Capability template for people and HR
Culture health check
A founder or people lead sees whether the company's stated values show up in daily work: each value written as behaviours you could watch for, read against the team's own evidence, with where practice matches, where it drifts, and two or three rituals to try. Reported by team and theme, never about a person.
The outcome
Leadership stops guessing whether the values on the website still describe the company. Each value becomes two or three things a new starter could actually notice in an ordinary week, and each of those is set against what the company already knows: the staff survey, what leavers said on the way out, how the last few big decisions were really made, and what people get thanked for. Every behaviour gets a plain reading (matches, mixed, drifts, or not enough evidence) with the numbered evidence behind it, so nobody has to take the verdict on trust.
The check names the patterns that quietly wear a culture down, such as decisions that reach a team late and without reasons, but only where two separate pieces of evidence point the same way. It then proposes two or three rituals sized for the company, each tied to the drift it answers, and lists the questions only leadership can settle.
It is about the company, never about a person. Results are shown by team and theme; small teams are merged so nobody's answer can be picked out; names are removed from every note. Leadership decides what changes and what staff hear.
Who starts the work, and when
The founder, the people lead or the head of operations runs it after a staff survey closes, before an annual look at the values, after a merger or a spell of fast hiring, or when exit conversations keep circling the same complaint. It runs again after the next survey to see what moved. The founders or the leadership team read it, confirm or reword the behaviours, and choose what to try.
What they supply
- The stated values, with the description the company uses for each, and any behaviours already written down.
- The staff survey export, counts by team and question, with the question wording, and last time's results if there are any.
- Exit notes, one per leaver, ideally with names already taken out.
- Notes on recent decisions that changed how people work: who proposed, who was asked, who decided and how it was told.
- What people were recognised for, as counts by theme.
- A few settings: headcount, the rituals already in place, the smallest group the company is willing to report, and how high a survey score has to be to count as a match.
Any one source of evidence is enough to start; more sources make the readings firmer.
What comes out
One culture health check, ready for leadership:
- the evidence, numbered, so every reading can be traced;
- each value written as behaviours, with drafted ones marked for leadership to confirm;
- where practice matches and where it drifts, behaviour by behaviour, with any team that sits well below the company;
- a view by team with small teams merged, and what changed since the last survey, the largest fall first;
- the anti-patterns the evidence shows, the ones with only one sign, and the ones not seen;
- two or three rituals to try, each with what it answers, who runs it and what to watch next time;
- the questions for leadership, and what was left out and why.
Nothing is shared with staff, sent or posted.
An invented example
Illustrative example with invented data. The company, the people, the client, the survey and every figure below are invented. Harrowfinch Studio is an invented 64-person design and web agency with four teams. Its people lead runs a culture health check after the October staff survey, for the studio's two founders. First come the inputs the people lead supplied, in full, then the check the capability returned.
The inputs
1. Settings
Company: Harrowfinch Studio, a design and web agency
Headcount: 64, in four teams: Design, Engineering, Client Services, Operations
Today: Tuesday 10 November 2026
Period: October 2026 survey; exit notes and decisions from April to
October 2026; recognition from July to September 2026
Run by: People lead
Leadership: the two founders
Rituals already in place: monthly all-hands with a founders' update; an
onboarding buddy for each new starter; a weekly meeting in each team;
a shout-outs channel
Last survey's results shared with staff: yes, a company summary went to
all staff in May 2026
Smallest group, thresholds, team gap, notable change: use the defaults
2. The stated values (from the staff handbook)
Candour: We say what we see, early.
Ownership: We see things through.
Craft: We make work we are proud of.
No behaviours written yet.
3. The survey questions
Q1 I can tell my manager about a problem without worrying how it will land.
Q2 I know what the studio is aiming for this quarter and how my work fits in.
Q3 When a decision affects my work, I hear why it was made.
Q4 When work crosses into another team, someone still sees it through to the end.
Q5 Work is checked properly before a client sees it.
Q6 I would tell a friend this is a good place to work.
