Power & politics at work: a step-by-step HR playbook (with scripts)

Most HR training treats politics as something to deplore. That’s a luxury position. In real organizations, politics is the residue of three structural facts: goals are interdependent, resources are scarce, and authority is ambiguous. Your job isn’t to wish that away — it’s to channel power so the work gets done and the people who do the work aren’t demoralized.

What workplace politics actually is

Jeffrey Pfeffer’s working definition (Managing With Power, 1992): ‘the processes, the actions, the behaviors through which potential power is utilized and realized.’ Politics is downstream of power. If you want to change the politics, you have to change who holds power, on what basis, and over what.

Step 1 — Build the power map

Before any intervention, write down the power map. For each player in the situation (usually 4–8 people), score their power on each base from 0 (none) to 3 (dominant). Use French & Raven’s five plus informational power.

The power map almost always reveals the surprise: the formal org chart and the real power chart are not the same. The CTO might be legitimately powerful but referent-weak; the principal engineer who joined four years ago might be expert + referent + informational, which is a stronger combination than the CTO’s title alone.

Step 2 — Name the political game

Most political behavior fits a small number of recurring patterns. Naming the game in writing — to yourself first, then to the sponsor — converts a vague ‘there’s a lot of politics on this team’ into a diagnosable problem.

Step 3 — Align the executive sponsor (script)

You cannot run a political intervention without an executive sponsor who understands the diagnosis and is willing to spend political capital. Misalignment here is the #1 reason HR interventions fail.

If the sponsor will not commit to both, you don’t have a sponsor. Stop the intervention. Document the conversation. Revisit when conditions change.

Step 4 — Confront the toxic influencer (script)

Toxic influencers are usually high-expert + high-referent + high-informational. Frontal attacks fail because their referent power converts colleagues into defenders. The conversation must be specific, behavioral, and tied to consequences the person actually cares about (which is rarely money — it’s status and access).

Three rules for this conversation: (a) describe behavior not character — never ‘you’re political,’ always ‘this specific action had this specific effect’; (b) end every item with what changes, not just what’s wrong; (c) leave time for pushback — the goal is behavior change, not a confession.

Step 5 — Re-arm the demoralized manager (script)

The manager who has been outflanked by an influencer is usually in one of two states: learned helplessness (‘nothing I do matters’) or over-correction (‘I’m going to fire them’). Neither is the right state. Your job is to restore their sense of agency and give them three concrete moves.

Step 6 — Signal to the bystanders (script)

The most underweighted audience in a political intervention is the bystanders — the 80% of the team watching to see whether HR will actually act. If you intervene invisibly, they assume nothing happened, and the next person who games the system gets the same green light.

You don’t need to publish the consequences. You do need to publish the standards. Use the next all-hands or team meeting to name — without names — the standard you’re holding everyone to.

Step 7 — Fix the structural conditions

If you don’t change structure, the same politics re-emerge with a different cast in 6–12 months. The four structural levers, in order of leverage:

1. Decision rights — write a RACI for the 5–10 most contested decision types on the team. Publish it.
2. Information flow — default-open docs, shared dashboards, recorded decisions with named owners. Information hoarding stops paying.
3. Reward criteria — make promotion criteria public and audit the last three promotion cycles for ‘visibility’ vs ‘contribution.’ If the political players got promoted, you’ve been training the behavior.
4. Comp transparency — at minimum, publish bands. Opacity rewards backchannel negotiation, which is itself a political game.

Three worked scenarios

Scenario A — The principal engineer who runs a shadow org

Power profile: expert 3, referent 3, informational 3, legitimate 1. Game: gatekeeping + coalition-building. Sponsor move: CTO publicly delegates architecture decisions to the EM, not the principal. HR move: Step 4 conversation focused on pipeline access and pre-meetings. Structural fix: published architecture decision records with named decider; the principal is a contributor, not the decider.

Scenario B — The director who keeps absorbing scope

Power profile: legitimate 3, reward 2, coercive 2. Game: empire-building. Sponsor move: CEO names a scope boundary and refuses headcount increases inside the contested zone. HR move: re-write the director’s scorecard around depth metrics (retention, promo rates from within) not breadth (number of functions reporting in). Structural fix: org-design review every 6 months with explicit ‘what should NOT be in this org’ section.

Scenario C — The PM who shifts blame in writing

Power profile: informational 3, referent 2, legitimate 2. Game: blame-shifting via documentation. Sponsor move: VP Product stops accepting post-hoc memos as evidence. HR move: introduce pre-mortems with named owners — the same person who would later write the blame memo now has to sign the pre-mortem. Structural fix: decision log with two columns: ‘who made this call’ and ‘who was consulted.’