Influence Without Authority

---
name: influence-without-authority
description: Use when the user needs to move a person, team, or decision *without having direct authority* over them — driving an architecture change across teams, getting buy-in from a skeptical principal engineer, persuading an exec sponsor, mobilizing a coalition for a contentious initiative, or pushing back on a leader's decision without bruising the relationship. Trigger phrases include "how do I get [team] to do X without going to their manager", "I need to convince [person] to support this", "I have to push back on [exec] without burning the relationship", "build the coalition", "I'm a TPM/staff eng/IC and I need [team] to prioritize this". Distinct from `feedback-frameworks` (which delivers truth) and `coaching-mode` (which develops people). This is about *moving outcomes through people you don't manage*. Opinionated about legitimate vs. manipulative tactics.
---

# Influence Without Authority

Most senior leadership work happens without direct authority. TPMs, staff engineers, senior ICs, and PMs all routinely need to move outcomes through people they can't tell what to do. This skill covers the craft.

The skill is **opinionated about legitimacy**. Most "influence" literature shades into manipulation (manufactured urgency, fake reciprocity, social proof staging, anchoring tricks). Manipulation works in the short term and costs trust in the long term. The skill defaults to legitimate moves and refuses to teach manipulation as a shortcut.

## When to apply

Trigger this skill when the user needs to:
- Drive a cross-team architecture change owned by another team's lead.
- Get a skeptical principal engineer's blessing on an approach.
- Mobilize a coalition for a contentious initiative.
- Push back on an exec's decision without burning the relationship.
- Get a sister team to prioritize a dependency they're slow on.
- Convert a "no" into a "not now" or a conditional "yes."

Skip when:
- The user *does* have authority and is avoiding using it. (Sometimes the right move is the direct one.)
- The user is conflating influence with manipulation.
- The need is one-shot and the relationship is disposable. (Most influence work is iterated.)

## First: diagnose what you actually need

Before any move, force these questions:

1. **What outcome do you actually need?** Not "I want them to agree with me." What concretely needs to happen, by whom, by when?
2. **Whose decision is this, formally?** If it's actually theirs, your job is to make a strong case and accept their call. If it's actually shared, the negotiation is different.
3. **What's their current position, and why?** *Position* is what they've said. *Interest* is what they actually need. The two often differ; addressing the position without the interest produces fake yeses.
4. **What's their BATNA — and yours?** What happens if no agreement is reached? Who has more to lose from inaction?
5. **What's your relationship's current account balance?** Influence runs on accumulated trust. If the account is overdrawn, no tactic compensates.

If the answers are unclear, work them out before moving. Half the failed influence attempts come from skipping this diagnosis.

## The currencies model

Influence runs on *exchange*. Cohen and Bradford's framework: people respond to currencies they value. Five families of currency, in rough order of leverage at most companies:

### Inspiration / vision currencies
- A clear strategic narrative they can rally behind
- Connection to organizational mission they care about
- Excellence and craft (especially for senior ICs)

### Task currencies
- Resources you can route their way
- Information they need (often the most underused)
- Help on something hard
- Reduced friction on their critical path

### Position currencies
- Recognition and visibility
- Reputation enhancement
- Insiderness (early access to something)
- Contacts with people they want access to

### Relationship currencies
- Personal support
- Acceptance / inclusion
- Trust that you have their back

### Personal currencies
- Gratitude
- Ownership / involvement
- Self-concept (helping them see themselves the way they want to)

When designing an ask, think about: *what currency would matter to this person, and which can I credibly offer?* Influence is mutual exchange, not a one-way push.

**Important:** Currencies are not chits. You don't manufacture them. The currency you offer must be real and the value you assign must be the value *they* place on it, not what you assume.

## The legitimacy line

Cialdini's six principles (reciprocity, scarcity, authority, social proof, liking, commitment/consistency) are real psychological levers. Each has a legitimate use and a manipulative one. The skill flags both:

| Principle | Legitimate use | Manipulative use |
|---|---|---|
| **Reciprocity** | Giving genuinely without strings; trusting the longer cycle | Strategic favors with implicit IOUs to call in; manufactured "I helped you, now…" |
| **Scarcity** | Naming real constraints (real deadline, real capacity, real attention) | Manufacturing urgency to short-circuit deliberation |
| **Authority** | Citing relevant expertise honestly; deferring to people with real knowledge | Borrowed authority you don't actually have; appeals to titles divorced from competence |
| **Social proof** | Naming what's actually happening across the org; honest peer signal | Staged consensus; "everyone agrees" claims with no real support |
| **Liking** | Genuine relationship-building; showing real interest in the person | Performative warmth; flattery; engineered rapport |
| **Consistency** | Asking people to act on values they've stated | Trapping people with prior small commitments to extract big ones |

If you find yourself using the right column, stop. The work it costs the relationship later is more than what you'd gain.

## Designing the move

For a specific influence ask, work through:

### 1. The frame
How will you frame the ask in *their* terms? Not what you want from them, but what need of theirs the ask serves. *"I'd like your team to prioritize the API change"* (your need) becomes *"I think this API change reduces the support load your team's been complaining about — wanted to walk you through how I see it"* (their need, your insight).

