Growth maturity — calibration, pillars, and next steps (v3)¶

Growth scores six pillars from four shared primitives, calibrates them against your organisation's own baseline rather than a global leaderboard, and lets HERC suggest — never auto-apply — your next step.

This page pairs with Growth: that page owns the home-page walkthrough, this one explains the model underneath — how v3 calibrates you, how pillars are built, and how next actions are chosen — deliberately without engineering detail.

Pillars6

Primitives4

StagesReactive → Prescriptive

Modelv3

What the model gives you¶

A baseline that fits you

Scores are read against your organisation's shape and history, not raw global counts. Everyone can start; the question is "how far along your expected path are you?"
One headline, one bottleneck

The composite blends all pillars (equal weights by default). When one area lags, it is called out separately — your headline is not "only as good as the worst pillar."
Confidence you can read

Coverage and projections carry a low / medium / high cue so the UI never overclaims from a thin or noisy baseline.
A next step, on request

HERC suggests a ranked next action with an expected lift — you choose whether to run it. Nothing is applied automatically.

The mental model

Think of Growth like a fitness tracker. Your journey is personalised to your organisation's size and industry; the model does not rank you on raw catalogue size. It asks how far along your own expected path you are, and where the next gain is cheapest.

How maturity is built¶

The same four primitives are measured for every pillar, rolled up into a per-pillar score, and blended into a composite headline. The weakest pillar becomes the bottleneck, and HERC reads the whole picture to propose a next step.

flowchart LR
    COV["Coverage"] --> ROLL
    DEN["Density"] --> ROLL
    USE["Usage"] --> ROLL
    QUA["Quality"] --> ROLL
    ROLL{{"Roll up per pillar<br/>(geometric mean)"}} --> PILLARS["Six pillars<br/>0–100 each"]
    PILLARS --> HEAD(["Composite headline<br/>all pillars, equal weights"])
    PILLARS --> BOT["Bottleneck<br/>weakest pillar,<br/>if ≥5 pts behind next-lowest"]
    HEAD --> HERC["HERC next-action<br/>suggestion"]
    BOT --> HERC
    classDef accent fill:#ff8400,stroke:#e67700,color:#ffffff,font-weight:600;
    class HEAD accent

Figure 1 — Four primitives feed every pillar; pillars compose into a single headline and expose a bottleneck; HERC reads the whole snapshot to suggest what to do next.

The four primitives¶

Every pillar is a 0–100 composite of the same four primitives, interpreted against your organisational baseline.

Primitive	Meaning (plain language)
Coverage	How much of what we expect for your org (terms, assets, contracts, products, active users) you actually have — a ratio, not a headline count.
Density	How interconnected things are: links, lineage, contract monitoring, graph relationships, fewer orphans.
Usage	Consumption of governed artefacts: HERC and dashboard activity, glossary lookups, steward engagement — not vanity production alone.
Quality	Durability under use: sensible glossary churn, fewer contract violations, freshness, steward responsiveness, answer satisfaction.

Primitives roll up per pillar with a geometric mean by default, so a single zero drags the pillar down honestly rather than being averaged away.

The six pillars¶

Pillar	What it reflects
Catalog coverage	How much of the Data Catalog is governed and represented.
Glossary completeness	Depth and reach of the Business Glossary.
Quality / contracts	Data contracts and the quality signals they monitor.
Automation footprint	Flows and assisted workflows in active use.
Intelligence adoption	Use of HERC and assisted workflows (see AI Fluency below).
Stewardship engagement	Steward activity and responsiveness.

Each pillar also carries a stage on the four-rung ladder — Reactive → Proactive → Predictive → Prescriptive — worst to best.

From pillars to the headline¶

The composite stage reflects all pillars, not just the weakest. When one pillar sits 5 or more points below the next-lowest, the home page shows a bottleneck callout with that pillar's name and the approximate lift if you closed the gap. Tight clusters hide the callout, so ordinary noise never looks like a crisis.

