PEAKTEAMS open item bank

Every performance item is answered on a 1 to 5 agreement scale: Strongly disagree, Disagree, Neutral, Agree, Strongly agree.

Items flagged reverseScored are flipped before aggregation: score = 6 - answer (1 becomes 5, 2 becomes 4, 3 stays 3).

Item mean = arithmetic mean of all team members' answers to the item after reverse scoring. Standard deviation is computed on the same 1 to 5 scale.

KPI score (1 to 5) = weighted mean of the item means mapped to the KPI, with the item weight (1 unless stated; the four bureaucracy items of Friction & Consumption carry 0.5 each).

KPI score = Σ(weight_i × item mean_i) / Σ(weight_i)

A KPI with three items therefore counts exactly as much as a KPI with two, and the Deep-Dive's twelve or thirteen items per KPI do not outweigh the Snapshot's two or three.

The 18 KPIs sit in four pillars of the input-process-output model. A pillar score is the mean of its KPI means: one KPI, one vote.

Input · 5 KPIs

Purpose & Objectives, Values & Vision, Environment & Resources, Team Organization, Leadership Principles

Mean of the 5 KPI means in the pillar (one KPI = one vote, however many items it has).

Process · 10 KPIs

Communication Culture, Trust & Safety, Social Cohesion, Delivery Methodology, Conflict Navigation, Feedback & Development, Innovation & Growth, Motivation & Focus, Identification & Engagement, Inner Drive

Mean of the 10 KPI means in the pillar (one KPI = one vote, however many items it has).

Output · 2 KPIs

Team Output, Organizational Impact

Mean of the 2 KPI means in the pillar (one KPI = one vote, however many items it has).

Friction · 1 KPI

Friction & Consumption

Friction is one KPI, so the pillar score is that KPI mean (1 to 5, higher = more friction).

Overall performance score (1 to 5) = mean of the 17 value-creating KPI means (Friction & Consumption excluded).

Team Net Value treats friction as a tax on the value a team creates: it multiplies rather than subtracts, so a team creating little value cannot look average just because friction is low, and maximal friction can take a team to zero. Exactly as implemented:

Value Score = (Input + Process + Output) / 3
Value Points = (Value Score - 1) / 4 x 100
Friction Load = (Friction - 1) / 4
Friction Tax = Friction Load ^ 2
Team Net Value = Value Points x (1 - Friction Tax), rounded to an integer and clamped to 0..100

The tax is convex (exponent 2): friction at “Neutral” (3.0) costs 25% of the value, at “Agree” (4.0) 56%, at “Strongly agree” (5.0) 100%.

Bands: Critical 0–20 · Poor 20–40 · Average 40–60 · Good 60–80 · Excellent 80–100. The band labels are provisional until benchmark data allows calibrating them.

Friction & Consumption is the one negatively keyed KPI: a higher score means more friction. Its items are never reverse-scored into a "good" direction; instead the KPI mean enters Team Net Value as a tax on the value the team creates. Friction items are excluded from the overall score, from strengths and from growth opportunities; in the consensus matrix their quadrant is assigned on the flipped score (6 - mean) while the displayed mean stays as answered. The item "Our meetings are really creating value." is positively phrased inside this KPI and is therefore reverse-scored.

Each identity item shows two statements on a 5-point bipolar scale: 1 = fully the left statement, 3 = centre, 5 = fully the right statement. The team mean per item is rescaled to -5..+5 via score = (mean - 3) x 2.5 (implemented as the 1..10 intermediate (mean - 1) x 2.25 + 1, then (x - 5.5) x 10/9, which is the same line), so 1 maps to -5, 3 to 0 and 5 to +5. For items whose poles are swapped (leftMapsTo equals the dimension's right pole code) the sign is flipped.

Dimension score (-5..+5) = mean of its three sign-corrected item scores (weight 1 each). Negative values point to the left pole, positive values to the right pole.

The four-letter team identity code takes, per dimension, the left pole code when the score is below 0 and the right pole code otherwise (an exact 0 resolves to the right pole by convention and is flagged as leaning). Four dimensions with two poles each give 16 identities.

item score     = (team mean − 3) × 2.5          // −5 … +5, sign flipped when poles are swapped
dimension      = mean(item scores)              // 3 items, weight 1
pole           = score < 0 ? left : right
leans          = |score| < 1.0
confidence     = max(0, 1 − sd(respondent scores) / 5)

A dimension with |score| below 1.0 is reported as "leans" towards its pole rather than as a settled trait.

Confidence (0..1) = max(0, 1 - sd / 5), where sd is the standard deviation of the individual respondents' sign-corrected scores on the -5..+5 scale (with a single respondent, the spread of the three item scores is used instead).

Team Cohesion (0..100) = (1 - min(mean item variance / 4, 1)) x 100, where the variance (population variance) is computed per performance item across all respondents and averaged over the items with at least two answers; 4 is the maximum variance on a 1 to 5 scale (half the team at 1, half at 5). Identity items are excluded.

Team Cohesion = (1 − min(mean(variance_i) / 4, 1)) × 100

100 = every member gave the same answer to every item; 0 = maximal disagreement.

Per item, agreement is "high" when the standard deviation is below 0.8, "low" at or above 1.2, "medium" in between. The consensus matrix uses a score cut of 3.5 and an agreement cut of 1.

At item level, a mean of 4.0 or above is reported as a strength, below 3.0 as a growth opportunity and below 2.5 as a critical challenge.

Results appear only once at least five team members have answered; the owner's own answer does not count. Filters that would isolate fewer respondents are refused, so no individual answer can be read back from a team result.

With 8 or fewer responses one person can move an item's standard deviation across the agreement cut, so agreement statistics are shown as hints rather than verdicts for small teams.