Начать анализ игры

Start analyzing a game

Public analysis helps you explore a game's audience, activity and rating.

How to use it

Open Analysis and find a game by name or Roblox URL.
Add the game and select Refresh.
Select two to ten games for comparison. View their metrics over the same period.

Advice and practical context

Public-analysis boundary

Sources provide public CCU, Visits, favorites, votes, metadata, and observed history. MARPLA derives metrics only from those inputs and exposes their formulas. Private D1, session time, acquisition sources, and revenue cannot be reconstructed by guessing.

Data quality first

Before comparing two to ten games, verify Universe IDs, snapshot times, and window coverage. A game added later has shorter history. Do not read an incomplete series as low activity or compare one game's current snapshot with another's mean.

Account for vote count in ratings, the cumulative nature of Visits, and the lack of release detail in updatedAt. These three limits prevent common false conclusions. Keep source and formula available beside every derived signal.

Technical detailsCalculations, permissions and behavior

Provenance and boundaries

Public means a public Roblox fact. Derived means a calculated estimate. Owner-confirmed requires an explicit owner grant; private retention, revenue, PTR/qPTR and acquisition data are not copied from other modules.

Unavailable and a dash do not mean zero. History begins with the first independent observation. CCU is concurrent players; Visits is accumulated entries, not DAU.

Server count covers only a complete list of public root-place servers, excluding private/reserved. The sample does not cover all Roblox.

Metric formulas and limitations

The chart heading contains the metric name, guide icon and AI advice button. Descriptions and formulas are available here through the book icon; they are not repeated below AI advice, including country, language and player-activity breakdowns.

Public score

derived. Formula: weighted mean(available group percentiles) − risk / 100 × maxRiskPenalty; without groups: weighted mean(clamp(log1p(CCU)/log1p(1000)*100), Wilson rating) − penalty. Example: See the formula for inputs and units.

Higher is a stronger signal. Below 3 groups or 40% weight is provisional; without a cohort confidence ≤20, activity is not growth.

Data confidence

derived. Formula: min(data completeness score, available weight) × cohort confidence. Example: See the formula for inputs and units.

Higher is a stronger signal. Completeness and comparability, not growth probability.

Risk

derived. Formula: min(100, sum(observed risk flags)). Example: See the formula for inputs and units.

Lower means fewer observed limitations. Not a claim of bots, manipulation or fraud.

Current CCU

public. Formula: playing (Roblox Games API). Example: 250 CCU means 250 concurrent players.

Current audience size. Not Visits or daily unique players.

Visits

public. Formula: visits (Roblox Games API). Example: 100,000 accumulated visits, including repeat visits.

Scale, not quality. Repeated visits count; not DAU.

Activity age

derived. Formula: (asOf − sustainedActivityAt) / 86400000; fallback firstPublicAt, then createdAt. Example: 10 days since observed sustained activity.

Context, not a score. Age basis shown; already-active discoveries are left-censored.

Universe age

derived. Formula: (asOf − createdAt) / 86400000. Example: Universe created 90 days ago.

Context. Not the public release date.

Observation age

derived. Formula: (asOf − firstPublicAt) / 86400000. Example: Two days of independent observations.

More history. Does not prove a recent launch.

Mean CCU · 1 hour

derived. Formula: Σ(CCUᵢ × Δtᵢ) / ΣΔtᵢ. Example: 100 CCU for 10 minutes and 200 for 20 minutes gives 166.7 CCU.

Higher is a stronger signal. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

CCU growth · 1 hour

derived. Formula: (mean(current window) / mean(previous equal window) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Zero base unavailable; baseline <25 is flagged.

New Visits · 1 hour

derived. Formula: Visits(end) − Visits(start). Example: 102000 - 100000 = 2000 new visits.

Higher is a stronger signal. Both boundaries required; counter reset means unavailable.

Mean CCU · 6 hours

derived. Formula: Σ(CCUᵢ × Δtᵢ) / ΣΔtᵢ. Example: 100 CCU for 10 minutes and 200 for 20 minutes gives 166.7 CCU.

Higher is a stronger signal. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

CCU growth · 6 hours

derived. Formula: (mean(current window) / mean(previous equal window) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Zero base unavailable; baseline <25 is flagged.