Production vs consumption¶

The production vs consumption insight compares governed output to downstream use (analytics, HERC, references). A healthy band is often around 0.5–2.0, tuned per org. When the ratio sits outside that band, the home page folds the imbalance into the bottleneck callout as a single editorial line — for example:

How the imbalance reads on home

"Consumption is outpacing production by 1.4× — this pillar would also lift adoption."

Useful for deciding whether to publish more or activate what already exists. The standalone dial moved to the /dna panel in the visual uplift; the home keeps the same insight in fewer pixels.

Confidence¶

Coverage and projections carry a low / medium / high cue, so the interface never prescribes from an unstable baseline.

Band	Coverage — what it means	Treat it as
Low	Little baseline yet, or a noisy fit.	Directional only.
Medium	Observed or hybrid baseline with enough confidence for prioritisation.	Good enough to prioritise.
High	Strong hybrid baseline and industry priors (after opt-in) — the priors are required to reach this band, not just a refinement.	Sharp — act with confidence.

Headline confidence is the weakest pillar band, so the UI does not overclaim on the strength of one well-calibrated area.
Goal and pace projections use R² on recent history: fewer than 14 days or R² below 0.30 → low; 0.30–0.70 → medium; 0.70 or above → high.

A high headline needs industry priors

Observation and daily snapshots on their own lift the headline to medium at most. Reaching high requires opting in to industry priors on top of a strong hybrid baseline — the sharper denominators are a prerequisite for the top band, not merely a nice-to-have. Until you opt in, the headline stays at medium even when every pillar is well calibrated.

Confidence is not fixed — it climbs as daily snapshots accumulate and (after opt-in) as industry priors sharpen the denominators, and it can slip when signals turn noisy or the baseline is still thin.

stateDiagram-v2
    direction LR
    [*] --> Low
    Low --> Medium: more daily snapshots
    Medium --> High: strong baseline + industry priors
    High --> Medium: signals turn noisy
    Medium --> Low: baseline still thin
    classDef accent fill:#ff8400,stroke:#e67700,color:#ffffff,font-weight:600;
    class High accent

Figure 2 — Confidence matures with observation and (optional) industry priors, and can regress when the signal weakens. Low is directional; medium is enough to prioritise; high is sharp.

Setting goals¶

You can set one active goal per pillar — a target stage or score and a deadline. Pace and horizon estimates are projected against that goal, and HERC can discuss goals via the assistant when your permissions allow.

Default goals on day one¶

Every organisation starts with six default goals — one per pillar — seeded on the day onboarding completes (or when an admin skips onboarding). Each default goal:

Targets the next stage up from your current pillar stage (Reactive → Proactive → Predictive → Prescriptive), never beyond Prescriptive.
Sets the target score at the middle of that stage band.
Carries a 60-day deadline for your two weakest pillars (HERC gives the bottleneck the tighter loop) and a 90-day deadline for the rest.
Wears a "Default goal — tweak anytime" badge on the Goals surface and on the spotlight goal card on home.

Editing any field on a default goal (target stage, target score, or deadline) immediately clears the badge — it becomes yours. You can also archive the goal and create a different one on the same pillar.

For organisations that existed before this change, a one-shot backfill ran on deploy: every org that lacked an active goal on a pillar received the matching default. Any goals you had already set were preserved unchanged.

You should never see an empty Goals page

Seeding is idempotent — pillars that already have an active goal are skipped. If you ever land on a Goals surface with no rows (for example, an org that joined before seeding shipped), the home page seeds the six defaults for you on the next visit.

Next actions from HERC¶

When conditions allow, HERC may suggest a ranked next action — what to do, the expected lift, and a rough time horizon. Each suggestion carries a confidence cue aligned with headline and baseline reliability (the same low / medium / high idea as elsewhere).

flowchart LR
    SNAP["Daily snapshot"] --> HASDATA{"Snapshot<br/>available?"}
    HASDATA -- no --> SUP1["Suppressed<br/>'not enough data yet'"]
    HASDATA -- yes --> CONF["Headline confidence<br/>= weakest pillar band"]
    CONF --> GATE{"Medium<br/>or high?"}
    GATE -- low --> SUP2["Suppressed<br/>'confidence still building'"]
    GATE -- "medium+" --> PICK(["Ranked next action<br/>+ confidence cue + CTA"])
    PICK --> YOU["You choose to run it<br/>never auto-applied"]
    PICK -. records .-> AUDIT["Audit trail<br/>suggested + acknowledged"]
    classDef accent fill:#ff8400,stroke:#e67700,color:#ffffff,font-weight:600;
    class GATE accent

Figure 3 — A suggestion only appears when there is a snapshot and headline confidence is medium or higher. Suggestions never auto-apply, and every one is recorded for accountability.

