Skip to main content

Conceptadvanced

Architecture Presentations

Overview

An architecture presentation exists to produce something: a decision, an alignment, a commitment, a course correction.

the question that structures everything
"what needs to happen by the end of this meeting?"

Most technical presentations don't answer that question. They present content — context, analysis, architecture, plan — and end with nobody knowing what was asked for.

The result is predictable: the meeting ends, people vaguely agree, and nothing happens.

Problem

The pattern:

34 slides
context (8), problem (6), current architecture (5),
proposal (9), plan (4), next steps (2)
time: 45 minutes, with 40 of presenting
result: "very good, we'll evaluate it"

No ask was made. "We'll evaluate it" is what people say when they don't know what was asked for — and what gets evaluated afterwards is what each person remembered, which is little.

And there is a second problem, specific to technical presentations: density. A slide with a twenty-component diagram occupies everyone's attention for two minutes and communicates little, because each person is trying to decipher a different part.

Core Concepts

Start with the ask

"I am asking for approval to allocate 25% of engineering
capacity for 12 months to an initiative. I will explain why,
and what happens if we don't."

First slide, or first thirty seconds. That orients all the listening: people start evaluating what they hear against the ask, instead of trying to guess where the presentation is going.

And it has an additional effect: if the ask is unacceptable for some structural reason, that shows up in minute one rather than in minute forty.

See communication.

A four-part structure

1. the ask what I need from you
2. the reason, with a number why
3. what happens if we don't the real alternative
4. the risk and the mitigation what can go wrong

Four slides, or four paragraphs. The technical detail goes into an appendix, and it is used if asked about.

That structure works because it answers the questions in the order whoever decides asks them. The engineering structure — context, analysis, conclusion — answers in the order whoever built the analysis produced it, which is a different thing.

Reserve half the time for discussion

a 45-min presentation 20 of presenting, 25 of discussion

The discussion is where the decision happens. A presentation that takes 40 of the 45 minutes leaves five for what matters — and the result is "we'll evaluate it", because there was no time to evaluate.

And there is a practical consequence: if you can't present in twenty minutes, the material is too dense or the scope is wrong.

One diagram, one message

bad one diagram with the whole architecture
good three diagrams, each showing one thing

A diagram in a presentation gets a few seconds of attention. If it requires a minute of study, it doesn't communicate — it occupies.

What works: a diagram with the minimum needed for that moment's message, with the rest removed. See diagram quality.

The likely objections go in the material

Anticipating the three most likely objections and addressing them before they are raised demonstrates preparation and saves discussion time:

"you will probably ask why we didn't use vendor X's
solution. We evaluated it, and it doesn't meet the
data residency requirement — it's in appendix B."

That also protects against the dynamic in which a known objection brings the proposal down by appearing unconsidered.

The individual conversations come first

Decision meetings rarely change positions — they confirm positions formed beforehand. That means the persuasion work happens before, individually.

"I spoke with security, operations and finance. Security has
a reservation about X, which is addressed on slide 4."

Arriving at the meeting knowing each participant's position is what avoids the surprise that blocks the decision. See stakeholder management.

A document before slides, for relevant decisions

slides tolerate gaps in reasoning
a document exposes the weak argument before the meeting

A two- to four-page document, read beforehand, produces far better meetings: people arrive with questions instead of with confusion, and the time is spent on what matters.

It is also what remains afterwards — slides, with no narration, communicate nothing.

End with what was decided

"we decided X. So-and-so is responsible for Y, by deadline Z.
W was left open, and comes back next time."

Thirty seconds at the end, with the record sent afterwards. Without it, each participant's memory diverges — and the divergence shows up weeks later, when it is expensive.

Mental Model

Start with the ask, reserve half the time for discussion, and end by recording what was decided. The detail goes into an appendix.

When to Use

  • When there is a decision to produce, an alignment to build or a commitment to obtain.
  • With the ask stated in the first thirty seconds.
  • After the individual conversations.

When Not to Use

With no explicit ask — an informational presentation should be a document.

With presenting taking all the time.

With dense diagrams.

Without having talked beforehand to whoever can block it.

Without recording what was decided.

When a document would do — for many cases, it does it better.

Alternatives

  • A document read beforehand, with the meeting dedicated to discussion. It is superior for complex decisions.
  • Individual conversations only — when there is no need for a collective decision.
  • A demonstration — when there is something working, showing it is worth more than presenting.
  • Nothing — not every decision needs a meeting; many are resolved asynchronously.

The first is the standard in organizations that have adopted it, and the change is usually perceived as one of the most effective a technical group can make.

Trade-offs

PresentationDocument
Interactive, adjusts liveForces clarity, persists
Tolerates gapsRequires that they read it
Better for aligningBetter for deciding
Detail in the bodyDetail in an appendix
Demonstrates depthKeeps the focus
Consumes discussion timeRequires more preparation

Failure Modes

No ask. "We'll evaluate it."

A long presentation. No time to decide.

A dense diagram. Occupies attention without communicating.

A known objection not addressed. It brings the proposal down.

No prior conversations. A surprise that blocks.

No final record. Divergent memories.

Common Mistakes

Building it in engineering order.

Putting the technical detail in the body.

Not reserving discussion time.

Presenting to mixed audiences with the same version.

Not sending the record of what was decided.

Real-World Example

An insurance company's architecture group had a proposal approval rate of 31%. The presentations were considered good — the material was praised — and the decisions didn't come.

A review of twelve presentations found the pattern:

presentations with an explicit ask at the start 2 of 12
average presenting time 38 min of 45
slides per presentation, median 29
presentations preceded by individual
conversations with every decider 3 of 12
presentations that ended with a record
of what was decided 1 of 12

The changes were about format, not content:

A three-page document sent 48 hours in advance, with the four-part structure. The meeting came to start with five minutes of silence for anyone who hadn't read it.

The ask in the first paragraph, always.

Presenting for at most 15 minutes, with the rest of the time in discussion.

Mandatory individual conversations with everyone who can block it, before the meeting.

Three likely objections addressed in the document, in a section of their own.

A record of what was decided sent the same day, with owners and deadlines.

Results after 14 months, across 19 proposals:

approval rate from 31% to 79%
average time from proposal to decision from 6.4 to 2.1 weeks
proposals rejected 4, all for a structural
reason identified
in the prior conversation
proposals that came back to the table
from a late objection 0 (against 5 in the previous
period)

The four rejected are the result the group considers most valuable: all were rejected before the meeting, in the individual conversations, which saved weeks of preparation on proposals that wouldn't advance.

The recorded lesson: the technical content of the proposals didn't change. What changed was that it stopped taking up the meeting — and that the ask started being made.

And the five minutes of silence at the start, for reading, was the strangest change and the most cited: it ensured everyone arrived at the same starting point, which the presentation never achieved.

Practical Exercise

Take the last architecture presentation you gave and answer: what was the ask, and in what minute was it made?

If there was no ask, or if it came after minute thirty, you have found the reason the decision didn't come.

Interview Questions

  • Why does starting with the ask change how the whole presentation is heard?
  • Why do decision meetings rarely change positions?
  • Why does a document read beforehand produce better meetings?

Further Reading

  • Minto, Barbara. The Pyramid Principle. Pearson, 2009.
  • Hohpe, Gregor. The Software Architect Elevator. O'Reilly, 2020.
  • Duarte, Nancy. Resonate. Wiley, 2010.
Finished reading this document?Your progress is saved in this browser only.