What a point of view is

Supports: Developer, weeks 4 to 6 (defending the roadmap and presenting the Product Package) and the Master Developer capstone. Time: about two and a half hours.

1. Why this matters

When you own a product, people judge it before they use it. A customer reads a page. A sponsor reads a note. An employer reads your repository. Most never meet you. If your thinking is visible, they can see how you decide, what you notice and what you refuse to claim. That builds trust faster than any slogan.

Making your thinking visible has a name: thought leadership. The name makes some people uncomfortable, because it sounds like a person who shouts big ideas. We use a smaller meaning. In the Dojo, it is the habit of sharing a clear point of view in public, with the evidence behind it, so that others can test it and answer. You do not need to be famous. You need to be clear and honest.

A Product Development Manager (PDM) holds a whole product in mind: the customer, the money, the design, the risk and the build. A writer who knows one discipline sees one slice. You can say something that crosses them, and that is often where the useful ideas live.

This lesson is the base for the other two. You decide what you believe, check what you want to quote and write it as one claim.

The four questions for this lesson

1. What do I decide here? What I believe, how strongly, what I am willing to say in public, and which outside facts I trust enough to repeat.
2. What do I delegate to agents, and with what exact brief? Challenging my claim, finding the best argument against it, and finding where a statistic came from (section 5).
3. How do I verify what comes back? I open every source myself, read what it really says and check it against my own evidence.
4. Which other disciplines are involved? Brand, design and support (will a customer read it the same way?), finance (is any money claim true?), legal and privacy (what may I say?).

Rung on the agent ladder: use. You work with one agent, step by step, as a sparring partner. You do not hand it a whole task yet. Lesson 2 moves to delegate.

2. Key ideas

A point of view is a claim

A point of view is a statement that:

• says something about the world, not only about you;
• could be wrong, so a reasonable person could disagree;
• is supported by something you did, saw or built;
• is useful to a reader who has to decide or act.

Compare five pieces of writing about the fictional product TrailMug, which makes reusable cups for cafés.

Kind	Example	Point of view?
Topic	“Reusable cups and cafés”	No. It names a subject
Summary	“Here is what we built this month”	No. It reports; it does not claim
News	“TrailMug opens a second trial café”	No. It is a fact; useful, but not an argument
Pitch	“TrailMug is the best way to cut cup waste”	No. It cannot be tested and has no evidence
Point of view	“A cup swap scheme lives or dies at the counter: if the swap is slower than a paper cup, cafés stop doing it”	Yes. It can be wrong, and a trial can support it

Someone can answer the last one with “I disagree: price matters more than speed.” That is a gift. It tells you what to look at next.

Where a point of view comes from

It rarely arrives as a flash of genius. It comes from your own work: what you decided, what surprised you and what you changed your mind about. A decision record is a good start, because it already holds a choice, the options you rejected and the trade-off you accepted. So is an interview where you heard something unexpected, or a roadmap order you argued about.

Writing it down is part of finding it. A well-known essay (see the resources) says that putting an idea into words shows how much of it was still unclear. Expect your first sentence to be wrong, and rewrite it until it says what you mean.

The qualities worth aiming for

A thought-leadership guide in the Dojo’s source material lists several qualities of writing that earns attention. We restate them in our own words, as a checklist for your claim:

• Insightful. It shows something the reader had not seen, or sees an old thing in a new way.
• Relevant. It matters to this reader now, and it gets to the point. It is timely and engaging.
• Forward-looking. It says what might happen next, or what to do about it.
• Impactful. The reader can act on it. It is original, and it is short enough to finish.

Nothing there says you must be loud, shocking or approved by experts. A modest, specific claim from real work can meet every quality.

Check a number before you quote it

The guide that supplied that list also quoted survey percentages about which qualities readers rate as important. It gave no source. A number with no source is a rumour in a tie. The habit you need is simple: trace it to the original, open it, and read how it was made.

Here is how we did it for this track, with no numbers repeated. It is a worked source check you can copy.

