Engineering as marketing: the weekend free-tool playbook

The best-performing marketing asset I have built for Favors.dev is not an article. It is a free landing page roast tool that took a weekend, and in our most recent Search Console reading it drew more impressions than 29 of the 32 blog posts that ranked at all. This is the playbook for building one, scoped so that it stays a weekend and does not quietly become a second product.

Most writing on this topic is a list of other companies' famous tools illustrated with the numbers those companies chose to publish. That proves the technique works at a scale neither of us operates at, and it tells you nothing about what to build on Saturday. So here is the version from a company of one: our own figures, the five shapes a small tool can take, the exact features to refuse, and the part where the tool started beating the article written to support it.

What is engineering as marketing?

Engineering as marketing is shipping a small free tool whose job is to attract the people who would buy your product. It is not a trial, it is not a demo, and it is not a PDF behind a form. It is a working thing at its own URL that solves one narrow problem completely, for anybody, for nothing.

The reason it suits solo founders specifically is that it plays to the one resource you hold in surplus. You cannot outspend a funded competitor on ads and you cannot out-publish a content team. You can absolutely out-build both on a tool small enough that neither of them considers it worth a sprint.

Why do free tools out-rank articles?

Tools win the queries where somebody wants a job done rather than a topic explained, because they satisfy that intent completely instead of describing how to satisfy it. Nothing about a tool gets preferential treatment from a search engine. It simply matches a different, and usually more valuable, kind of query.

Look at how the two query shapes differ. Someone searching a noun phrase wants to understand something, and an article is the correct answer. Someone searching a phrase that ends in a verb wants to perform an action, and every article they open is a detour. Google's own guidance on people-first content frames the test as whether a reader leaves feeling they need to search again. A tool that does the job is the one page in the results where the answer to that is definitively no.

There is a second effect that matters more over time. Articles get read and forgotten. Tools get bookmarked, and they get linked from forum replies, from Slack channels and from other people's articles, because linking a tool is a favor to the reader in a way that linking another explainer is not. That is the compounding part, and it is why a tool is worth more than the weekend it costs even when its traffic looks unremarkable in month one.

What our free tool does in Search Console

Our roast tool and the article written to support it went live together and target the same theme, which makes them a clean comparison. Every figure below comes from the same Google Search Console property on the same dates, each covering the preceding 28 days.

Three readings from that table are worth more than the headline.

The tool earned clicks the article did not. For three consecutive readings the tool page sat between positions 7 and 11 and converted a click each time, while the article held a deeper position on roughly twice the impressions and converted none. A page that lets somebody do the thing gets chosen out of a results list more often than a page that explains it.

The crossover took six weeks. Up to mid-July the article out-impressed the tool about two to one. By the August reading that had inverted: 42 impressions for the tool against 11 for the article. Nothing changed on either page in between. The tool kept accumulating the long-tail queries a single interactive page can match, and the article did not.

The tool's average position got worse as it won. It went from 7.5 to 38.0 while its impressions more than doubled. That looks like a collapse and is not one. A page that starts matching a much wider spread of queries picks up a long tail of deep placements, and the average is dragged down by the new ones rather than by anything lost. Watch the impression count and the number of distinct queries first. Average position will mislead you here.

The honest caveat: these are small numbers on a young property, and one tool is a data point rather than proof. What I would defend is the direction, which held across four readings and matches what our templated directory pages did on a larger sample. Pages that do something out-perform pages that describe something, on this property, consistently.

How do you choose which tool to build?

Take the core noun your product deals in and run it through five tool archetypes. One of the five will be obviously buildable and obviously useful, and that is your tool. This takes about ten minutes and it is the highest-leverage ten minutes in this playbook, because the wrong idea is what turns a weekend into a quarter.

Our core noun is a founder's landing page. Checker gave us a technical audit, which devtools already does. Generator gave us a copy writer, which is a chat window with extra steps. Grader gave us a scored roast against a fixed rubric, which is a real opinion nobody else was publishing. That is the one we built.

