Where sources disagree

Same attribute (net_income) for the same entity (KeyCorp) at the same observation time (2026-07-21) has two different values: 387 USD and 472 USD. No period differentiation or other contextual factors are present to explain the discrepancy.

Both facts report diluted EPS for KeyCorp on the same date (2026-07-21) with identical metadata (no period specified, same observation timestamp), but claim different values: 0.35 USD_diluted vs 0.44 USD_diluted. This is a direct value conflict on the same metric.

Fact A reports 74 USD while Fact B reports 70 billion USD for assets_under_management—a difference of approximately 946 million times. Although Fact B is from an earlier period (January 2026) compared to Fact A (July 2026), the magnitude of discrepancy is implausible as a normal business change. For KeyCorp, a major financial services company, 70 billion USD is realistic and consistent with known scale; 74 USD is not credible for this entity. This indicates a data quality issue—likely a unit conversion error, decimal point error, or misidentified entity.

Same entity (KeyCorp) reports two different net_income values (486 USD vs 387 USD) at identical timestamp (2026-07-21 00:00:00) with no period qualifier to explain the difference. This indicates either a data quality issue or conflicting sources for the same attribute.

Both facts report net_income for KeyCorp at the identical timestamp (2026-07-21 00:00:00), but with conflicting values: 486 USD vs 472 USD. The 14 USD difference (~2.9% variance) cannot be reconciled without additional context. This indicates either a data entry error, calculation discrepancy, or conflicting sources reporting different values for the same metric at the same point in time.

Cramer praised Dollar General as 'terrific and remains terrific' in Feb 2025, but by July 2026 expresses skepticism about an acquisition by Fairbank (likely DG CEO Tom Fairbank), claiming it requires explanation and rationalization. This suggests diminished confidence in the company's decision-making, creating tension with the earlier blanket endorsement. The contradiction is a value conflict rather than a direct logical contradiction—Cramer can praise a company while criticizing specific decisions, but the language shift from unqualified praise to requiring rationalization indicates a meaningful change in assessment.

Same entity (Brent crude) and attribute (commodity_price) have conflicting values: 88.45 vs 104 USD_per_barrel. FACT A is timestamped to 2026-07-20, but FACT B lacks observation timestamp (None), making it impossible to determine if they represent the same time period. If from the same date, this is a clear data quality issue. If from different periods, the missing timestamp on FACT B prevents proper reconciliation.

FACT A is explicitly timestamped to 2026-07-01, showing Brent crude at 71.57 USD/barrel. FACT B provides the same attribute with a significantly different value (104 USD/barrel) but has no observed timestamp. The 45% price difference is plausible for different time periods (crude prices fluctuate daily), but the missing timestamp on FACT B creates ambiguity: if both values claim to represent the same period, this is a direct contradiction; if they represent different periods, the timing metadata is incomplete. The contradiction is primarily one of data quality/missing context rather than an absolute logical impossibility.

The same entity (Regeneron Pharmaceuticals) reports two different stock prices for the identical timestamp (2026-05-18 00:00:00): 629.68 USD vs 698.25 USD. This represents a 10.9% discrepancy ($68.57 difference). Both facts claim to be observations of the same attribute at the same moment, which is logically impossible—a stock can only have one price at a given instant. This indicates a data quality issue, likely stemming from different data sources, incorrect timestamps, or a data ingestion error.

Both facts reference Netflix's margin attribute observed on the same date (2026-07-16), but report different values: 31.5% vs 33.4%. Since the observation date, period, and entity are identical, only one value can be correct. The 1.9 percentage point difference is material and cannot be reconciled as rounding or measurement variance.

Both facts report a margin value for Netflix, Inc. on 2026-07-16, but with conflicting values: 34.1% vs 31.5%. The same attribute cannot have two different values for the same entity at the same timestamp.

Both facts reference the same entity (Netflix, Inc.), same attribute (margin), and same observation timestamp (2026-07-16 00:00:00), but report conflicting values: 34.1% vs 33.4%. These represent a 0.7 percentage point difference for what should be a single margin value at a specific point in time.