1. Find the sentence. The guide’s README stated the percentages and called them findings from “research”.
2. Search for its exact wording. A web search of the phrases led to an announcement on the website of a professional body for IT, published on 7 September 2021. It repeated the same statements and said they came from the “prior research” of a company called Thinkers360.
3. Follow it back one more step. The company’s own press release of 26 October 2020 describes a survey it ran in the autumn of 2020 among its own members and other readers and writers of business thought leadership. The press release repeats only some of the figures. The full report is for paying members.
4. Ask who asked whom. The people surveyed were members of a community built around thought leadership, plus others the company reached. They are likely to care about the topic. That does not make the answers false, but they may not describe readers in general.
5. Ask how many, and how. The open pages do not say how many people answered or how they were chosen. Without that, you cannot judge how much weight a percentage can carry.
6. Decide. The original is a company survey whose full method is behind a paywall. We cannot verify the figures. So we use no numbers from it. We use the ideas it names only as plain-language qualities, and we say where they came from.

Your rule: if you cannot find and open the original, or cannot see who was asked and how many, leave the number out. You can still make the point in words (“readers value clear, useful writing”) or by showing your own evidence (“three of the five cafés I asked said the same thing”).

The same rule applies to an agent. Agents state numbers and study names with great confidence, and some are invented. A figure from an agent is unverified until you open the original.

What would change your mind

A point of view that nothing could change is a belief. Write one sentence about what evidence would make you drop or change the claim. It protects you from your own pride and tells readers you are honest. Keep the claim small, too: one decision in one product, backed by what happened, is a good start.

3. Step by step

Step 1: Collect your raw material

Open your Product Package folder. List up to five moments where you decided something, were surprised or changed your mind. Link each to a file.

Step 2: Draft three claims

Write one sentence for a moment that says something about the world, not about you. Aim for sentences that could be wrong.

Step 3: Apply the four tests

For each claim, ask:

• Could a reasonable person disagree? If not, it is a fact or a platitude.
• What did I do or see that supports it? If nothing, it is a guess.
• Who needs this, and what could they do with it? If nobody, it is a diary entry.
• What would change my mind? If nothing, it is a belief.

Keep the claim that passes best.

Step 4: List every outside fact you want to use

If your claim leans on an outside fact (a statistic, a study, a rule), write it in a table: the fact, where you read it and the link to the original. If you cannot open the original, strike the fact out.

Step 5: Run the source check

For each fact left, follow the six steps in section 2. Write the result in a log, even if the result is “left out”.

Step 6: Rewrite the claim in plain words

Read it aloud. Cut every word a second-language reader might not know. Keep it to one or two sentences.

Step 7: Challenge it

Use an agent as a sparring partner (section 5). Keep the objections that point to something real and write what you will do about them.

Step 8: Save it in git

Commit the statement, the log and your notes. The history shows how the claim changed.

4. Worked example: TrailMug

TrailMug is fictional. The notes below are invented to show the method. The founder has run a small trial in two cafés, and she wants to find her point of view.

Raw material. Her decision record chose a flat fee for partner cafés. Her interview notes say that two café owners both mentioned the morning rush. Her demo script notes: “the swap step felt slow when I tested it with a queue.”

Three draft claims.

1. “Reusable cups are good for cafés.” (Topic-ish. Nobody disagrees.)
2. “TrailMug is the easiest cup swap on the market.” (Pitch. No evidence. Cannot be tested.)
3. “A cup swap scheme lives or dies at the counter: if swapping a cup is slower than handing over a paper cup, cafés quietly stop.” (Passes.)

Four tests for claim 3. Someone could disagree: “price matters more than speed.” Two owners and her own queue test support it. Café owners and other founders can act on it by timing the swap. She would drop it if cafés keep a slower swap because the fee is low.

The outside fact. She wanted to write that “most customers care about waste”, a claim she had seen in an article. The article gave no study. A search led to a blog citing a survey of its own readers, and that survey page was gone. Her log reads: “Not verified. Left out. Use what two owners told me, and say it is two.”

The statement she saves (thought-leadership/01-point-of-view.md):

# Point of view (draft 2, 3 November 2026)

**Claim.** A cup swap scheme lives or dies at the counter. If swapping is
slower than handing over a paper cup, cafés quietly stop.

**Why I think so.** Two owners in my trial named the morning rush as
their main worry (interview notes, files 03 and 04). My own test with a
queue felt slow at the swap step (demo script, line 12).

**What I do not know.** Whether price matters more. I have two cafés, not
a sample.

**What would change my mind.** A trial where cafés keep a slower swap
because the fee is low.

**Outside facts used.** None. One claim about customer attitudes was
removed because I could not find its source (see source-check log).