Prefer acting on medium or higher when you need a crisp priority.
Suggestions never auto-apply — you choose what to run.
The platform keeps an audit trail of what was suggested and acknowledged.
If headline confidence is low, suggestions may be suppressed so we do not prescribe from unstable baselines.

AI Fluency¶

AI Fluency is a sublayer of Intelligence Adoption — not a seventh pillar. It reflects how deeply HERC and agents learn your org over time: chat depth, query diversity, retry patterns, satisfaction trend, and related signals.

Depth over volume

Depth and balanced use matter more than raw chat volume alone — a few deep, varied, satisfying conversations move AI Fluency further than many shallow ones.

Industry priors (opt-in)¶

Your administrator can opt in to contributing aggregate telemetry (industry and size-band buckets) so Datahub can refresh industry priors — the denominators that help promote baseline confidence toward high for organisations of a similar shape.

Aspect	What applies
Privacy	Payloads are aggregate-only (bucketed counts / ratios, k-anonymity safeguards on the collector) — not raw assets, terms, names, or user identifiers.
Default	Off for everyone; you can turn the toggle off at any time.
Stub mode (rollout)	While the collector base URL is unset, the app stays in stub mode: no outbound POST and no payloads leave the environment, regardless of opt-in. Opt-in still records intent; live delivery starts only once the URL and signing keys are configured.
Retention	Previously submitted aggregates are not recalled from the product UI; operational retention is defined by your collector and release process.

What you see over time — day 1 / 10 / 30¶

The Growth home answers one question from first load: "where am I heading next?" You never land on a stack of "HERC is calibrating" empties.

A "Where you're heading next" panel sits near the top and shows up to four milestones HERC projects on your horizon. Each step carries:

A day chip — Today, Tomorrow, or in N days.
A confidence badge — gaining clarity, on track, or confident.
A headline — the pillar and stage you are about to cross (e.g. "In 12 days: Proactive on Catalog Coverage").
A body line — one concrete signal that would move you across.
An optional Act on this button that deep-links into the catalog / glossary / contracts surface where that signal is created.

When there is not yet enough signal to project a date, the copy pivots to an optimistic calibrating voice — never "we don't have your data yet". It reads "First personalised milestones in about a week — keep working in Datahub and the picture sharpens."

Day	What you see	Why it matters
Day 1 (baseline just captured / skipped)	Forward-milestones panel with a `calibrating` step + 6 default goals, each with the "Default goal — tweak anytime" badge.	You have a starting horizon and a starting plan from the first login. Nothing empty.
Day 10 (first snapshots in)	At least one milestone flips to `steady climb` copy as HERC sees your primitives move. Pillars with positive slope start projecting precise dates.	You see the first real pace signal and learn which pillars are already compounding.
Day 30 (calibration threshold crossed)	Milestones flip to `reachable` with concrete `in N days: Stage on Pillar` headlines. Your 60-day bottleneck goal is typically two-thirds of the way to target.	The home transitions from aspirational to precisely predictive — the forward panel stays as the ongoing 30/60/90 motivator.

The Growth home, section by section¶

Retained for continuity

These layout notes describe the same data through a redesigned shell. The day-to-day home-page walkthrough lives in Growth; they are kept here so calibration and presentation stay documented together. Same data, same numbers, same hooks — different shell.

The home is a tiered grid, not a full-width stack, designed for business users and analysts who land here for guidance rather than diagnostics.