New Visits · 6 hours

derived. Formula: Visits(end) − Visits(start). Example: 102000 - 100000 = 2000 new visits.

Higher is a stronger signal. Both boundaries required; counter reset means unavailable.

Mean CCU · 24 hours

derived. Formula: Σ(CCUᵢ × Δtᵢ) / ΣΔtᵢ. Example: 100 CCU for 10 minutes and 200 for 20 minutes gives 166.7 CCU.

Higher is a stronger signal. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

CCU growth · 24 hours

derived. Formula: (mean(current window) / mean(previous equal window) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Zero base unavailable; baseline <25 is flagged.

New Visits · 24 hours

derived. Formula: Visits(end) − Visits(start). Example: 102000 - 100000 = 2000 new visits.

Higher is a stronger signal. Both boundaries required; counter reset means unavailable.

Mean CCU · 3 days

derived. Formula: Σ(CCUᵢ × Δtᵢ) / ΣΔtᵢ. Example: 100 CCU for 10 minutes and 200 for 20 minutes gives 166.7 CCU.

Higher is a stronger signal. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

CCU growth · 3 days

derived. Formula: (mean(current window) / mean(previous equal window) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Zero base unavailable; baseline <25 is flagged.

New Visits · 3 days

derived. Formula: Visits(end) − Visits(start). Example: 102000 - 100000 = 2000 new visits.

Higher is a stronger signal. Both boundaries required; counter reset means unavailable.

Mean CCU · 7 days

derived. Formula: Σ(CCUᵢ × Δtᵢ) / ΣΔtᵢ. Example: 100 CCU for 10 minutes and 200 for 20 minutes gives 166.7 CCU.

Higher is a stronger signal. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

CCU growth · 7 days

derived. Formula: (mean(current window) / mean(previous equal window) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Zero base unavailable; baseline <25 is flagged.

New Visits · 7 days

derived. Formula: Visits(end) − Visits(start). Example: 102000 - 100000 = 2000 new visits.

Higher is a stronger signal. Both boundaries required; counter reset means unavailable.

Mean CCU · 14 days

derived. Formula: Σ(CCUᵢ × Δtᵢ) / ΣΔtᵢ. Example: 100 CCU for 10 minutes and 200 for 20 minutes gives 166.7 CCU.

Higher is a stronger signal. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

CCU growth · 14 days

derived. Formula: (mean(current window) / mean(previous equal window) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Zero base unavailable; baseline <25 is flagged.

New Visits · 14 days

derived. Formula: Visits(end) − Visits(start). Example: 102000 - 100000 = 2000 new visits.

Higher is a stronger signal. Both boundaries required; counter reset means unavailable.

Mean CCU · 30 days

derived. Formula: Σ(CCUᵢ × Δtᵢ) / ΣΔtᵢ. Example: 100 CCU for 10 minutes and 200 for 20 minutes gives 166.7 CCU.

Higher is a stronger signal. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

CCU growth · 30 days

derived. Formula: (mean(current window) / mean(previous equal window) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Zero base unavailable; baseline <25 is flagged.

New Visits · 30 days

derived. Formula: Visits(end) − Visits(start). Example: 102000 - 100000 = 2000 new visits.

Higher is a stronger signal. Both boundaries required; counter reset means unavailable.

Early impulse

derived. Formula: meanCCU24h / activityAgeDays. Example: 250 / 5 = 50 CCU per age-day.

Higher is a stronger signal. Unavailable before a full day; age-matched cohorts.

CCU per 100k Visits

derived. Formula: meanCCU24h / totalVisits × 100000. Example: 200 / 500000 * 100000 = 40.

Higher is a stronger signal. Not retention. Age and visit-size normalization.

Estimated session

derived. Formula: meanCCU24h × 1440 / ΔVisits24h. Example: 100 * 1440 / 12000 = 12 minutes.

Higher is a stronger signal. Estimate only; ≥100 new visits, 0.1–240 minutes, ≥80% coverage; score needs confidence ≥50.

Log trend · 7d

derived. Formula: OLS slope(log1p(CCU), time in days). Example: A positive log-slope indicates an upward trend.

Higher is a stronger signal. Not a future forecast; coverage required.

