Results, Presentation and Concall: What If the Numbers Do Not Agree?
Reconcile quarterly filings, investor presentations and earnings calls using period, scope, currency and metric definitions, with dated Infosys examples.
When a quarterly result, investor presentation and earnings call appear to disagree, first check whether they describe the same period, reporting scope and metric. A difference that remains after those checks deserves investigation, not an average of the competing numbers.
The practical problem often starts with a perfectly accurate summary. An analyst copies “revenue growth” from a filing, another copies it from a slide and a third captures it from the call. Once the qualifiers disappear, three different measures look like one contradictory fact.
For a research team, the task is to preserve those qualifiers and build a reconciliation before updating the model. Agreement is useful only when the sources are actually comparable.
Give each source the right job
For a statutory historical line item, start with the applicable filed financial statements and notes. Check whether the figures are audited or reviewed, standalone or consolidated, and whether a later correction or restatement supersedes them.
An investor presentation can provide operating definitions that do not appear as separate statutory line items: volumes, subscribers, order intake, store counts or adjusted measures. Capture its definitions and calculation boundaries. A polished slide is not a reason to replace a different statutory measure.
The concall supplies explanations, qualifications and forward statements. It can help explain a cost increase or a change in guidance, but a spoken approximation should not silently overwrite a precisely reported historical amount. Check official transcripts against audio where a consequential passage is ambiguous.
| Research question | Starting source | What to retain |
|---|---|---|
| What was reported revenue? | Filed financial statement | Period, currency, scale, scope and notes |
| What happened to operating volumes? | Defined company disclosure or presentation | Unit, included businesses and comparable period |
| Why did margin change? | Statement notes, presentation and call | Reconciliation plus attributed explanation |
| What does management expect next? | Dated outlook disclosure and call | Horizon, range, definition and conditions |
This is a source-selection method, not a rule that every filing is flawless or every presentation is unreliable. The relevant document depends on the question.
Three growth rates can all be correct
Infosys’s disclosures for the quarter ended 31 March 2025 provide a useful historical example. Its release dated 17 April 2025 reported quarterly revenue of ₹40,925 crore and distinguished the following year-on-year rates:
| Q4 FY25 revenue growth measure | Reported rate |
|---|---|
| Reported rupee growth | 7.9% |
| Reported US-dollar growth | 3.6% |
| Constant-currency growth | 4.8% |
These figures appear in the company’s IFRS INR earnings release, page 1. They are dated historical disclosures, not a statement about the latest quarter.
The rates answer different questions. Reported rupee and dollar growth use their respective reporting currencies. Constant-currency growth adjusts for currency effects using the company’s methodology. It is not automatically organic growth: acquisition effects are a separate issue.
The official earnings-call transcript, page 5 explicitly identifies the 4.8% Q4 year-on-year measure as constant-currency growth. A summary that drops those words creates a false conflict with the rupee figure.
Do not average 7.9%, 3.6% and 4.8%. Store them as different metrics. Also do not describe their percentage-point gap as a complete FX bridge without the company’s currency methodology and underlying figures.
Same company and quarter can still mean different entities
In the same FY25 results package, Infosys reported Q4 revenue from operations of ₹40,925 crore on a consolidated basis and ₹34,136 crore on a standalone basis. Both are in ₹ crore for the quarter ended 31 March 2025. The difference is reporting scope. See the audited results package, consolidated statement on PDF page 12 and standalone statement on PDF page 18.
An automated extraction that finds the first revenue row in a PDF can pick the wrong scope even when it reads the digits correctly. Check the statement heading, not just the company name and date.
The subtraction, ₹40,925 − ₹34,136 = ₹6,789 crore, is an arithmetic gap. It is not necessarily a ready-made “subsidiary revenue” line. Consolidation can include intercompany eliminations and other adjustments. A model needs the relevant notes, not a renamed residual.
Keep the same basis across the comparison window. If you must change basis because coverage differs, explain the break instead of presenting a seamless growth series.
Check the period before explaining the business
Quarter, half-year, nine-month and annual columns often sit next to each other. An apparently large surprise may be a cumulative figure pasted into a quarterly field.
In a hypothetical example, Q1 revenue is ₹100 crore and first-half revenue is ₹230 crore on the same basis. Derived Q2 revenue is ₹230 − ₹100 = ₹130 crore. The ₹230 figure is not Q2 revenue.