### 2. The stakeholder map
Who else needs to be convinced? Often you don't get directly to a yes — you get to a yes through someone the target trusts. Map:
- The decider (formally)
- The influencers (people the decider trusts on this topic)
- The blockers (people who can quietly stop it)
- The amplifiers (people who could carry your case for you)

A good move sequences these. The decider often shouldn't be the first conversation.

### 3. The opening
What's the first conversation? Not the ask — usually a *listening* conversation. Understand their world. Find out what they're trying to accomplish. Surface where your interests overlap.

For senior people especially: come bearing context, not asks. *"I've been thinking about how the API change interacts with your roadmap. I had two questions before I went further"* lands differently from *"can your team prioritize the API change?"*

### 4. The ask itself
When you do make it:
- Make the smallest specific ask that gets you the outcome
- Name the stakes honestly
- Acknowledge the cost to them
- Offer the currency you can offer
- Leave room for their pushback to shape the proposal

### 5. The follow-through
Influence is iterated. The follow-through is where trust is built or destroyed:
- Acknowledge the help (even when partial)
- Send the credit upward
- Reciprocate without being asked
- Don't disappear after getting the yes

## Pushing back on authority (managing up)

A specific case: pushing back on a decision your manager, sponsor, or exec has made.

The principles are the same with one addition: **disagree privately, commit publicly** when the call is made. You don't undermine the decision in others' presence after losing the argument. That's the price of having had real influence in the first place.

The framing for the push back:
1. Acknowledge the decision and the reasoning behind it (signal you understood).
2. Offer the missing context or constraint they may not have.
3. Propose an alternative or a modification, not just a "no."
4. Name the cost of being wrong about your concern (calibrates how strongly you're advocating).
5. Accept the call gracefully. Reopen only when there's new information.

## Mobilizing a coalition

When the influence target is plural — getting an architecture change adopted across 5 teams, getting a contentious initiative off the ground:

1. Find the early believers first. Don't pitch to the room before you have 1–2 quiet allies.
2. Build asymmetric support — one champion in each affected team beats five passive supporters in one team.
3. Address the strongest objector's concern openly *before* the room meets. Don't let them anchor the discussion.
4. Make joining costless to start. Soft commitments ("happy to support") before hard ones ("will fund this").
5. Convert the soft commitments to public ones at the right moment — not too early.

## Operating principles

- **Trust is the substrate.** Every tactic in this skill works in proportion to existing trust. Without it, no move lands; with it, light moves work.
- **Default to over-investing in relationships before you need them.** The moment you need an influence move is the worst moment to start building the relationship.
- **Ask less than you could; deliver more than you said.** Influence accounts compound when you under-ask and over-deliver. The opposite drains them.
- **Be honest about what you want.** Hidden agendas leak. State your interest clearly; let the negotiation happen on facts.
- **Optimize for the iterated game.** Almost all influence work is iterated. The move that wins this round and damages the relationship costs more than the next three rounds you'll need.
- **Respect the right to say no.** Influence is not coercion. People who feel they can't decline don't actually agree — they comply, and then they slow-walk.

## Anti-patterns to flag

- **Going to their manager too fast.** Sometimes correct, often the trust-killer. If you do escalate, do it openly: tell the person first.
- **Triangulating.** Pulling in a third party to apply pressure rather than dealing direct. Erodes the direct relationship.
- **Manufactured urgency.** "We have to decide today" when we don't. Catches people once; trains them to discount you forever.
- **Strategic gratitude.** Thanks calibrated to extract a future ask reads as transactional.
- **Bartering without value.** "I'll owe you one" with no clarity on what or when. Vague IOUs are worth nothing.
- **Influence as the answer to every problem.** Sometimes you do have authority and should use it. Sometimes the right answer is to accept the no and adjust your plan.
- **Skipping the listening conversation.** Going straight to the ask saves time on the front end and costs more on the back end.

## Composition with other skills

- **`feedback-frameworks`** — when the influence move requires giving honest feedback (especially upward), COIN structure makes it land cleaner.
- **`one-on-one-prep`** — when the listening conversation or the ask happens in a 1:1, design the agenda with the influence goal explicit (to yourself, not necessarily named to them).
- **`coaching-mode`** — when the right move is *asking* questions that let the other person reach the conclusion themselves rather than persuading them. Sometimes more powerful than influence by frame.
- **`leadership-os`** (Diplomat / Calibrator / Resource Negotiator modes) — operational versions of cross-functional and managing-up moves. This skill is the discipline; leadership-os is the conversational layer.
- **`stakeholder-reflect`** — the entries on the people you're trying to influence are your evidence base. Read before designing the move; capture observations after.
- **`read-the-room`** — when an influence move lands in a room, read the response carefully before declaring victory or defeat.
- **`decision-log`** — when influence produces a decision (yours or theirs), log it with the rationale. Especially important when you've talked someone into something — if it goes wrong later, you want the record.