Netflix EPS is stated as both 3.60 USD (FACT A) and 0.56 USD (FACT B). These represent a ~6.4x difference in a key financial metric. While FACT B has a specific observation date (2025-12-31, year-end 2025), FACT A lacks temporal context (no period, no observed date), making it ambiguous whether they represent different reporting periods. However, if both claims purport to represent current EPS values for the same entity, they directly contradict. The 85% difference magnitude suggests data quality issues, source misalignment, or missing period metadata that would explain the discrepancy.

Two significantly different EPS values for Netflix: 3.60 USD (no observation date) vs 0.56 USD (observed 2026-01-20). While both lack explicit period information, the 6.4x difference indicates either: (1) they represent different reporting periods (quarterly vs annual, or different fiscal quarters), (2) one value is stale/outdated, or (3) a data quality issue. The presence of an observation timestamp on Fact B but not Fact A suggests Fact B is more recent and likely more reliable.

The same attribute (EPS) has significantly different values for the same entity: 3.60 USD vs 0.55 USD. This represents a ~6.5x difference with no clear temporal or definitional distinction provided. Fact A lacks an observation date, while Fact B has a specific date (2026-02-11), suggesting either different reporting periods (which would require explicit period metadata to resolve) or conflicting data from different sources. The magnitude of the difference cannot be attributed to normal period-to-period variation.

Two different EPS values reported for Netflix (3.60 USD vs 3.85 USD). While Fact B includes an observation timestamp (2026-05-26), Fact A lacks observation metadata, making it unclear if these represent the same reporting period. If both claim the same time period, this is a direct value conflict. If from different periods, this is expected variation but still represents conflicting claims without proper period disambiguation.

Same attribute (EPS) for the same entity has two different values (3.60 vs 2.52 USD). While EPS naturally varies across reporting periods, both facts lack explicit period information. Fact B has a specific observation date (2026-02-11) but Fact A has none, suggesting they may represent the same point in time or lack proper temporal context. This indicates either: (1) data from different quarters/years without clear period labels, or (2) duplicate/conflicting data sources reporting different values for the same period.

Two different EPS values (3.60 USD vs 0.79 USD) are claimed for Netflix without explicit period information. FACT A lacks an observation date, making it unclear how current it is, while FACT B provides a recent observation date (2026-05-26). If these represent the same metric for the same fiscal period, this is a clear contradiction. The large discrepancy (78% difference) suggests either: (1) they measure different periods (e.g., annual vs quarterly, different fiscal years), (2) they use different calculation methods (basic vs diluted EPS), or (3) one value is stale/outdated.

Both facts claim to represent the same attribute (stock_price) for Capital One, but use incompatible formats. Fact A expresses the value as '20 percent' (a percentage without units or context), while Fact B expresses it as '174 USD' (an absolute price). A stock price cannot be both a percentage and an absolute value. Additionally, Fact A lacks a timestamp (Observed: None), preventing temporal reconciliation. This suggests either a data quality issue or that Fact A represents a different concept (e.g., price change) but was mislabeled.

Stock price cannot be simultaneously represented as an absolute value in USD (259) and a relative percentage (20%). These are incompatible units for the same attribute. Fact A expresses stock_price as an absolute price; Fact B appears to express it as a percentage, which is either a percentage change, a percentage of something else, or malformed data. Additionally, Fact B lacks an observation timestamp, making temporal comparison impossible.

FACT A describes hotel development pipeline growth and FeePAR-accretive hotel additions—characteristics of a hospitality/real estate company. FACT B describes a technology deal with SAP regarding AI and enterprise data capabilities. These facts appear to describe entirely different business domains and are unlikely to coherently characterize the same entity. At least one fact appears to be misattributed or confused with a different entity. 'Simply Wall St Community' is a financial analysis platform, not a hotel operator or direct SAP partner, suggesting both facts may be misapplied.

FACT A discusses a company with real estate/hospitality development pipeline, hotel contract signings, and net margins—characteristics of a hospitality/real estate enterprise. FACT B specifically references 'Dana' with a stock valuation ($38.29 fair value vs. $32.55 current price). Dana Inc. is a well-known automotive parts supplier, not a hospitality company. FACT A contains no mention of Dana and describes business operations incompatible with auto parts manufacturing. The facts appear to concern different entities, making it impossible to assess contradiction at the claim level.