4. The survey export, October 2026
period,team,question,respondents,agree,neutral,disagree
2026-10,Design,Q1,20,15,3,2
2026-10,Design,Q2,20,12,5,3
2026-10,Design,Q3,20,9,6,5
2026-10,Design,Q4,20,14,4,2
2026-10,Design,Q5,20,17,2,1
2026-10,Design,Q6,20,14,4,2
2026-10,Engineering,Q1,16,13,2,1
2026-10,Engineering,Q2,16,9,4,3
2026-10,Engineering,Q3,16,7,4,5
2026-10,Engineering,Q4,16,7,5,4
2026-10,Engineering,Q5,16,13,2,1
2026-10,Engineering,Q6,16,11,3,2
2026-10,Client Services,Q1,12,7,3,2
2026-10,Client Services,Q2,12,9,2,1
2026-10,Client Services,Q3,12,5,4,3
2026-10,Client Services,Q4,12,6,3,3
2026-10,Client Services,Q5,12,9,2,1
2026-10,Client Services,Q6,12,7,3,2
2026-10,Operations,Q1,4,2,1,1
2026-10,Operations,Q2,4,3,1,0
2026-10,Operations,Q3,4,2,1,1
2026-10,Operations,Q4,4,3,0,1
2026-10,Operations,Q5,4,3,1,0
2026-10,Operations,Q6,4,2,2,0
5. The last survey's export, April 2026
period,team,question,respondents,agree,neutral,disagree
2026-04,All teams,Q1,48,36,8,4
2026-04,All teams,Q2,48,29,13,6
2026-04,All teams,Q3,48,31,11,6
2026-04,All teams,Q4,48,28,12,8
2026-04,All teams,Q5,48,39,6,3
2026-04,All teams,Q6,48,33,10,5
6. Exit notes, April to October 2026 (typed by the people lead)
1. Engineering, left July. Priorities changed after the June team change
and nobody explained why. Valued having work reviewed before clients
saw it.
2. Client Services, left May. Felt that raising client problems with
their lead got them seen as negative. Wanted a clearer path to
promotion.
3. Design, left June. Moving to another city. Praised the studio's
quality bar.
4. Engineering, left August. Support tickets bounced between teams with
nobody owning them. Heard about the June team change from a client
before hearing it from the founders.
5. Client Services, left September. Raised a client issue and was told
by their lead, Wren Kafton, to "keep it positive". Said pay was below
what other agencies offered.
6. Engineering, left October. Big decisions were made by the founders
in private and the reasons were never clear.
7. Decision notes (typed by the people lead)
D1. June 2026. Engineering reorganised into two groups by product line.
Proposed by the founders. No one outside the founders was asked.
Decided by the two founders in one leadership meeting. Announced to
all staff in a message on the day it took effect; the message gave
no reasons.
D2. August 2026. A peer review step added before any client delivery.
Proposed by the design leads. Discussed with Design and Engineering
in a joint team meeting. Decided by the head of design two weeks
later. Reasons written up and given to Design and Engineering in
their team meeting before the step began.
D3. September 2026. The studio stopped work for one client after
repeated scope disputes. Raised by a Client Services account lead at
the weekly meeting the week after the second dispute. The founders
asked Client Services for the history and decided within a week.
Client Services was told in its team meeting, with reasons.
Engineering was not told and found out when that client's tickets
stopped.
D4. October 2026. Who picks up out-of-hours support requests between
Client Services and Engineering. Raised by both team leads.
Discussed by both teams in two meetings. No owner named; still open.
8. Recognition notes (shout-outs channel, July to September 2026)
9 shout-outs in all.
6 thanked someone for working late or over a weekend.
2 passed on praise from a client.
1 thanked someone for helping another team with a problem.
The output
CULTURE HEALTH CHECK: Harrowfinch Studio, October 2026 survey
For: People lead. Leadership decides: the two founders. Prepared Tuesday 10 November 2026.
Read from: 3 values; survey export (24 rows, 52 respondents, 4 teams); last survey (6 rows, 48 respondents); exit notes (6 leavers); decision notes (4); recognition notes (9 recognitions); no survey comments. Settings: defaults used for the smallest group (5), the match threshold (70%), the drift threshold (50%), the team gap (10 points) and the notable change (10 points).