The agent’s challenge (section 5) pointed out that she had tested the swap only herself, not with real staff. She added “tested only by me” to the statement and put a staff test on her roadmap.

5. Working with AI agents on this

What I decide: the claim, my evidence and which objections matter.

What I delegate, with an exact brief. The agent is a sparring partner, not an author. Paste your statement after this brief:

Act as a sceptical reader. Below is a short statement of a point of view
I plan to share in public. Do NOT rewrite it and do NOT suggest a better
claim.
Do three things and label them 1 to 3:
1. OBJECTIONS. The three strongest honest objections a reasonable reader
   would raise.
2. GAPS. What evidence the statement lacks for its claim.
3. WORDS. Any phrase that a reader whose first language is not English
   might misunderstand, or that sounds like marketing.
Rules: do not invent facts, studies, names or numbers. If you refer to a
study or statistic, say "unverified" next to it and give me the exact
words I should search for. If you are unsure, say "not sure".

How I verify what comes back.

• I ask of each objection: is it about my claim, or a different one?
• I keep and act, or reject with a written reason.
• I open any study the agent names. If I cannot find it, it is not evidence.
• I check that the “gaps” are gaps in my evidence, not in the agent’s imagination.

Which disciplines are involved. Design and support can tell me how a customer would read the claim. Finance checks any claim about money. Brand checks whether it fits the promise I make elsewhere. Legal and privacy check what I say about people and data. I ask them. The agent does not stand in for them.

Disclose. In the log, one line: “An agent raised objections to the draft; I kept two and rejected one, with reasons in this file.” See Disclose AI use.

Where this sits on the ladder. This lesson is on the use rung: one agent, one step at a time. Later, a sparring partner can become several agents, each arguing a different side. That is orchestration, and it works only if each session reads the same evidence file. Governance would be a written rule, for example “no post is published until a named person has read the claims table”. Lesson 3 shows a small version.

6. Common mistakes

• Choosing a topic, not a claim. “Reusable cups” is not a point of view.
• Making it a pitch. If it could be a slogan on your product page, it is not a claim.
• Quoting a number you cannot trace. Search for the original. If you cannot open it, drop the number.
• Trusting a repeated number. Ten websites repeating one figure is one source, not ten.
• Forgetting who was asked. A survey of a company’s own community may not describe everyone.
• Letting the agent write the claim. You would publish its opinion under your name.
• Hiding what you do not know. The “what I do not know” line makes the claim more credible.

7. Practice: write your point of view and keep the source log

Deliverable: three files in your product’s git repository, in thought-leadership/.

1. 01-point-of-view.md: the statement, in the shape shown in section 4. One claim, why you think so with links to files in your Product Package, what you do not know, what would change your mind, and which outside facts you used (or “none”).
2. 01-source-log.md: a table with one row per outside fact you considered. Columns: the fact, where you read it, the link to the original, whether you opened it, who was asked and how many (if it is a survey), and your decision (use or leave out). If there are no outside facts, write “none considered”.
3. 01-agent-review.md: the exact brief you used, the agent’s raw answer and your verdict for each objection (keep or reject, with a reason).

Add one AI-use line to the statement. Commit with a clear message, for example Add point of view and source log.

Check: give the statement to one person and ask: “In your words, what do I believe?” If they cannot tell, rewrite it.

Stretch: run the source check on one statistic quoted about your market. Write two lines on what you found.

8. Checklist

• My claim says something about the world, and someone could disagree.
• It is supported by something I did, saw or built, and I linked it.
• I wrote what I do not know and what would change my mind.
• Every outside fact has a link to its original, and I opened it.
• I left out every number I could not trace.
• I wrote down who was asked and how many for any survey I used.
• I used an agent only to challenge the claim and wrote my verdicts.
• A person read it and told me what they understood.
• I recorded my AI use.

9. Further learning

The first essay below shows why writing is a way of thinking, so draft before you decide you have nothing to say. The plain-writing essay and the essay collection give models of clear claims in short pieces. The article collection from a firm of product advisers shows how practitioners state opinions; read it selectively and compare it with your own customers. The MIT course material helps you study your reader. In the next lesson you turn your claim into a short post.