Sidecar products: free tools that bring in customers

The term came out of product-led growth writing around 2022, and nobody has a clear claim to coining it. The earliest public use I can find is Kyle Poyar, then at OpenView, who listed “free tools/sidecar products” as a route to near-zero acquisition cost in September 2022, with Snyk’s vulnerability database as the example. Anthony Pierri credits Poyar with the idea, and Ben Williams, who ran product-led growth at Snyk, defines sidecar products as “free, ungated products that offer value in the same or closely adjacent space to your core product”. More recently Emily Kramer at MKT1 has applied it to the tools, templates, directories and quizzes replacing the ebook as the thing a brand gives away. SaaS Hill draws the line between the core product, which people pay for, and the sidecar, which is free, smaller and points back at the core.

It is an old idea with a new name. Gabriel Weinberg and Justin Mares called it engineering as marketing in their book Traction: build something useful, give it away, and let it find your customers for you. HubSpot was doing it in 2007.

A sidecar is different from a lead magnet in one way that matters. An ebook gets read once, if at all. A tool gets used, often more than once, and it usually gives a result about the person using it: their website score, their quote, their exchange rate. That personal result is what makes people come back, share it and trust the company that built it.

Until recently a free tool needed a developer, a budget and a quarter. Now a marketer can build one in an afternoon with Lovable, v0, Bolt, Replit or Claude Code. Alex Shartsis in Growth Unhinged walks through an ROI calculator built in under two hours, and Kramer makes the point that most marketers can now build these on their own.

That changes the maths. The bar used to be “is this worth an engineering sprint?”. Now it is “is this worth a Friday?”, so you can test several small tools and keep the one that works.

It also changes outbound. Eric Nowoslawski built a custom ROI calculator for each target account at KlientBoost, Clay and Snitcher: scrape the prospect’s site, find their pricing, have AI write the build prompt and send a preview link in the first email. Showing the result before asking for the call is a sidecar used one account at a time.

Most sidecars fall into one of six shapes. The best type is the one that matches the moment your buyer is in.

Start from the job your buyer does just before they need you. A renovation client wants a rough price before they call a builder. A founder wants a business name before a store. A marketer wants a headline before a post. Build the tool for that job, and make its result point at what you sell.

Then keep it small. One input screen, one result, one next step. Ungated tools get used far more: Kyle Poyar has found ungated products get two to three times as many people starting to use them as free trials do. If you want the email, ask for it after you have shown something useful, such as the full report or a saved quote.

Give it its own page and its own name, so it can rank and be shared on its own, and so it can be switched off without hurting the main brand. Eraser made DiagramGPT look different from Eraser for exactly that reason.

Build it with what you already have. On this site the digital psychology directory is a sidecar: fifty entries in one data file, rendered as pages, and it now brings in most of the site’s Google impressions. I build these in Claude Code on Astro; Lovable and v0 work well if you want a hosted tool without a code base.

Volume is not intent. A free tool brings in many more people than a demo form, and most of them will never buy. Judge it on sign-ups or qualified leads that come from it, not visits.

Too far from the core product. A clever tool that attracts the wrong audience is a traffic stunt. Shopify’s name generator works because people naming a business are about to need a store.

Nobody owns it. Several of the tools in the list above have moved, been gated or been switched off since they were saved. A broken calculator is worse than none. Give each one an owner and a check every quarter.

Copying the format, not the moment. A calculator is not useful because it is a calculator. It is useful because it answers a question the buyer has right then.