# Why do two dashboards show different numbers? > Usually both are right by their own hidden definitions — the real defect is that neither number's derivation can be produced on demand. A ten-minute test. - Canonical: https://hadalinstruments.com/research/why-do-two-dashboards-show-different-numbers/ - Published: 2026-08-03 - Author: Hadal Research - Answers the question: "why do two dashboards show different numbers" - Coins the term: **The Orphan Number** — A figure on a dashboard whose derivation nobody can produce on demand — source, filters, joins, definition, refresh time — and which therefore cannot be distinguished from a wrong number by anyone reading it, however correct it happens to be. --- ## The short answer Because each dashboard is answering a subtly different question and neither can prove which. The same business word — revenue, active users, conversion — gets operationalised twice: different filters, different join paths, different treatment of refunds or test accounts or timezones, different refresh moments. Both surfaces compute their own definition faithfully; the divergence lives in the undocumented space between definitions. That much, the better explainers already say. What they stop short of is the diagnostic that actually bites: ask for either number's full derivation — source query, transformations, definition, refresh stamp — and watch whether anyone can produce it. A figure whose derivation nobody can produce on demand is what I call **an orphan number**, and an estate that runs on orphans doesn't have a dashboard problem; it has an epistemic one that dashboards merely display. **WHAT THIS IS — AND WHAT IS NOT PUBLISHED.** This article is method, and its subject is the reader's own estate — no vendor, tool or platform is measured or ranked here, and dashboard tooling is not the defect: orphanhood is tool-agnostic. Where a specific estate needs its lineage established, that is an audit — see the close of this piece. Status of any measured claim: `NOT YET COMPUTED`. ## Prerequisite Knowledge You need viewing access to the two disagreeing surfaces and the standing to ask one question of whoever owns them: "walk me from this figure back to the primary record." No SQL required for the test itself — the test is whether the walk can be produced at all, and by whom, and how long it takes. ## Where the divergence actually enters Four gaps, in rough order of frequency. **Definitions:** the semantic gap — one surface counts an order at placement, the other at payment; one excludes internal accounts, the other never heard of them. Nothing is broken; two truths diverge. **Transformations:** joins that duplicate or drop rows under conditions nobody re-checked, filters applied by default in one tool's data model. **Time:** refresh cadence and caching, so the surfaces describe different moments while wearing the same date. **Source drift:** the upstream table changed shape or meaning, and one pipeline noticed. Every one of these is invisible on the dashboards themselves — which is the structural point: the surface is where divergence is *noticed*, never where it can be *diagnosed*. ## The ten-minute test Pick the single number that matters most on either surface. Ask its owner to produce, in writing: the source it draws from, the query or transformation chain, the definition in business terms, and the timestamp of its last refresh. Then apply the only scoring that matters: **produced within a day, or not.** If it can be produced, your divergence is a solvable afternoon of comparing two derivations. If it cannot — if the answer involves archaeology, a departed colleague, or "the tool handles that" — you have found an orphan, and you have learned something more important than why the dashboards disagree: you've learned that *agreement* would have been unverifiable too. Run the same test on the other surface. Two derivations resolve a dispute; one derivation wins it by default; zero derivations mean the dispute was never about data at all. ## The Observable Mechanism Everything here is checkable inside your own organisation with no external access and no tooling: the test's output is a document that either exists or doesn't within a stated time. That is deliberate — lineage is not a property a vendor can sell you retroactively; it is a property your estate either practises or lacks, and ten minutes of asking establishes which. ## What This Does Not Establish (The Limits) This article establishes how to detect orphanhood — not which of two numbers is right, which requires recomputation from the primary record, and not that traceability guarantees correctness: a fully-documented pipeline can faithfully implement a wrong definition. It also does not establish that every number deserves the full test — estates run on thousands of figures, and the test is for the ones decisions actually lean on, which is a shorter list than the dashboard suggests. ## Where this leads The audit-grade version of the walk — every load-bearing figure traced to its source with the gaps named, and FABRICATED available as a first-class finding for the numbers that trace to nothing — is the [Dashboard Lineage Audit](/instruments/dashboard-lineage-audit/)'s whole job. The stronger property, correctness by independent recomputation, is the [Reproducible Verdict Kernel](/instruments/reproducible-verdict-kernel/)'s ground. And the vocabulary is in the [lexicon](/lexicon/) — the orphan number joins it with this article as its receipt, because a term, unlike a