Fact A claims strong earnings potential with higher net margins and pipeline growth, while Fact B projects a PE ratio of 973.0x. This is logically inconsistent: companies with robust margin growth and earnings potential should have lower (normalized) PE ratios. A 973x PE ratio indicates the market is pricing in either severely depressed current earnings or extraordinary future growth expectations—contradicting the narrative of already-strong margins. Additionally, Fact A discusses hotel development pipelines while Fact B specifically references 'Summit' revenues, raising questions about whether both facts actually concern the same entity.

Fact A describes a hotel/real estate development business (contract signings, pipeline growth, FeePAR-accretive hotels), while Fact B describes a telecommunications business (mobile revenues, fixed line operations, FX headwinds on overseas operations). These are fundamentally different business models and cannot both accurately describe the same entity's core operations. This appears to be a misattribution—the facts reference different industries and operational metrics.

Fact B explicitly references 'NTT's investment in proprietary technologies' and positions NTT to benefit from 5G/6G, clearly attributing the forecast to NTT (Nippon Telegraph and Telephone). This contradicts the stated entity 'Simply Wall St Community'. Additionally, Fact A discusses a hospitality/real estate company's development pipeline and hotels, which doesn't align with either Simply Wall St Community (a research platform) or NTT (a telecommunications company). These appear to be forecasts about different entities that have been mislabeled as the same entity.

Fact B explicitly references 'NTT shares' with Japanese yen pricing (¥159.0), indicating this analysis is about a specific Japanese company (likely Nippon Telegraph & Telephone or similar). Fact A discusses hotel pipeline growth and development, which does not align with NTT's core business. These facts appear to be about different entities, suggesting either a misattribution of the entity label 'Simply Wall St Community' or a data integrity issue where Fact B was incorrectly associated with this analysis.

FACT A describes a company with hotel development pipeline, contract signings, and FeePAR-accretive metrics (characteristics of hospitality/real estate), while FACT B explicitly references Simpson Manufacturing—a building products/fastener company with no hotel development operations. These represent fundamentally incompatible business models and cannot both describe the same entity.

FACT A reports user_count as '7 USD' (a currency amount), while FACT B reports user_count as '7000000 users' (a numeric count). These values are incompatible both numerically (7 vs 7,000,000) and by unit (currency vs count). FACT A appears to contain miscategorized or corrupted data—a monetary value incorrectly labeled as user_count.

FACT A assigns a currency value ('7 USD') to the user_count attribute, while FACT B assigns a user count ('7000000 users'). These are fundamentally incompatible units for the same attribute. FACT A appears to contain a data entry or attribute assignment error — it has the wrong unit type entirely.

Both facts measure the same attribute (user_count) for the same entity but report vastly different values. Fact A reports '7 USD' which is nonsensical as a user count (appears to be a price/cost metric). Fact B reports '7000000 users' which is a valid user count. These cannot both be true for the same metric.

Fact A claims user_count is '7 USD' (currency units, nonsensical for user count), while Fact B claims user_count is '7000000 users' (numeric users). Both reference the same attribute but with incompatible units and vastly different magnitudes. Fact A appears to be a data quality error where a monetary value was assigned to a user count attribute.

FACT A reports user_count as '7 USD' (malformed: currency unit instead of count), while FACT B reports '7000000 individual_investors'. These values differ by six orders of magnitude and use incompatible units. FACT A appears to be corrupted or misattributed data—user_count should not be denominated in currency. FACT B is a properly formatted count with an observation timestamp. If both refer to the same entity and timeframe, they cannot both be accurate measurements of user_count.

Both facts claim to represent 'user_count' for the same entity, but assign incompatible values: FACT A claims 7 USD (with a currency unit that doesn't align with user counts), while FACT B claims 7,000,000 users. The numerical values differ by three orders of magnitude (7 vs 7,000,000). FACT A's USD unit suggests potential data corruption, miscategorization, or mislabeling of the attribute—it may actually represent price, subscription cost, or revenue rather than user count.

FACT A claims user_count is '7 USD' (malformed data with incorrect units for a user count attribute), while FACT B claims user_count is '7,000,000 users' observed on 2026-04-26. These values differ by 6 orders of magnitude. FACT A's unit (USD) is invalid for a user count attribute, indicating data corruption or entry error. The numeric values (7 vs 7,000,000) represent an irreconcilable contradiction.