Behaviours: 6. Matches: 1. Mixed: 2. Drifts: 2. Not enough evidence: 1. Anti-patterns spotted: 2. Rituals to try: 3. Questions for leadership: 5.
THE EVIDENCE
S1. Q1 "I can tell my manager about a problem without worrying how it will land." Company: 37 of 52 agree (71%). Last survey: 75%.
S2. Q2 "I know what the studio is aiming for this quarter and how my work fits in." Company: 33 of 52 agree (63%). Last survey: 60%.
S3. Q3 "When a decision affects my work, I hear why it was made." Company: 23 of 52 agree (44%). Last survey: 65%.
S4. Q4 "When work crosses into another team, someone still sees it through to the end." Company: 30 of 52 agree (58%). Last survey: 58%.
S5. Q5 "Work is checked properly before a client sees it." Company: 42 of 52 agree (81%). Last survey: 81%.
S6. Q6 "I would tell a friend this is a good place to work." Company: 34 of 52 agree (65%). Last survey: 69%.
X1. Exit theme: decisions, and the reasons for them, not explained to the people they affect. Raised by 3 of 6 leavers. Counts. Concern.
X2. Exit theme: raising a problem got people seen as negative. Raised by 2 of 6 leavers. Counts. Concern.
X3. Exit theme: work moving between teams with no one owning it. Raised by 1 of 6 leavers. Noted only. Concern.
X4. Exit theme: the quality bar and the review of work before clients see it. Raised by 2 of 6 leavers. Counts. Valued.
X5. Exit theme: no clear path to promotion. Raised by 1 of 6 leavers. Noted only. Concern; outside the stated values.
X6. Exit theme: pay below other agencies. Raised by 1 of 6 leavers. Noted only. Concern; outside the stated values.
C. Comment themes: none supplied.
D1. June 2026: Engineering reorganised into two groups by product line. Proposed by: the founders. Asked: no one outside the deciders. Decided by: the two founders, in one leadership meeting. Told: all staff, by message on the day it took effect, with no reasons given.
D2. August 2026: a peer review step added before any client delivery. Proposed by: the design leads. Asked: Design and Engineering, in a joint team meeting. Decided by: the head of design, two weeks later. Told: Design and Engineering, in their team meeting before the step began, with the reasons written up.
D3. September 2026: the studio stopped work for one client after repeated scope disputes. Proposed by: a Client Services account lead, at the weekly meeting the week after the second dispute. Asked: Client Services. Decided by: the two founders, within a week. Told: Client Services, in its team meeting, with reasons; Engineering was not told and found out when the client's tickets stopped.
D4. October 2026: who picks up out-of-hours support requests between Client Services and Engineering. Proposed by: both team leads. Asked: both teams, in two meetings. Decided by: no one: still open. Told: not applicable.
R1. Recognition: 6 of 9 were for working late or over a weekend.
R2. Recognition: 2 of 9 were for passing on praise from a client.
R3. Recognition: 1 of 9 was for helping another team with a problem.
VALUES AS BEHAVIOURS
Candour: "We say what we see, early."
B1. Looks like: a problem on a project is raised at the team's weekly meeting in the week it is spotted. (draft for leadership to confirm)
B2. Looks like: when leadership makes a decision that changes a team's work, that team hears the reason before the change takes effect. (draft for leadership to confirm)
Does not look like: a concern voiced only in private after the meeting, or softened until no one acts on it.
Ownership: "We see things through."
B3. Looks like: work passed to another team has a named owner until it is closed. (draft for leadership to confirm)
B4. Looks like: a team that spots a problem outside its own area follows it up until someone has it in hand. (draft for leadership to confirm)
Does not look like: "not our team" hand-offs that leave a client or a colleague waiting.
Craft: "We make work we are proud of."
B5. Looks like: client work is checked by a peer before the client sees it. (draft for leadership to confirm)
B6. Looks like: when a deadline puts quality at risk, the team tells the client lead before the work is sent. (draft for leadership to confirm)
Does not look like: skipping the check to hit a date without telling anyone.
WHERE PRACTICE MATCHES AND WHERE IT DRIFTS
B1. Problems raised in the week they are spotted. Reading: Mixed.
For: S1: 71% agree they can tell their manager about a problem, at or above 70%.
