Your company means four different things by the same word.
“Active user” means one thing in your analytics, another in finance, a third in the deck, a fourth in the churn dashboard — and a number quoted to your board is computed two ways by two teams who are both, locally, right. Driftvice is the first glossary that knows it: derived from where your language actually lives and reconciled against it every day — never a document someone has to remember to update.
Drift · metric-computation · high
term active user
BI where last_seen > now()-30d
warehouse …and internal = false
diverges at where (internal filter)
blast radius 4 surfaces · Q3 board deck
owner team:growth — routed
settle it once → watched everywhere →
The category
A glossary page decays on contact with reality. Driftvice is the opposite idea.
Language fragmentation isn't carelessness — it's entropy, the natural result of many people defining meaning in many places over years. The usual defense is a glossary page, and it's wrong the day after it's written, because it's disconnected from the surfaces where your language actually lives and changes.
Driftvice owns the gap no existing tool does — the cross-surface drift layer between defined meaning and used meaning. It is not a data catalog (it covers language, not data assets), not a semantic layer (it reads metric definitions and reasons about them; it never serves your business numbers), and not a writing checker (it detects fractured meaning, not prose quality). It connects your tools with your own access, reads the meaning in use on each, and continuously diffs them.
One subscription · one product
Seven surfaces over one company-language graph.
Derived Glossary
True by construction.
An always-current glossary computed from where your language actually lives — every term's agreed definition, approved aliases, scope rules, owner, and live drift status. Nobody maintains it; it updates itself against reality.
Drift Feed
See all four meanings at once.
Continuous detection across the full taxonomy — conflicting definitions, a metric computed two ways, one concept under three names, a board number that quietly changed in a commit. Each surfaced with evidence side by side, scored for blast radius, and routed.
Reconciliation
Make drift someone's to fix.
Every conflict routed to the owner who can settle it, with pre-drafted resolutions: declare canon, merge two terms, split a name doing two jobs, retire a stale alias, codify the inclusion rules — or formally accept a divergence. The ambiguity becomes a decision.
Canon & Governance
Settled stays settled.
Resolved terms become versioned, owned canonical objects with explicit scope rules and an executable canonical computation. Once canonical, usage is watched everywhere — so the drift you resolved can't quietly creep back.
Metric Lineage
Why the two numbers differ.
Trace a metric from the slide it's quoted on, back through the BI definitions, down to the warehouse computations — and see the exact clause where two paths diverge: a missing filter, a different window, a different denominator.
Coherence Scoring
Converging, or fragmenting?
A score for how coherent your company's language is — per surface, per domain, company-wide, with trends — plus the heavily-used terms nobody has governed yet, so cleanup targets the language that matters most.
Surfaces & Fixes
Read it where it lives. Fix it there.
Connect your own tools with your own access — analytics, BI, warehouse, docs, code, templates — self-serve, revocable. Then close the loop: propose fixes, sync canon into the tools that hold definitions, and stop a governed metric from being silently redefined in code before it ships.
The drift taxonomy
Every way meaning fractures. Caught between surfaces, where nobody is watching.
Each version looks fine in isolation. The conflict only exists between surfaces. Driftvice watches there — classifies each divergence, scores how widely it reaches and whether a real decision is exposed, and sends it to someone who can resolve it.
Definitional conflict
One concept, materially different definitions across surfaces — "active user" meaning four things.
Metric-computation drift
The same metric computed differently — a filter here, a window there — located down to the exact clause.
Naming drift
One concept wearing many names, or one name covering two genuinely different concepts.
Scope / boundary drift
Agreement on the concept, disagreement on what it includes — trial users? internal accounts? bots?
Temporal drift
A definition or computation that changed on one surface and never propagated — including a silent change in a commit.
Canon-vs-usage divergence
A surface that quietly ignores the definition everyone agreed to.
Your board numbers reconcile.
Trace a metric from the slide back through the BI definitions to the warehouse, and see the exact clause where two computations diverge. The disagreement stops being a suspicion and becomes a line you can point at — before it's on a slide.
Cleanup becomes a state you hold.
Reconcile a term once, and its usage is watched across every connected surface. The moment a new surface diverges, it's caught against canon immediately. Language cleanup stops being a project you redo every year.
One answer for your agents.
Your agents consume your data at scale. Your canonical definitions, names, scope rules, and computations are queryable as ground truth, so an agent resolves the authoritative meaning before it reports a number — and can tell settled from drifting.
Stop finding out about fragmented language in the meeting where two numbers don't match.
Start free — connect a couple of the tools where your language already lives, and watch your real drift surface on your own terms in minutes.
Trustvice 360Declare as regras. Deixe cada execução ser verificada.
Agentes, evidências, governação, comércio e divulgação estão no Trustvice 360. Registe-se grátis e abra todas as suítes a partir de um só painel.