Fact A reports user_count as '7 USD' (nonsensical for a user count metric, appears to be a currency amount), while Fact B reports user_count as '7000000 investors' — fundamentally incompatible values for the same attribute. Additionally, Fact A lacks an observation date while Fact B is dated 2026-05-01, suggesting one is stale or corrupted data.

Both facts claim to measure 'user_count' for Simply Wall St Community, but report incompatible values: Fact A reports '7 USD' (which is nonsensical for a user count) while Fact B reports '7000000 users' with a timestamp of 2026-05-21. These represent a direct value conflict—the two measurements differ by several orders of magnitude and use incommensurate units.

Both facts claim to measure 'user_count' for the same entity but report drastically different values: 7 USD vs 7000000 users. Additionally, Fact A's value (7 USD) is semantically invalid for a user_count attribute—user counts should be unitless numbers, not currency amounts. Fact B's value and unit (users) are logically consistent with the attribute name. Either Fact A contains corrupted/mislabeled data, or the values were sourced from incompatible systems.

Both facts claim to represent user_count for the same entity, but assert dramatically different values: FACT A reports '7 USD' (malformed data mixing currency with count), while FACT B reports '7000000 investors'. Ignoring units, the numerical values (7 vs 7,000,000) are incompatible. FACT A also exhibits a data quality issue—'USD' is inappropriate for a user count metric.

FACT A reports user_count as '7 USD' (invalid unit for user count, appears to be corrupted data), while FACT B reports '7000000 users' (7 million). Setting aside the unit mismatch in FACT A, the numerical values conflict dramatically (7 vs 7,000,000). FACT A also lacks a valid timestamp ('None'), indicating data quality issues. FACT B has a proper observed date and valid units.

FACT A assigns user_count a value of '7 USD' (a monetary amount), while FACT B assigns it '7000000 users' (a numeric count). These are fundamentally incompatible units for the same attribute. User counts must be expressed as numbers of users, not currency. FACT A appears to be mislabeled—the '7 USD' value likely represents a different attribute (e.g., price or subscription cost), not user_count.

Both facts claim to measure the same attribute (user_count) for the same entity, but provide completely incompatible values. FACT A states '7 USD' (currency, likely a data entry error) while FACT B states '7000000 users' (7 million users). These cannot both be true for the same metric. Additionally, FACT A uses an incorrect unit (USD) for a user count measurement.

The same attribute (user_count) for the same entity has two fundamentally incompatible values: 'Fact A' assigns a monetary value (7 USD), while 'Fact B' assigns a user count (7000000 users). These are different units/types and cannot both be true simultaneously. This likely indicates a data quality issue, misattribution, or confusion between an unrelated metric (price/cost) and the actual user count.

FACT A claims user_count is '7 USD' (appears to be malformed data mixing a quantity with currency), while FACT B claims user_count is '7000000 investors' (7 million). These are incompatible values for the same attribute. FACT A's value is also logically invalid—a user count should not be expressed in USD currency units. The magnitude difference (7 vs 7,000,000) and the data quality issue in FACT A indicate a likely data corruption or incorrect field mapping in FACT A.

The same attribute 'user_count' has two incompatible values: 'FACT A' shows '7 USD' (monetary unit), while 'FACT B' shows '7000000 users' (user count unit). Beyond the unit mismatch, the numeric values differ drastically (7 vs 7,000,000). FACT A appears to be a data entry error—USD is not a valid unit for user_count. FACT B is the logically correct representation.

Same attribute (user_count) has fundamentally incompatible values: FACT A reports '7 USD' (a currency measurement) while FACT B reports '7000000 users' (a user count). FACT A appears to contain a malformed or incorrectly labeled value—user_count should never be expressed in USD currency units. This indicates either data corruption in FACT A or a labeling/schema error where FACT A belongs to a different attribute (e.g., subscription_price or cost_per_user).

Fact A states user_count as '7 USD' (malformed—currency unit applied to a count metric), while Fact B states user_count as '7000000 millions' (7 billion). These represent incompatible values for the same attribute. Fact A's value is nonsensical (USD is not a valid unit for user count), and Fact B's value is implausibly large for a community platform. Even accounting for possible data quality issues, these cannot both be correct.

