Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
The Structural Argument
Explanation is the content type generated answers replace most cleanly. If your page explains how something works, a system can read it and tell the person the same thing without sending them to you. A tool that computes a result from the reader's own inputs cannot be substituted that way, because the answer does not exist until they supply the numbers.
Be precise about the strength of that claim. It is a mechanism argument, not a measured finding: nobody has published a controlled comparison of tool pages against article pages under generated answers. The reasoning is sound and the evidence is absent, which is the honest way to hold it.
What is better established is the second benefit. Tools are among the most reliably referenced things a small site can own, because pointing at a working calculator is easier for a writer than explaining the method. That puts them in the same category as reference tables and original data, discussed in whether anything on your site is worth linking to.
Choosing One Worth Building
The test is whether people already do this badly, by hand, in a spreadsheet or on paper. If they do, a tool removes friction from something they were going to do anyway, which is a much lower bar than persuading them to want something new.
| Tool type | Works when | Example of the job |
|---|---|---|
| Calculator | The maths is fiddly and the inputs are personal | Cost, sizing, break-even, dosage, capacity |
| Checker | People want to know whether something is right | Validating a format, a setting, a configuration |
| Generator | Output follows rules people cannot remember | Producing correctly formatted text or markup |
| Lookup | Facts are scattered and checked repeatedly | Rates, thresholds, compatibility, deadlines |
| Quiz or assessment | Rarely; usually a lead form wearing a costume | Only where the result is genuinely diagnostic |
The last row deserves the scepticism. Assessments that produce a flattering score and a sales call are recognisable as lead capture and get treated accordingly. If the result would be the same whatever the answers, you have built a form, and people can tell.
Gate the Extra, Never the Answer
Demanding an email before showing any result kills the two things that make a tool worth building. Nobody links to a page that refuses to do anything, and most visitors leave rather than pay with an address for something they have not seen.
The pattern that works is showing the result and offering something genuinely additional for the address: the full breakdown, a saved or shareable version, a comparison against benchmarks, or ongoing monitoring of whatever the tool measured. The visitor has already received value, which is what makes the exchange reasonable.
This also determines whether the tool functions as a linkable asset at all. A writer will link to something a reader can use immediately and will not link to a form. Gating the answer converts a potential reference into a landing page, which is a poor trade for a modest increase in captured addresses.
Where Tools Fail
Three failure modes account for most abandoned tools, and all three are decided before any code is written.
- Nobody had the question.
The commonest failure. Check that people search for the job before building the answer, using the demand research in keyword research, not by assuming your problem is theirs.
- The result is not actionable.
A number with no interpretation leaves the user where they started. Say what the result means and what to do about it, or you have automated a calculation without answering a question.
- Maintenance was not budgeted.
A tool that produces wrong output because a rate changed is worse than no tool, and people who cited it are now pointing at something incorrect. Set a review interval at build time.
- Nobody can find it.
A tool on a page nothing links to is invisible. It needs internal links from the articles about the same problem, which is ordinary internal linking and is routinely skipped for tools.
- It scaled beyond its usefulness.
Accounts, storage, and live data turn a marketing asset into a product with support obligations. Keep the first one stateless if you can.
What a Tool Tells You That an Article Cannot
The inputs are research. Every use is somebody telling you their budget, their scale, their constraint, or their configuration, in numbers, unprompted and at volume.
That is genuinely valuable and frequently ignored. Aggregated honestly, it is original data about your market that nobody else holds, which is the strongest citable asset a small business can publish. The method for turning it into something defensible, including the privacy constraints on doing so, is in publishing original research.
It also qualifies better than a form. Someone who put real numbers into a calculator has told you the size of their problem, which makes the follow-up conversation specific rather than generic. That is the conversion argument for tools, and it holds independently of whether the AI-resistance argument turns out to be right.
Questions About Building Free Tools
- Why build a tool instead of writing an article?
Because an answer engine can restate an explanation and cannot do arithmetic on the reader's own numbers. A calculator, checker, or generator produces a result specific to the person using it, which is a job rather than a piece of information. That is a structural difference, though it is a mechanism argument rather than a measured finding.
- What makes a tool worth building?
It should do something people currently do badly in a spreadsheet, produce a result they would act on, and be finishable in days rather than months. The failure mode is building something impressive that answers a question nobody was asking, which is the same failure as writing content nobody searched for.
- Should I gate the tool behind an email?
Gate the extra, never the answer. Show the result, then offer something genuinely additional in exchange for an address: the full breakdown, a saved version, or ongoing monitoring. Demanding an email before showing anything converts worse and removes the reason anyone would link to it.
- Do free tools actually earn links?
Free tools are among the more reliably linkable things a small site can build because citing a tool is easier than explaining the calculation. That is the same property that makes reference tables and original data citable, and it is why tools belong in a linkability audit rather than a content calendar.
- How much engineering does this need?
Much less than people assume. Most useful tools are a form, some arithmetic, and a clear result. The expensive versions are the ones that need accounts, storage, or live data, and those are worth avoiding for a first attempt because the maintenance obligation outlasts the enthusiasm.