Growth acceleration

derived. Formula: growth24h(now) − growth24h(previous window). Example: 50% - 20% = 30 percentage points.

Higher is a stronger signal. Three complete windows required; low-base effects.

Stability · 24h

derived. Formula: meanCCU24h / maxCCU24h. Example: 100 / 200 = 0.5.

Higher is a stronger signal. High stability does not imply growth.

Volatility · 24h

derived. Formula: time-weighted standard deviation / meanCCU. Example: Standard deviation 20 / mean 100 = 0.2.

Lower is steadier; early growth can be volatile. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

Daily activity coverage

derived. Formula: P10(hourly mean CCU) / P90(hourly mean CCU). Example: Hourly P10 of 50 / P90 of 100 = 0.5.

Higher is a stronger signal. ≥20 complete hours required. Not geography.

Peak preservation · 24–48h

derived. Formula: meanCCU(24–48h after peak) / peakCCU. Example: 80 / 200 = 0.4 peak preservation.

Higher is a stronger signal. Only after the window matures; cause unknown.

Peak preservation · 72–96h

derived. Formula: meanCCU(72–96h after peak) / peakCCU. Example: 80 / 200 = 0.4 peak preservation.

Higher is a stronger signal. Only after the window matures; cause unknown.

Peak preservation · 7–8d

derived. Formula: meanCCU(7–8d after peak) / peakCCU. Example: 80 / 200 = 0.4 peak preservation.

Higher is a stronger signal. Only after the window matures; cause unknown.

Roblox favorites

public. Formula: favoritedCount (Roblox Games API). Example: 500 public favorites.

Size. Public cumulative counter.

Favorites / 1000 Visits

derived. Formula: ΔFavorites24h / ΔVisits24h × 1000. Example: 20 new favorites / 2000 new visits * 1000 = 10.

Higher is a stronger signal. ≥100 new visits; not a unique-player conversion.

Upvotes / 1000 Visits

derived. Formula: ΔUpvotes24h / ΔVisits24h × 1000. Example: 10 / 2000 * 1000 = 5.

Higher is a stronger signal. Votes can be withdrawn; negative counters are excluded.

Rating

public. Formula: up / (up + down) × 100. Example: 90 / (90 + 10) = 90%.

Higher is better. Consider sample size; scoring uses Wilson.

Rating Wilson lower bound

derived. Formula: 95% Wilson lower confidence bound, z = 1.96. Example: 90 positive votes out of 100 gives a lower bound of about 82.6%.

Higher is a stronger signal. Conservative positive-vote share, not game quality.

Text originality

derived. Formula: 1 − max(Jaccard(name + description, peers)). Example: Maximum text similarity 0.3 gives originality 0.7.

Higher is a stronger signal. ≥10 games; ≥3 shared meaningful words; not a plagiarism finding.

Genre opportunity

derived. Formula: tanh(genreGrowth7d/100) + growingShare − cloneDensity − top3Share. Example: A growing niche with lower concentration receives a higher index.

Higher is a stronger signal. Observed sample only; ≥10 games with week history.

Top 3 concentration

derived. Formula: Σ top3 meanCCU24h / Σ genre meanCCU24h. Example: 600 / 1000 = 0.6.

Lower concentration. Not all Roblox.

Genre growth · 7d

derived. Formula: (Σ comparable mean7d / Σ previous mean7d − 1) × 100. Example: 1200 / 1000 - 1 = 20%.

Higher is a stronger signal. Comparable fixed set, not the entire platform.

After update · 24h

derived. Formula: (meanCCU(24h after observed update) / meanCCU(7d before) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Temporal association, not causation; updatedAt is not a full release log.

After update · 72h

derived. Formula: (meanCCU(72h after observed update) / meanCCU(7d before) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Temporal association, not causation; updatedAt is not a full release log.

After update · 7d

derived. Formula: (meanCCU(7d after observed update) / meanCCU(7d before) − 1) × 100. Example: 150 / 100 - 1 = 50%.

Higher is a stronger signal. Temporal association, not causation; updatedAt is not a full release log.

Observed updates · 30d

derived. Formula: count(observed updatedAt changes in 30d). Example: Three observed updatedAt changes.