Three filters decide between the survivors. The tool has to be adjacent to your product rather than a slice of it, because a free slice of the paid thing teaches people to use the free version. It has to deliver value in one screen, since anything needing a tutorial will not be shared. And its output has to be worth showing somebody: a score, a verdict, a generated artifact, a list. A tool whose result is a private number nobody would screenshot cannot spread, and spreading is the whole point.

How do you scope a tool to a weekend?

You scope it by deciding in advance what you will refuse to build. The features below are the ones that turn a two-day project into a two-month one, and every one of them is optional in a free tool. Write this list down before you open the editor, because you will want all six by Saturday afternoon.

What survives that cut is small enough to describe in one sentence: a URL that takes one input, does one computation and returns one result page. That is the entire product. Our roast takes a landing page URL, scores it against a fixed rubric and returns the score with the reasoning. There is no account, no history and no settings screen, and after several months live I have not once wished there were.

One thing worth real time inside that scope: the result page. It is the artifact people screenshot and send to each other, so it deserves more design attention than everything upstream of it. If you have a spare hour on Sunday, put it there rather than into another input field.

How do you turn tool users into product users?

Give the result away completely, then offer one specific next step that genuinely requires an account. Gating the result itself is the most common way founders ruin a tool that was working, because it converts interest into a bounce at the exact moment the visitor was most engaged, and it kills the sharing that made the tool worth building.

Ours works in two moves. The roast is free, instant and complete: you get the score and the reasoning with no account. What the account unlocks is a step-by-step fix plan generated from that specific roast, which is a different artifact rather than the same one held hostage. By the time somebody sees that offer they have already read a real critique of their own page, so they are deciding whether the next thing is worth an email, not whether we are worth trusting.

Three rules I would keep whatever you build. The next step should follow from the result the visitor just saw, so it reads as a continuation rather than a toll booth. It should be one option, not a plan comparison. And the tool should keep working forever for people who never take it, because a tool that decays into a nag stops being linked, and the links were the asset.

The wider version of this argument, and where a tool sits against the other free channels available to a company of one, is in SaaS marketing on no budget. If you want to see the bridge working rather than described, our landing page roast is the live example this whole post is drawn from, and the thinking behind its rubric is in the roast post.

How do you launch a free tool?

Launch it as a product, because to every venue that matters it is one. Tools are actively easier to launch than products: there is no pricing objection, no signup friction and no risk to the person trying it, which removes most of the reasons a launch post gets ignored.

• Show HN is built for exactly this. Its published guidelines ask for something people can play with, and explicitly rule out blog posts and sign-up pages, which is a free tool's natural advantage. Our Show HN launch guide covers the title format and the timing.
• Directories accept tools, and many prefer them. The submission process is the same as for a product and the acceptance rate is usually better. Work through the startup launch directories guide and treat the tool as the listing.
• Make the page itself indexable. A tool that renders entirely in JavaScript under a one-word heading has nothing to rank. Give it a real title, real copy above the interactive element explaining what it does and who it is for, and mark it up with schema.org's WebApplication type so machines can describe it accurately.
• Post it where the problem is discussed, not where tools are posted. A thread asking the exact question your tool answers is worth more than a listing nobody reads, provided you answer the question in the comment and link the tool as the shortcut rather than the point.
• Then leave it alone and let it compound. Ours took six weeks to overtake its own supporting article and has needed no maintenance since. The weekly rhythm that keeps everything else moving in the meantime is in SEO for solo founders.

One last thing that costs nothing. A free tool is the easiest asset in your portfolio to put in front of other founders, because you are handing them something useful rather than requesting a favor. That asymmetry is worth using deliberately: show it to people who will actually use it and tell you where it breaks, and you get your first real feedback and your first real links out of the same conversation.

Frequently asked questions