Both facts reference the same attribute (user_count) but with conflicting values and incompatible units. FACT A reports '7 USD' (price in currency, not a user count), while FACT B reports '7000000 users' (a proper count). FACT A's units (USD) are inconsistent with the attribute type (user_count), suggesting corrupted or misattributed data. The numeric values differ by orders of magnitude (7 vs 7,000,000).

Fact A claims user_count is '7 USD' (invalid data—currency unit instead of a count), while Fact B claims it is '7000000 millions' (7 trillion—implausibly high for any community; world population is ~8B). Both values are conflicting claims for the same attribute, and both exhibit data quality issues: Fact A has wrong units/corrupted data, Fact B has likely a multiplication error or magnitude mistake. Neither is internally coherent, but they definitively contradict each other numerically.

The same attribute (user_count) is assigned two radically different values: 7 (with an invalid USD unit for a user count) vs 7,000,000 users. FACT A appears malformed—'7 USD' is a currency value, not a user count. Even if the USD unit is ignored, 7 vs 7,000,000 is a massive numerical conflict. FACT B represents the legitimate user count; FACT A indicates a data quality issue.

FACT A reports user_count as '7 USD' (a currency value), while FACT B reports it as '7000000 users' (a user count). These are incompatible value types for the same attribute. FACT A appears to be corrupted or mislabeled data—a monetary value cannot represent a user count. This is a clear data quality issue, not a temporal discrepancy.

Both facts claim to describe the 'user_count' attribute for Simply Wall St Community, but provide dramatically different values: Fact A states '7 USD' (with an invalid unit for a user count), while Fact B states '7000000 users'. These represent irreconcilable values for the same attribute. Additionally, Fact A uses currency (USD) as a unit for what should be a user count, indicating a data quality issue or attribute misidentification.

FACT A claims user_count = 7 USD (with currency units), while FACT B claims user_count = 7,000,000 users. These are incompatible values for the same attribute. FACT A is semantically invalid—currency (USD) is inappropriate for measuring user count. The values differ by six orders of magnitude and use contradictory units (currency vs. user count).

The same attribute (user_count) has two incompatible values: FACT A expresses it as '7 USD' (a monetary value), while FACT B expresses it as '7000000 users' (a quantity). This is a direct value conflict. Additionally, FACT A's value is logically incoherent—user counts cannot be measured in currency. FACT A appears to contain a data quality issue where monetary and user metrics were conflated.

Both facts claim to measure the same attribute (user_count) for the same entity, but FACT A assigns a monetary value (7 USD) while FACT B assigns a user count (7,000,000 users). These are incompatible units for the same metric. FACT A's value appears to be data from a different attribute (likely price or cost) that was misattributed to user_count.

Fact A claims user_count = '7 USD' while Fact B claims user_count = '7000000 users'. These contradict in two ways: (1) the numeric values differ dramatically (7 vs 7,000,000), and (2) Fact A uses an inappropriate unit (USD, a currency) for a user count metric, which should be dimensionless or measured in 'users'. Fact A appears malformed—either the attribute is wrong, the unit is wrong, or the value is corrupted.

FACT A reports user_count as '7 USD' (measured in currency), while FACT B reports '7000000 users' (measured in user count). These are fundamentally incompatible: they measure the same attribute using different units and have drastically different magnitudes (7 vs. 7,000,000). FACT A appears to be corrupted or misattributed data—user_count should never be expressed in USD.

Fact A reports user_count as '7 USD' (a monetary value), while Fact B reports it as '7000000 users' (a population count). These are fundamentally different data types and orders of magnitude. 'USD' is not a valid unit for user counts, suggesting Fact A contains a data type error or was misclassified. Fact B's observation timestamp (2026-03-01) provides a temporal anchor, while Fact A has no observation date.

FACT A describes the U.S. Adhesive Films Market (industrial/materials sector), while FACT B provides growth drivers rooted in healthcare (neurological disorders, stroke incidence, rehabilitation equipment). These domains are fundamentally incompatible—adhesive films are not driven by healthcare factors. This suggests SNS Insider either conflated two separate markets or misattributed market drivers to the wrong sector.

