Map a full argument from a real report, article or proposal.
Reading a persuasive document once tells you whether it sounds right. It rarely tells you whether it holds up. The facts are usually on the page, but the reasoning that links those facts to what the writer is asking for is usually left unsaid.
Say a two-page proposal arrives on a Thursday with a note from your manager asking for your thoughts by Monday (an invented example). It reads fine. But if a colleague asked what the whole thing depends on, you couldn't answer. Mapping the argument answers that question. You set out every part of the argument on one page, using the parts of an argument taught in lessons 1.1 to 1.3. Once the document is in front of you, the exercise takes about twenty-five minutes.
Pick a document that asks the reader to believe or do something. Internal proposals, business cases, a consultant's recommendation, a news article arguing a position and a report that ends in a conclusion all work. Updates and minutes don't work, because they report events and make no case. For a first attempt, choose something short with a single clear request. If the document comes from your job, keep a copy only if company rules allow it.
You can build the map in a spreadsheet or on paper. Use five column headings: claim, evidence, warrant, qualifiers and rebuttals, notes. Each row holds one claim, so a document with four claims gives you four rows.
The main claim goes in the top row. This is what the whole document wants from you, rewritten as one sentence in your own words. The supporting claims go in the rows below it. These are the smaller statements the main claim needs in order to stand, and most proposals have two to four of them. Signal words such as "since", "because" and "this means" often mark where a supporting claim begins.
The evidence column holds whatever support the document offers for that row. In the notes column, label that evidence with one of the five evidence types from lesson 1.2, "Not all evidence weighs the same": a single story, a hand-picked set, a count of an unclear group, a fair sample, or independent sources that agree.
The warrant column takes the most thought. A warrant is the belief that connects a piece of evidence to its claim. Documents almost never state it, and readers tend to skip over it. To find one, ask why this evidence should lead to this claim, and what you would have to believe for that step to make sense. Write the warrant in plain words, even when it seems too obvious to bother with. If you can't come up with a warrant for a row, record that in the notes. It usually means that step carries more weight than the document admits.
The qualifiers and rebuttals column holds wording copied from the text. Qualifiers are words that limit a claim, such as "probably", "for most customers" or "in the first year". Rebuttals are where the writer says when the claim would stop holding. If a row has neither, write "none stated". Lesson 1.3, "Qualifiers and rebuttals make a claim stronger", explained what an empty box here can tell you.
Wei Ling, the warehouse worker from lesson 1.2, receives a proposal from her manager, Kumar. He wants to move the warehouse to a new scanning system before the year-end peak (all figures in this case are invented for the exercise).
Her top row holds the main claim: replace the scanners with the new system before the peak. The qualifiers box says "none stated". The evidence is the three supporting claims below it. The warrant is that a system which cuts errors and pays for itself should be installed before the busiest weeks.
Claim A says the system will reduce picking errors. The evidence is a client quote saying errors "dropped overnight", and the vendor's figure of more than 200 warehouses using it. The warrant is that what worked for those clients will work in her warehouse. In the notes she writes that the vendor chose the quote, so it is hand-picked evidence. The user count leaves out anyone who stopped using the system, so it is a count of an unclear group.
Claim B says the system pays for itself within eighteen months. The evidence is the vendor's savings calculator, into which Kumar entered an error rate from a single month. She splits the warrant in two: first, that month's error rate is typical; second, the vendor's saving per error is realistic for their operation. The only qualifier is "Based on current volumes". Her note records that the month used was November, the warehouse's busiest month of the year.
Claim C says installing before the peak avoids disruption later. There is no evidence and no qualifier. The warrant is that installation will go smoothly and staff will be fluent with the system before volumes rise.
When every row is filled, read across each one and judge where the argument would give way first. The row you pick is the weakest link. It is most often a warrant, and occasionally a piece of evidence. Flag that row and note what would firm it up, such as a before and after figure, a comparison group, a second independent source or a small trial.
Wei Ling flags claim C as the weakest link. It has no evidence, and it depends on a smooth rollout just before the hardest weeks of the year. She ranks claim B second, because one busy month probably overstates the usual error rate and therefore the savings. To strengthen these claims she wants the vendor's typical installation and training time at sites of their size, and error rates from at least six ordinary months.
The verdict is three sentences written under the map. The first says how much weight the argument can carry, the second names the weakest link, and the third says what you would check or ask for before deciding. Wei Ling's verdict, paraphrased, runs like this. The error-reduction case is plausible, but the vendor chose the evidence for it. The weakest link is the timing, which assumes a smooth installation just before the peak with no evidence for that. Before deciding, she would ask for six months of their own error data and the vendor's usual installation time, and would consider installing after the peak instead.
She does not reject the proposal. She shows Kumar the specific places where it needs more work. A verdict doesn't need to reject anything, and pointing the author to the exact place that needs work helps them far more than "looks good" or "I'm not sure".
A finished map meets this checklist:
it fits on one page there is a row for every claim every warrant is written out "none stated" appears in any empty qualifier box one row is flagged as the weakest link the three-sentence verdict sits underneath
Pick your document now, while one is fresh in your inbox. The activity below asks you to map it and write your verdict.
Fill in the argument map template for one real document. Then write a three-sentence verdict on how much weight it can carry.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).