Do not apply that subtraction mechanically to EPS or balance-sheet balances. EPS uses a weighted-average share denominator. Cash at a reporting date is a stock, not cumulative cash flow. See quarterly, year-to-date and annual reconciliation for the detailed boundaries.
A presentation may also compare an exit-month run rate with a quarterly average. Both can be informative, but they must retain different labels.
Reconcile adjusted measures without erasing statutory results
Suppose, in another hypothetical example, the filed operating profit is ₹18 crore while a presentation shows adjusted operating profit of ₹21 crore. The presentation identifies a ₹3 crore expense excluded from its measure.
The bridge is ₹18 + ₹3 = ₹21. That explains the difference. It does not establish that the exclusion is appropriate for your forecast or that the expense will never recur.
Retain the reported figure, the adjustment and the adjusted figure separately. Record the company’s rationale. Ask whether similarly described costs occurred in previous periods, whether the definition changed, and whether cash was actually paid. A non-cash charge can still have economic meaning.
Distinguish an expense added back before tax from an after-tax earnings adjustment. Do not apply the same amount to operating profit and PAT without a tax and attribution bridge. EBITDA can also exceed EBIT because depreciation and amortisation are excluded, without any inconsistency.
Keep guidance apart from actuals
A call that discusses an annual growth range is not reporting quarterly growth. A target for an eventual margin is not necessarily guidance for the current financial year. Record the horizon, whether the statement is a target or outlook, the exact metric and the conditions attached to it.
If a slide mentions an ambition and management qualifies it in the Q&A, keep both. Do not convert an aspiration into a hard model input by removing the condition. Our guide to reading a concall like an analyst covers the language comparison. The reconciliation task here is to ensure that the model uses the right kind of observation.
When the difference remains unresolved
After aligning period, entity, currency, units, accounting definition and document version, a material residual is a real research question.
Check for an updated exchange filing, a company correction, revised presentation or transcript clarification. Recalculate using unrounded data where available. A small rounding difference can be explainable, but establish a tolerance appropriate to the disclosure’s precision. Do not widen it until a material difference disappears.
Retain both source observations with their locators and publication dates. Record what you checked, what remains unexplained and the clarification needed. Use investor relations for a material unanswered question. A research flag is not evidence of fraud or an accounting failure.
If the unresolved value would materially affect the model, leave it unavailable or explicitly retain a dated earlier estimate with its limitation. Do not call that estimate a reported actual. An AI-generated midpoint is not a reconciliation.
Keep a short record that another analyst can reproduce
A useful reconciliation record contains the metric, raw value, units, covered period, reporting basis, source locator, publication date, proposed explanation and review status. The reviewer should be able to reopen the evidence and reconstruct the conclusion.
For example, “currency basis differs” is a completed explanation only after the relevant labels and methodology are checked. “Probably currency” is still an open question. A status field should reflect that distinction.
Once the inputs reconcile, use the quarterly-result-to-model-update workbook to preserve the old estimate, enter the completed quarter and explain the revised annual forecast. The download provides a compact evidence and assumption record, not a substitute for source review.
Where Altys helps
Altys’s source-linked research, guidance history and exportable analysis help teams bring company disclosures into a repeatable checking process. Keep reported facts, management explanations and analyst assumptions distinguishable, and use Excel to inspect the calculations outside the platform.
The goal is not to make every source look identical. It is to understand which differences are explained, which questions remain open and which inputs can responsibly enter the next model version. Request Altys access to explore that research and monitoring workflow.
Educational accounting and research methodology only. Altys Labs is not a SEBI-registered research analyst or investment adviser. These historical examples are not recommendations, forecasts, price targets or portfolio-allocation guidance.
Frequently asked questions
Why can results, presentations and concalls show different numbers?
They may cover different periods, entities, currencies or definitions. Reported rupee growth, constant-currency growth and adjusted profit answer different questions. Align the labels and verify the bridge before treating a difference as an error.
Which source should an analyst use when numbers conflict?
For a statutory historical line item, start with the applicable filed financial statement and notes, including later corrections. Use the presentation for defined operating metrics and the call for explanations and guidance. If a material like-for-like conflict remains unresolved, retain both observations and seek clarification.
Should an analyst average conflicting financial numbers?
No. Averaging incompatible observations does not reconcile them. Preserve the original values and source locators, identify the difference in scope or definition, and leave an unresolved model input unavailable rather than inventing a midpoint.