Free cash flow values differ by ~440x (8700 vs 19.7 USD) across only 1.5 years. While financial metrics can vary, this extreme divergence suggests a data quality issue—likely different calculation methods, unit mismatches (one possibly in millions while stated in dollars, or vice versa), or an extraction/measurement error rather than legitimate business performance change.

Two different EPS values (2.29 USD vs 2.14 USD) are recorded for the same entity with no period specified. While EPS naturally varies across reporting periods, both facts show 'Period: N/A', which is anomalous. If both represent the entity's 'current' or 'latest' EPS at their respective observation dates, the discrepancy without period clarity suggests either: (1) data quality issues, (2) different calculation methodologies, or (3) missing period information that would explain the difference. The ~3.5 month gap between observations makes a legitimate period change plausible, but this cannot be confirmed without period data.

Same entity and attribute (eps) have two significantly different values (2.27 USD vs 2.86 USD, ~26% difference). Both have Period: N/A, indicating they should represent the same reporting period. One has an observation timestamp (2026-07-15), the other has none, suggesting potential data source or timing misalignment. This represents a material conflict in financial data.

Both facts represent EPS (earnings per share) for the same entity with significantly different values: 2.27 USD vs 11.55 USD (~5.1x difference). This is a substantial numerical discrepancy for a critical financial metric. The contradiction is weakened slightly by: (1) missing observation date on Fact B, (2) both facts missing explicit period information (N/A), which could indicate different reporting periods, calculation methods (basic vs diluted), or data sources. However, the magnitude of difference suggests a genuine conflict that requires investigation.

The same entity has two significantly different EPS values: 2.27 USD (observed 2026-07-15) vs 12.40 USD (no observation date). A 5.5x difference in earnings per share is material. The contradiction is clear, though both facts lack period specification, and Fact B lacks a timestamp, which leaves some ambiguity about whether these represent different measurement contexts or methodologies.

Both facts reference the same entity and attribute (eps) but report substantially different values (2.27 USD vs 11.54 USD, a 5x difference). The absence of period information for both facts, combined with one having an observation timestamp (2026-07-15) and the other having none, suggests these may refer to different fiscal periods or data sources. Without explicit period identifiers, having two unreconciled EPS values for the same entity indicates a data quality issue that should be investigated.

The EPS values differ significantly (2.27 USD vs ~11.44 USD, approximately 5:1 ratio). However, the contradiction confidence is moderate because FACT B lacks period information. If FACT A represents quarterly EPS and FACT B represents annual/TTM (trailing twelve months) EPS, they would not contradict—they'd be different aggregation levels. If both represent the same period and measurement type (e.g., both Q2 2026 or both TTM), this is a clear value conflict. The missing observation date and period for FACT B introduces ambiguity about whether these should represent identical values.

Same attribute (eps) for the same entity shows significantly different values: 2.29 USD vs ~11.44 USD (~5x difference). Both facts lack period information, preventing determination of whether these represent different time periods. Fact A is timestamped (2026-07-15), while Fact B has no timestamp. The large magnitude of the discrepancy suggests either data quality issues, different reporting bases (e.g., diluted vs. basic EPS), or data from different periods that should be clarified.

The EPS value for Johnson & Johnson Services, Inc. shows a significant discrepancy (2.29 USD vs 11.54 USD, ~80% decrease). Observed on different dates (2026-07-15 vs 2026-03-03), these values cannot both represent the same reporting period. The critical issue is that both facts have 'Period: N/A', which suggests they may be claiming to represent current/general EPS values rather than specific quarterly or annual periods. If so, the entity cannot legitimately have two different current EPS values. If these are different reporting periods (Q1 vs Q2 2026), that would be normal time-series variation and not contradictory—but the missing period metadata creates ambiguity.

The same entity has two significantly different EPS values (2.29 USD vs 11.54 USD, a 5x difference). Without explicit period information for either fact, these cannot be reconciled as different reporting periods. The values are too divergent to be explained by rounding or minor data variance. Fact A was observed on 2026-07-15, while Fact B has no observation timestamp, suggesting potential data quality issues or sources reporting different metrics under the same attribute name.

