50%

الدرس 2 من 4

Influence Without Authority

Architects rarely command anyone. Decisions stick when teams are convinced, and the techniques for that are learnable rather than innate.

قراءة 10 دقيقة

In most organisations an architect has no line authority over the engineers whose work they shape. You cannot instruct anyone. Every decision you make survives only to the extent that people who could ignore it choose not to. This is uncomfortable to discover and it is also the healthiest arrangement available — a decision that survives scrutiny is usually a better decision.

Four techniques that reliably work

From proposal to shared decision

  1. 1

    Make the constraint visible before proposing the answer

    Teams resist conclusions handed to them and accept conclusions they help reach. Show the latency budget, the coupling graph, or the cost of the outage first. Very often the team then proposes your design themselves — which is a far better outcome than being right at them.

  2. 2

    Present at least two credible options, honestly

    A single recommendation reads as a decision already made and invites opposition. Two real options with their real costs turns the conversation into a joint evaluation. The alternative must be genuinely credible; a strawman is detected immediately and costs you the room.

  3. 3

    Name what your preferred option costs

    Volunteering the weakness of your own proposal is the single most effective credibility move available. It signals you have actually thought it through, and it stops the meeting turning into a hunt for the flaw you were hiding.

  4. 4

    Write down the decision and the rejected alternatives

    Six months later nobody remembers why. Without a record, the debate reopens with worse information, usually driven by whoever joined most recently. The record is what makes a decision durable.

Two ways to run the same design review

Decision handed down

The architect arrives with the design, presents it, and asks for questions.

  • Fast in the meeting itself
  • Clear about what has been decided
  • The team has no ownership and will not defend the decision under pressure
  • Objections surface later, in code, rather than in the room
  • You lose the team's context, which is usually better than yours on the details
  • Every future decision starts from a slightly more defensive position

Choose when: The decision is genuinely non-negotiable — a regulatory or contractual constraint — and you say so plainly rather than pretending to consult.

Constraint shared, decision reached

The architect presents the forces and two credible options, then works the trade-off with the team.

  • The team owns the outcome and will hold the line when it is inconvenient
  • Surfaces context the architect did not have
  • Objections arrive while they are still cheap to address
  • Builds the team's own architectural judgement for next time
  • Slower in the room, sometimes by a lot
  • Requires you to genuinely accept a different answer than the one you arrived with

Choose when: Nearly always. The cost is one longer meeting; the benefit is a decision that survives the first deadline that pressures it.

A team pushes back on your design

As a developer

Defend the design. You thought about it carefully, their objection has a hole in it, and finding the hole feels like doing your job well.

As an architect

Assume the objection contains information you do not have. A team pushing back usually knows something specific about the code, the operational reality or the delivery timeline that you did not factor in. Extract that first. Sometimes the design changes; often it survives with a caveat you did not know it needed; occasionally you discover the objection is really about workload rather than architecture, which is a different conversation worth having on its own terms.

In practice

Winning by losing the first argument

An architect proposed replacing three teams' ad-hoc HTTP integrations with an event bus. The teams objected: extra infrastructure, harder local development, and a new failure mode they would have to learn.

Constraints

  • Three teams, none reporting to the architect
  • A previous architecture initiative had been abandoned mid-way
  • One team was already behind on a committed deadline
  • The coupling problem was real and getting worse

Decision

Withdraw the proposal. Instead, spend two weeks instrumenting the existing integrations and publish a single chart: how often a change in one service forced an unplanned change in another, over the previous six months.

Why

The teams were not wrong about the costs, and arguing the costs down would have been arguing against people with better information. The disagreement was really about whether the coupling problem was severe enough to be worth paying for — and that was a question evidence could answer where opinion could not.

What it cost

Two weeks of the architect's time and a visible retreat that some read as weakness. The chart showed 40% of changes triggered an unplanned change elsewhere. Two of the three teams then proposed the event bus themselves and the third adopted it a quarter later. The design shipped roughly a month later than it would have if mandated, and it shipped with three teams who wanted it — which is why it was still in place two years on.

You are convinced a team is about to make a costly mistake, and you cannot persuade them. What now?Reveal

Separate two questions: is it costly enough to escalate, and is it reversible? For a reversible decision, let them proceed and agree in advance on the signal that would mean it went wrong — being demonstrably right later buys far more influence than being loudly right now, and you might be the one who is wrong. For an expensive and irreversible decision, escalate explicitly and in the open: state the risk, state that you disagree, and let the accountable person decide. Escalating is legitimate and quiet non-escalation is not, but escalation is a limited resource and spending it on reversible decisions is how architects become people teams route around.

Key takeaways

  • Architects usually have no authority, so decisions survive only if teams are genuinely convinced.
  • Show the constraint before the conclusion; teams accept what they helped reach.
  • Volunteering what your own proposal costs is the strongest credibility move available.
  • Fake consultation is detected quickly and is more damaging than an openly imposed decision.
  • Escalate for expensive irreversible decisions; let reversible ones run and agree the signal that would prove them wrong.