Context. Changes before first observation are unknown.

Days since update

derived. Formula: (asOf − updatedAt) / 86400000. Example: Two days since the reported update.

Context. Roblox timestamp, not proof of significant content.

Run ≥100 CCU

derived. Formula: longest consecutive observed interval with CCU ≥100. Example: 36 consecutive observed hours above 100 CCU.

Higher is a stronger signal. Gaps break runs; milestones require ≥3 hours.

Public servers

public. Formula: count(complete Public root-place server list). Example: 62 servers in a complete public root-place list.

Context. Partial lists unavailable; excludes private/reserved.

Public server occupancy

derived. Formula: Σ players in public root servers / Σ capacity ×100. Example: 50 players / 100 available slots = 50%.

Higher is a stronger signal. Complete public root-place list only, not global CCU.

24h completeness

derived. Formula: Σ accepted sample intervals / 24h ×100. Example: 20 observed hours / 24 = 83.3%.

Higher is a stronger signal. Intervals need two valid bounding samples.

MEDIAN CCU · 1 hour

derived. Formula: median(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MIN CCU · 1 hour

derived. Formula: min(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MAX CCU · 1 hour

derived. Formula: max(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P10 CCU · 1 hour

derived. Formula: p10(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P25 CCU · 1 hour

derived. Formula: p25(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P75 CCU · 1 hour

derived. Formula: p75(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P90 CCU · 1 hour

derived. Formula: p90(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MEDIAN CCU · 6 hours

derived. Formula: median(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MIN CCU · 6 hours

derived. Formula: min(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MAX CCU · 6 hours

derived. Formula: max(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P10 CCU · 6 hours

derived. Formula: p10(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P25 CCU · 6 hours

derived. Formula: p25(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P75 CCU · 6 hours

derived. Formula: p75(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P90 CCU · 6 hours

derived. Formula: p90(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MEDIAN CCU · 24 hours

derived. Formula: median(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MIN CCU · 24 hours

derived. Formula: min(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MAX CCU · 24 hours

derived. Formula: max(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P10 CCU · 24 hours

derived. Formula: p10(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P25 CCU · 24 hours

derived. Formula: p25(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P75 CCU · 24 hours

derived. Formula: p75(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P90 CCU · 24 hours

derived. Formula: p90(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MEDIAN CCU · 3 days

derived. Formula: median(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MIN CCU · 3 days

derived. Formula: min(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MAX CCU · 3 days

derived. Formula: max(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P10 CCU · 3 days

derived. Formula: p10(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P25 CCU · 3 days

derived. Formula: p25(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P75 CCU · 3 days

derived. Formula: p75(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P90 CCU · 3 days

derived. Formula: p90(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MEDIAN CCU · 7 days

derived. Formula: median(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MIN CCU · 7 days

derived. Formula: min(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MAX CCU · 7 days

derived. Formula: max(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P10 CCU · 7 days

derived. Formula: p10(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P25 CCU · 7 days

derived. Formula: p25(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P75 CCU · 7 days

derived. Formula: p75(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P90 CCU · 7 days

derived. Formula: p90(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MEDIAN CCU · 14 days

derived. Formula: median(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MIN CCU · 14 days

derived. Formula: min(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MAX CCU · 14 days

derived. Formula: max(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P10 CCU · 14 days

derived. Formula: p10(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P25 CCU · 14 days

derived. Formula: p25(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P75 CCU · 14 days

derived. Formula: p75(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P90 CCU · 14 days

derived. Formula: p90(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MEDIAN CCU · 30 days

derived. Formula: median(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MIN CCU · 30 days

derived. Formula: min(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

MAX CCU · 30 days

derived. Formula: max(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P10 CCU · 30 days

derived. Formula: p10(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P25 CCU · 30 days

derived. Formula: p25(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P75 CCU · 30 days

derived. Formula: p75(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

P90 CCU · 30 days

derived. Formula: p90(CCU weighted by observed duration). Example: See the formula for inputs and units.

Distribution context. Own observations only; ≥80% time coverage; gaps >20 minutes are not filled.

update effect · Ccu · 1d

derived. Formula: (meanCCU(1d after) / meanCCU(7d before) − 1) × 100. Example: See the formula for inputs and units.