Choosing topics and intent
Topics come from questions people actually ask, not from a keyword list. A page exists only if there is a real question behind the query and a structural answer we are allowed to give.
The starting point is intent classification. A query such as brex card is brand-led and definitional: the reader wants to know what category of product this is. A query such as card limits is operational: the reader already has a program and wants to know how a mechanism behaves. A query such as corporate card vs credit card is comparative. Each of those needs a different page shape, and mixing them produces a page that answers nobody.
Once intent is fixed, the page is placed in a cluster — cards, spend management, expense management, solutions, comparisons, guides or company reference — and its links up, sideways and outwards are decided before the prose is written. That ordering matters: it prevents the same explanation being repeated across four pages, and it makes the missing pages obvious.
The pipeline
-
Define the question
Write the single question the page answers in one sentence. If it takes two sentences, it is two pages.
-
Classify the intent
Definitional, operational, comparative or contextual. The classification determines the section outline and the block types used.
-
Gather reference material
Publicly available provider documentation, payment-industry and regulatory reference material, accounting and internal-control practice, and standard program-administration patterns.
-
Separate the claim types
Each factual sentence in the outline is marked as definitional, structural, mechanism, practice or provider-specific. Provider-specific claims are removed and replaced with a pointer to the source.
-
Draft structurally
Write to the outline using definitions, tables, steps and split panels so the structure carries the explanation rather than paragraph volume.
-
Place the internal links
Hub link, sibling links, guide links and at least one comparison link, positioned inside the prose where they answer the reader’s next question.
-
Review against the checklist
Apply the fact-checking checklist: no figures, no provider terms, no trust signals, no advice framing, no transactional language.
-
Publish and date
The page carries an update date, and later revisions move it. Reported errors re-enter the pipeline at the claim-classification step.
Classes of source
We work from four classes of material. None of them is quoted as authority for a commercial claim, because none of them can be.
- Public provider documentation
- Openly published help material and product descriptions, read to understand vocabulary and how features are named — never to reproduce pricing, eligibility or contract terms.
- Payment-industry and regulatory reference material
- General descriptions of how authorisation, merchant category classification, settlement and card issuance work at the network and scheme level.
- Accounting and internal-control practice
- Standard treatment of expense documentation, approval separation, audit evidence and period close, described as general practice rather than as a rule for any company.
- Program-administration patterns
- Widely used approaches to role-based issuance, limit templates, budget ownership, receipt thresholds and periodic access review.
Pages that lean heavily on one of these classes carry a short sources and reference basis list naming the class, so a reader can see the type of material behind the explanation without being handed a fake citation to a specific document.
How claims are handled
This table is the operative part of the method. Every factual sentence in a draft is assigned one of these rows, and the row decides whether it survives.
| Claim type | Example shape | How it is handled |
|---|---|---|
| Definitional | "A charge card settles in full each cycle." | Published, phrased as a category definition |
| Structural | "Corporate cards are typically issued in the company’s name." | Published with a hedge that reflects real variation |
| Mechanism | "Category rules are evaluated at authorisation." | Published, described as behaviour rather than as a feature |
| Practice | "Receipt thresholds are commonly set per policy." | Published as a common pattern, never as a requirement |
| Provider-specific commercial | "Provider X charges Y." | Removed; replaced with a pointer to the provider |
| Quantitative | Any rate, fee, limit, count, share or date | Removed unless it counts something on this site |
| Evaluative | "The best corporate card is…" | Removed; rewritten as criteria the reader applies |
The same table is used during review, which is why the fact-checking page reads as its mirror image.
Why no pricing or eligibility data
This is the most common question about the method, so it is worth stating the reasoning rather than just the rule. Commercial terms are volatile, applicant-specific and jurisdiction-specific. A fee schedule copied today can be wrong within weeks, and a reader who acts on a stale figure has been actively misled, not merely under-served.
- Terms change without notice and we have no mechanism to detect the change
- Eligibility depends on the applicant, not on the product alone
- Availability and wording vary by country and by entity type
- A wrong figure is worse than no figure, because it looks like an answer
- Providers publish this information themselves, authoritatively and for free
So the method substitutes explanation for data. Instead of a rate, we explain what the rate applies to and when it is charged, so the provider’s own schedule becomes readable. That is the intended value of pages such as business credit card vs charge card.
Comparison frameworks and structure decisions
Comparisons are built criteria-first. We define the dimensions that genuinely differentiate card structures — liability, settlement, underwriting basis, administration model, control granularity, reporting depth — then describe how each category behaves on each dimension. There is no scoring, no weighting and no aggregate verdict, because a weighting is a statement about the reader’s priorities and we do not know them.
Page structure follows the same logic. Section order tracks the order in which a reader’s questions arrive: what is this, how does it work, what does it connect to, how do I decide. Internal links are placed at the moment the question they answer appears, and the closing related block collects the routes a reader is most likely to want next.
Pages are reviewed on a rolling basis, and immediately when a reported error or a change in a related page makes a revision necessary. Review moves the update date; cosmetic edits do not.
Limits of the method
Being explicit about what this method cannot do is part of the method.
What it does well
- Explains vocabulary and category boundaries clearly
- Describes mechanisms that are stable across providers
- Gives a reader criteria they can apply themselves
- Stays correct for long periods because it avoids volatile data
What it cannot do
- Tell you what any specific product costs or offers
- Predict an approval outcome or a credit decision
- Account for every jurisdiction, entity type or edge case
- Replace advice from an accountant, adviser or lawyer
FAQ
Frequently asked questions
Do you contact providers as part of the research?
No. Pages are written from publicly available material and general industry practice. We do not conduct interviews, request statements or accept provider input, which keeps the content independent but also means it can never speak to a provider’s current terms.
Why are hedging words like "typically" used so often?
Because card program design genuinely varies. "Typically" marks a common pattern rather than a rule, and removing it would turn an accurate observation into a false absolute. Where something is a definition rather than a pattern, the hedge disappears.
How often are pages reviewed?
On a rolling basis, and immediately when an error is reported or a related page changes in a way that makes an explanation inconsistent. Each page shows the date of its last substantive update.
Sources and reference basis
- Reference General payment-network descriptions of authorisation, settlement and merchant category classification.
- Practice Standard corporate card program administration: role-based issuance, limit templates, periodic access review.
- Practice General accounting and internal-control treatment of expense documentation, approval separation and period close.
- Policy Applied together with our editorial policy and fact-checking process.