> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vanticlab.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Friction Score
> Four structural dimensions, score bands with colour references, and why Vantic IQ stays quiet until confidence is justified.
export const diagnosticFrictionScoreCardLine = `How the Friction Score becomes a ${diagnosticWindowLabel} operational report.`;
## The four dimensions
Your Friction Score combines four independently calculated dimensions of operational friction. Each captures a different way time leaks from structure.
### Tool Fragmentation
How spread across applications a given class of work is. Many applications touched for one workflow usually means more context recovery and coordination overhead.
**Lower:** A tight toolset per workflow. **Higher:** Constant switching across many applications.
### Workflow Consistency
How repeatable your workflow sequences are. Stable A → B → C patterns are easier to execute reliably than ad-hoc ordering every time.
**More friction:** High variance run-to-run — a different order every time. **Less friction:** Predictable, repeatable sequences.
### Context Switching
How often you change focus in rapid succession. Rapid back-and-forth usually reads as coordination cost, interruption load, or cross-app copy-paste overhead.
**Lower:** Meaningful dwell time per application. **Higher:** Rapid, frequent switching.
### Workflow Duration
Sessions that run materially longer than your established baseline for comparable work. Duration spikes often flag blockage, rework, or handoffs that did not complete.
**Lower:** Stable, predictable session durations. **Higher:** Frequent unexplained spikes.
***
## Score bands
| Range | Band | Colour | What to do |
| ------ | ------------ | --------------- | ---------------------------------------------------------- |
| 0–29 | **Low** | Aqua `#06B8A9` | No urgent action needed. Review for drift over time. |
| 30–59 | **Moderate** | Amber `#BA7517` | Review the top one or two contributing patterns. |
| 60–79 | **High** | Red `#DC2626` | Prioritise the highest-scoring dimension this week. |
| 80–100 | **Critical** | Red `#DC2626` | Treat as urgent. The top pattern is costing material time. |
These four bands — Low, Moderate, High, Critical — are the only band labels
used in Vantic IQ. No other labels exist in the product. Hex values match
Brand-Identity-Stack v2.3 (Low band Aqua `#06B8A9`).
## The confidence gate
Vantic IQ does not publish the **composite** Friction Score (0–100) until the Pattern Engine has sufficient confidence. That means:
* A minimum number of workflow instances detected — repetition, not a single unusual day
* Enough dimensions crossing a reliability threshold independently
* Signal quality high enough that a published headline score is not misleading
The Pattern Engine uses a two-tier confidence model:
* **Engine gate:** The minimum bar to produce any result at all. Below it, the engine returns nothing — no score, no estimate.
* **Display gate:** A higher bar that must be cleared before the full composite Friction Score is published. Between the two gates, partial signal context may appear in the conversational interface without a composite score.
Note: before the full composite score is ready, you may see individual dimension signals in the conversational interface. Those are partial signals, not a partial Friction Score — the composite score is only published when all required dimensions reach the confidence threshold.
If the composite gate is not met, you do not get a headline Friction Score that implies the full model is ready — not a partial composite, not a "low" estimate.
That is separate from **early-window signals**: you may still see **Friction Pulse** signals (single facts, no multi-pattern claim), and some accounts may show **per-dimension** readouts before the composite is final, as described in [First use](/install-setup/first-use) and [Quickstart](/install-setup/quickstart). The confidence gate applies to the **full composite**, not to every individual signal detection.
## Why waiting is a feature
A score built from thin or non-representative data would be worse than no score. If your first week was unusually busy, unusually quiet, or atypical in any way, the Pattern Engine will wait for a clearer picture. This is by design.
## Optional benchmarking
If sector benchmarking is enabled on your account, your Friction Score is displayed alongside a contextual range for your industry vertical. This is opt-in and can be disabled at any time. See [Sector benchmarking](/concepts/sector-benchmarking).
***
{diagnosticFrictionScoreCardLine}
How your score is contextualised against patterns in businesses like yours.