dashboard figure, should carry its derivation with it. --- ## Claims examined ### Claim 01 — canonical: https://hadalinstruments.com/refutations/#claim-0a01c6f3 > "The numbers are different because one of the dashboards is broken." — our reading: Misleading In most real cases neither dashboard is broken: each is faithfully computing a differently-defined thing — different filters, different join logic, different refresh moments, different definitions of the same business word — and the divergence lives in the undocumented gaps between those choices. Hunting for the broken one presumes a shared definition that usually never existed. The productive question is not which surface failed but whether EITHER number's full derivation can be produced — because a divergence between two traceable numbers resolves in an afternoon, and a divergence between two orphans resolves never. **What is true:** Two dashboards usually disagree because they compute differently-defined quantities correctly, so the resolving act is producing both derivations — source, filters, joins, definition, refresh time — and comparing those, not adjudicating the surfaces. Evidence: https://hadalinstruments.com/research/why-do-two-dashboards-show-different-numbers/#claim-0a01c6f3 ### Claim 02 — canonical: https://hadalinstruments.com/refutations/#claim-808a9c2c > "We have a single source of truth, so our dashboards agree." — our reading: Unproven A shared source bounds one class of divergence and leaves the others untouched: divergence enters BETWEEN the source and the surface — in the semantic layer where the same word gets two definitions, in transformation logic, in caching and refresh cadence, in filters applied by default in one tool and not the other. Agreement is a property to verify at the surface, not a property inherited from the warehouse; declaring it from the architecture diagram is exactly the claim-without-receipt shape that keeps this genre of surprise alive. **What is true:** A single source of truth constrains where numbers start, not what they become on the way to a screen; whether two surfaces agree is verified by comparing their rendered numbers and derivations, never inferred from a shared backend. Evidence: https://hadalinstruments.com/research/why-do-two-dashboards-show-different-numbers/#claim-808a9c2c ### Claim 03 — canonical: https://hadalinstruments.com/refutations/#claim-6a970b61 > "The totals match across our dashboards, so the numbers are right." — our reading: False Matching establishes consistency, not correctness: two surfaces reading the same upstream table inherit its errors in perfect agreement, and two pipelines can share the same wrong transformation. Correctness is a different property with a different test — recomputing the figure from the primary record and getting the same answer. Agreement between derived surfaces is evidence the plumbing is consistent; it is silent about whether the plumbing carries the truth. The two properties fail independently, and the expensive failures are the consistent ones, because nothing on any screen looks wrong. **What is true:** Agreement between dashboards proves consistency; correctness is proven only by independent recomputation from the primary record — and a consistent estate can be consistently wrong, which is the failure mode nothing on screen reveals. Evidence: https://hadalinstruments.com/research/why-do-two-dashboards-show-different-numbers/#claim-6a970b61 ## Cite This Article APA BibTeX HTML Hadal Research. (2026). Why do two dashboards show different numbers?. Hadal Research. https://hadalinstruments.com/research/why-do-two-dashboards-show-different-numbers/ Version e472963, 2026-09-14. @misc{hadal_2026_why-do-two-dashboards-show-different-numbers, author = {Hadal Research}, title = {Why do two dashboards show different numbers?}, year = {2026}, url = {https://hadalinstruments.com/research/why-do-two-dashboards-show-different-numbers/}, howpublished = {Hadal Research}, version = {e472963}, note = {Published: 2026-08-03; version dated 2026-09-14} } Source: Hadal Research, Why do two dashboards show different numbers?. Original Research Copy Citation **Version e472963** identifies the commit that last changed this page in Hadal's content repository. That repository is not public, so the identifier does not resolve externally — it is published so a citation pins one specific state rather than a moving page. To obtain the exact version cited, use the [press and research route](https://hadalinstruments.com/press/). ## Explore further ### Instruments - [Dashboard Lineage Audit](https://hadalinstruments.com/instruments/dashboard-lineage-audit/) - [Reproducible Verdict Kernel](https://hadalinstruments.com/instruments/reproducible-verdict-kernel/) ### Concepts - [Point-in-Time Data](https://hadalinstruments.com/glossary/point-in-time-data/) - [Tick Data Provenance](https://hadalinstruments.com/glossary/tick-data-provenance/) [All Hadal research](https://hadalinstruments.com/research/)[This article as plain markdown](https://hadalinstruments.com/research/why-do-two-dashboards-show-different-numbers.md) --- ## Raw artifact — NOT PUBLISHED FOR THIS PAGE No downloadable artifact ships with this page. Eight published measurements do, each content-hashed so a reader can verify the figures independently. Where a measurement is published here without one, that is a gap rather than a policy, and it is stated rather than left to be noticed. [Measurements that ship their data](https://hadalinstruments.com/research/)