public research library explains the approach. The radar belongs to the paid membership; a score is not an instruction to buy. To turn this checklist into a routine, continue with our guide to reviewing an investment idea in twenty minutes.
Sources and editorial approach
Sources consulted on 2026-09-19. Examples and checklists are Momentu’s editorial frameworks, not validated strategies for generating returns.
General education, not personalised investment advice. Investing involves risk, including loss of capital. Read our editorial standards.