Tier	What you see	Why it is there
1 — Hero ring	Composite ring on a cosmic canvas with two satellites and one stage-aware narration line. The left satellite is `Stage N of 4` so you know where you are on the ladder; the right celebrates a recent win (a capability unlocked in the last week, or the pillar with the biggest 30-day jump, or — for a new org — a hopeful "First wins are weeks away"). The narration line names what unlocks at the next stage against your weakest pillar.	One question, one stage, one win, one next step. No jargon.
2 — Where you are heading next	A 1/2 + 1/2 row pairing the forward-milestones panel with either HERC's confident next-best action or your spotlight goal card.	The two forward-looking surfaces sit side by side, so the page reads as a plan, not a list.
3 — Where the maturity comes from	Six pillar cards in a 2-column grid — score, stage badge, full-width range bar with a stage-band gradient, narrative line, Open pillar button, and a primitives chevron. A pillar holding steady at zero movement shows a muted minus and a "Holding steady" label — never a red down-arrow.	Six tiles read as a dashboard; six full-width rows read as a wall of text.
4 — Your story	A 2/3 + 1/3 row. On the left, the journey timeline wrapped in a "How you got here" card (a friendly dotted line + aspirational ETA when there are no unlocks yet). On the right, a recent capability-unlock celebration, otherwise this month's contribution summary.	The journey reads as a section, not a floating timeline; the sidebar honours "celebrate wins" without full-width real estate.
5 — Try a small change	A full-width what-if insight tile and drawer for modelling small actions against the projected stage crossings.	The experimentation surface stays at the bottom so it does not interrupt the day-1 narrative.

A slim bottleneck banner appears between the stage strip and Tier 2 only when one pillar lags the next-lowest by 5 points or more, with the production-vs-consumption insight folded in as a single line.

What changed in the April 2026 visual uplift

The earlier redesign moved the home from information density toward storytelling — same data, same numbers, different shell.

Section	Before	Now
Hero	Composite number on a warm radial blob	Composite ring (WHOOP-style two-arc gauge) on a cosmic canvas with an "EXPLORE YOUR JOURNEY →" CTA. Five hero compositions exist (`ring`, `blob`, `vertical`, `compact`, `wide`) — `ring` is the default.
Production vs Consumption	Standalone chat-bubble dial competing with the hero	Removed from home; the insight folds into the bottleneck callout. The dial still lives on `/dna`.
Pillars	Compact rows with a flat range bar	Editorial rows: identity, full-width range bar with stage-band gradient + circle marker, narrative column, pace delta + "VIEW TREND →". A 30 day / 6 month window pill sits at the header.
Journey	Bare grey line with a text list	JourneyTimeline organism: dates above the line, milestones below, Today pulse marker, capability-unlock chips. Ladder and river variants exist.
Just unlocked	(no surface)	CapabilityUnlockCard rendered inline when an unlock crossed in the last 7 days, with two CTAs and a HERC-notified line.
Empty / stale states	Bare hint placeholders	MaturityStateCard — section-aware copy with an always-on CTA.
What if	Flat form drawer	1·2·3 layout: Pick a signal → Set the delta → Projected outcome with a stage-tinted range bar; the closed tile carries a sparkline preview.

Tour selectors are unchanged (growth-pillars, growth-journey, growth-whatif) so existing screenshots and support docs still land on the right anchors.

FAQ¶

Question	Answer
Why did my score drop?	Baselines and primitives move when expectations catch up to observation, when usage / consumption shifts, or when quality signals worsen. Check the bottleneck and primitive breakdown on home.
Why no recommendation today?	Often low headline confidence (baseline still thin or noisy) or engine suppression — finish onboarding, wait for more daily snapshots, or check the confidence labels on home before expecting a prescription.
Can I get a different baseline?	Re-run onboarding or let passive observation refine expectations; admins control industry priors opt-in where applicable.
What changed from before?	v3 replaces the weakest-pillar-only headline with a composite + bottleneck callout, adds primitives and baseline ratios, goals-first projections, the production vs consumption insight, AI Fluency, and optional industry priors — scores are not point-for-point comparable to older major versions.