Against: X2: 2 of 6 leavers said raising a problem got them seen as negative.
Noted, not counted: none.
Team gap: Client Services and Operations (combined), 56% against company 71%, on S1.
B2. Teams hear the reason for decisions that change their work. Reading: Drifts.
For: D2: the reasons for the review step were written up and given to both teams before it began. D3: Client Services was told, with reasons, that work for the client had stopped.
Against: S3: 44% agree they hear why a decision affecting their work was made, below 50%. X1: 3 of 6 leavers said decisions and their reasons were not explained. D1: the June team change was announced on the day it took effect, with no reasons. D3: Engineering was not told the client work had stopped and found out when the tickets stopped.
Noted, not counted: none.
Team gap: none.
B3. Work passed to another team keeps a named owner. Reading: Drifts.
For: none.
Against: D4: who picks up out-of-hours support requests between Client Services and Engineering was discussed twice and still has no owner.
Noted, not counted: S4: 58% agree, between the thresholds. X3: one leaver said work moved between teams with no one owning it.
Team gap: Engineering, 44% against company 58%, on S4.
B4. Problems outside a team's own area are followed up. Reading: Mixed.
For: none.
Against: none.
Noted, not counted: S4: 58% agree, between the thresholds. R3: one recognition, fewer than two, for helping another team with a problem.
Team gap: Engineering, 44% against company 58%, on S4.
B5. Client work is checked by a peer first. Reading: Matches.
For: S5: 81% agree work is checked properly before a client sees it, at or above 70%. X4: 2 of 6 leavers valued the quality bar and the review of work. D2: a peer review step before client delivery was proposed by the design leads, discussed with both teams it affects, and adopted.
Against: none.
Noted, not counted: none.
Team gap: none.
B6. Quality risks raised with the client lead before work is sent. Reading: Not enough evidence.
For: none.
Against: none.
Noted, not counted: R1: 6 of 9 recognitions were for working late or over a weekend, which may sit alongside this behaviour but does not describe it.
Team gap: none.
BY TEAM
Groups reported: Design (20 respondents), Engineering (16), Client Services and Operations (combined, 16). Combined: Client Services and Operations, because Operations had 4 respondents, below 5.
S1: company 71%; Design 75%; Engineering 81%; Client Services and Operations 56%. Gap: Client Services and Operations, 15 points below.
S2: company 63%; Design 60%; Engineering 56%; Client Services and Operations 75%. No gap.
S3: company 44%; Design 45%; Engineering 44%; Client Services and Operations 44%. No gap.
S4: company 58%; Design 70%; Engineering 44%; Client Services and Operations 56%. Gap: Engineering, 14 points below.
S5: company 81%; Design 85%; Engineering 81%; Client Services and Operations 75%. No gap.
S6: company 65%; Design 70%; Engineering 69%; Client Services and Operations 56%. No gap.
CHANGE SINCE THE LAST SURVEY
S3: 65% to 44% (-21 points). Notable fall.
S1: 75% to 71% (-4 points)
S2: 60% to 63% (+3 points)
S4: 58% to 58% (0 points)
S5: 81% to 81% (0 points)
S6: 69% to 65% (-4 points)
OTHER SURVEY RESULTS
S2: 63%, +3 points since the last survey. Bears on none of the stated values.
S6: 65%, -4 points since the last survey. Bears on none of the stated values.
ANTI-PATTERNS
Spotted:
1. Decisions out of sight. Evidence: D1, D3, X1, S3. Here: the June team change was decided with no one outside the founders asked and announced without reasons; Engineering learned second-hand that client work had stopped; 3 of 6 leavers raised it; S3 is at 44%, down 21 points.
2. Bad news unwelcome. Evidence: S1, X2. Here: one reported group is at 56% on S1, below 70%. Across the whole company, 2 of 6 leavers said raising a problem got them seen as negative.
One sign only: Values in name only: R1: 6 of 9 recognitions were for working late or over a weekend, which no stated value names.
Not seen: Culture left to spread by itself (onboarding buddies and a weekly meeting in each team are in place); Habits from a smaller company (no item shows it); Results never shared back (the April results went to all staff in May); Uneven standards (no item shows it).