Both facts report the adjusted_net_earnings attribute for the same entity (Johnson & Johnson Services, Inc.) at the same observation timestamp (2026-07-15 00:00:00), but with different values: 6699 USD vs 7081 USD. The difference of 382 USD (~5.7%) represents a material discrepancy that cannot be reconciled without additional context (such as different data sources, rounding, or revisions).

Same entity (Johnson & Johnson Services, Inc.) has two different adjusted_eps values (2.90 USD vs 2.77 USD) recorded for the identical observation timestamp (2026-07-15). This is a direct value conflict—adjusted EPS cannot simultaneously be both 2.90 and 2.77 for the same company on the same date.

Same entity (Johnson & Johnson Services, Inc.) reports two significantly different EPS values (2.29 USD vs 12.40 USD) without clear temporal differentiation. Both facts lack period information, making it unclear if they represent the same reporting period. Fact A has an observation timestamp (2026-07-15) while Fact B has none, suggesting potential data quality or source issues. The 5.4x value difference is substantial for a key financial metric.

Same entity and attribute (EPS) but with significantly different values: 2.29 USD vs 11.55 USD. Fact A has an observation timestamp (2026-07-15) while Fact B has no timestamp. Without period information or source metadata, it's unclear if these represent different time periods, data sources, or methodologies (adjusted vs unadjusted, diluted vs basic). The 5x difference is substantial and cannot be explained by minor calculation variations.

The two facts assert dramatically different values for the same attribute (free_cash_flow) on the same entity. Fact A reports $8,700 USD while Fact B reports $20 billion USD—a difference of approximately 2.3 million times. For a company like Johnson & Johnson with annual revenues exceeding $90 billion, the $8,700 figure is implausibly low and appears to be either corrupted data, missing unit conversion, or a measurement error. Fact B ($20 billion) is consistent with the financial scale of a Fortune 500 pharmaceutical company.

Same attribute (eps) has two significantly different values for the same entity: 2.29 USD (Fact A, observed 2026-07-15) vs 2.86 USD (Fact B, no observation date). The 20% variance (0.57 USD delta) is material. Both entries lack period information (N/A), making it unclear whether these represent different reporting periods or a data quality issue. Fact A's timestamp suggests it's more recent, but Fact B's missing observation date prevents determining which value is current or authoritative.

Fact A reports 8,700 USD in free cash flow while Fact B reports 18.54 billion USD — a ~2.13 million-fold difference for the same attribute. For an entity like Johnson & Johnson Services, Inc., both values are questionable (8.7K is implausibly low), but they cannot both be accurate for the same period. The timing discrepancy (Fact A has observation timestamp 2026-07-15, Fact B has none) suggests possible different periods or data sources, but without period clarity, this represents a material value conflict.

Fact A reports free_cash_flow of 8,700 USD while Fact B reports 35.5 billion USD for the same entity. These values differ by a factor of approximately 4 million. Both claim to measure the same attribute (free_cash_flow) for Johnson & Johnson Services, Inc., with no period specification to disambiguate them. One value is almost certainly incorrect or represents a data quality issue (e.g., missing scaling factors, unit conversion error, or corrupted value).

Same entity (Johnson & Johnson Services, Inc.), same attribute (free_cash_flow), same observation timestamp (2026-07-15 00:00:00), but conflicting values: 8700 USD vs 6214 USD. This represents a ~29% discrepancy and suggests either data from different sources, calculation method differences, or a data quality issue.

The same entity attribute (sales_territories) has two incompatible numeric values: 130 territories vs 80 territories. Both lack time period or observation timestamps, suggesting they represent the same state. The 38-territory difference (~38% variance) indicates either measurement error, different data sources, or outdated information.

The same attribute (margin) has two significantly different values: 46% vs 54.8%. Without period information or observation dates, it is unclear whether these refer to different time periods or represent a data quality issue. The 8.8 percentage point difference is substantial for a margin metric and suggests either different reporting periods, different calculation methodologies, or a data error.

Two different EPS values (0.44 vs 0.35 USD_diluted) are reported for KeyCorp at the same observation timestamp (2026-07-21). Since both facts refer to the same attribute, same entity, same time, and same measurement basis (diluted USD), they cannot both be true simultaneously. This indicates either duplicate data from conflicting sources, a data quality issue during extraction/collection, or an incomplete reconciliation of the facts.