RITUALS TO TRY
1. Decision notes in the open. Answers: B2; decisions out of sight. How often: for every decision that changes a team's work, posted before it takes effect. Run by: the founders. Takes: about ten minutes a decision. Watch: S3 at the next survey, and whether X1 comes up again in exit notes.
2. Problem first. Answers: B1; bad news unwelcome. How often: weekly, opening each team's existing meeting. Run by: each team's lead. Takes: about five minutes of each weekly meeting. Watch: S1 for each reported group at the next survey. Across the whole company, whether X2 comes up again in exit notes.
3. Hand-off owners. Answers: B3, B4. How often: weekly, inside each team's existing meeting. Run by: each team's lead. Takes: about five minutes of each weekly meeting. Watch: S4, including Engineering's share, and whether open hand-offs like D4 gain an owner.
QUESTIONS FOR LEADERSHIP
1. Confirm, reword or replace the six drafted behaviours, B1 to B6; none was supplied.
2. Which of the three rituals, if any, to try, and who owns each (B1 to B4; both anti-patterns spotted).
3. What staff are told from this check, and when (S1 to S6; the April results were shared in May).
4. Who owns out-of-hours support requests between Client Services and Engineering (D4), still open after two meetings.
5. Whether to add a survey question on raising quality risks before a deadline, since nothing supplied bears on B6.
LEFT OUT
Exit notes: one person's name removed; the theme is kept without it.
Exit notes: one leaver left to move city; that reason is not a culture theme.
Survey: Operations (4 respondents) combined with Client Services, the next smallest team.
Themes outside the stated values: X5 (promotion path) and X6 (pay), each from one leaver.
Survey comments: none supplied.
Prepared for the people lead. The two founders decide what changes and what staff are told. Nothing has been shared with staff, sent or posted, and no record has been changed.
Where it fits
- People and HR, which runs the survey, keeps the exit notes and protects the anonymity of both.
- Leadership, which owns the values and decides what the company changes because of them.
- Marketing and advertising, where agency culture is part of what clients buy and a tired team shows in the work.
- Nonprofits, where people join for the mission and leave when daily practice stops matching it.
- Professional services, where values are printed in every proposal and tested in every busy season.
- Software and technology, where teams split and reorganise often and habits from the first twenty people linger.
- Growth & scaling, when what everyone simply knew at fifteen people has to be made deliberate at sixty.
- Acquisitions & integration, when two companies with two sets of values start working as one.
What has to change for your company
Set the smallest group you will ever report; five is a careful floor and small, close-knit teams may want more. Decide the survey scores that count as a match and as a drift, and the gap between a team and the company worth showing. Map your survey questions to your values once so every run reads them the same way. Supply your own behaviours once leadership has confirmed them. Trim the ritual menu to what would suit your company and add rituals you already trust. Name who sees the check and who decides what staff are told. It runs in ChatGPT, Claude, Copilot or Gemini from the files you supply; it needs no connection to your survey tool or HR system.
What has been checked
Fully Powered AI wrote this capability template. It was editorially reviewed on 10 November 2026. It was tried on its invented example on 10 November 2026 in one run; the record is published with it, and lists the small edits made afterward that have not been run again. It has not yet been tested in a client environment. It has not yet been tested as a packaged version in ChatGPT, Claude, Copilot or Gemini.
Testing in your environment means running it on a past survey your leadership already discussed, with hard cases written in advance: a team of three, an exit note that names a manager, a theme raised by only one leaver, a request to say who answered. The people lead checks every share against the survey counts and every reading against its evidence before anyone relies on it. Pilot acceptance-test preparation sets those cases up, and what we test before we call an AI workflow ready describes the record we keep.
The full capability template
Everything above is the outline. The complete capability template is the working version: the roles, the inputs, the procedure, the exceptions and stop conditions, the output format, the reviewer checklist, the acceptance tests, the instructions for an AI assistant and what to customize for your company. It comes with a download and the link to its folder in Fully Powered’s public repository.
Free with an account. Sign in by email: no password and no card.
Next step
Bring a real piece of this work to a call. We’ll walk through what it takes to fit this capability to your rules, prove it on your work and put it in the hands of the roles that own it.