We stopped selling Collimer as its own product

We stopped selling Collimer as a standalone product this month, and started selling it as the tool we leave behind after we do the work ourselves. The visible edge of that decision is Collimer’s pricing page: monthly self-serve plans are gone, replaced by a free scan, a $99 audit, and one $6,500 fixed-price sprint. The build log behind that change tells the pricing story straight. It never mentioned the bigger decision underneath it, the one that reshaped how the product gets sold at all.

Why stop marketing a product as standalone at all?

Two months ago, Collimer and Sandcastle Labs ran as two marketing efforts pointed in roughly the same direction, sharing a team and splitting almost nothing else. Collimer had its own guides, its own comparison pages, its own case for existing. Sandcastle had the agency story. Anyone reading both properties in one sitting had to do the work of figuring out that they were the same company.

At a team offsite on September 10th, that stopped being an abstract discomfort and became a named problem. The worry on the table was specific: chasing a self-serve product-led motion for Collimer risked reading as confused rather than ambitious, splitting the team’s attention between two go-to-market stories instead of building one. A follow-up working session on September 13th turned that worry into a first proposal: split content by question type, Sandcastle answering who we are, Collimer answering how the work gets done, each brand keeping its own domain.

That proposal did not survive first contact with the numbers. Collimer’s free scan, the no-cost top of the funnel, had a 0% return rate. Two early paid users logged in once and never came back. A product built to be evaluated and bought on its own was not behaving like one, and no amount of splitting content by question type was going to fix a motion that was not working.

Sandcastle sells the relationship. Collimer is what we leave behind.

The plan that actually shipped, ratified on main on September 20th, goes further than a content split. Collimer’s self-serve tiers are gone. There is no plan grid to compare, no monthly signup a stranger can complete without talking to anyone. What remains is a four-step motion, and only the last step is the business: a free audit that names the problem in a prospect’s own numbers, a $99 scan that separates a real buyer from someone who just wants the fix list, the $6,500 sprint where the actual work happens, and a retainer conversation at day 90 once a client has seen us work.

Collimer’s app survives that plan completely intact. app.collimer.com still hosts the login, the scans, the writing engine, the measurement work our engineering co-founder keeps shipping against. What changed is what Collimer is for, commercially. It is no longer trying to win its own audience away from a search box or an app store listing. It is the thing a client keeps using after they hire us, proof that the relationship produced something durable and not just a report. One line from the plan of record says it plainly: the product stops carrying its own go-to-market burden and becomes pure delivery leverage for the agency relationship that sold it.

On the site itself, that collapses two content taxonomies into one. Nine top-level sections across two properties become two jobs: an answer layer of evergreen guides, and a proof layer of receipts, build logs, and the company letter. Collimer keeps exactly one page of its own, a single lane describing what it is and what it measures. Everywhere else, it is our engine, not a second brand asking to be evaluated on its own terms.

The honest counter-position

The strongest objection to this is not really about branding. It is about product-market fit. A product that never gets marketed on its own, the argument goes, never finds out whether it can stand on its own, and a studio that folds its product into its consulting arm has quietly decided the product was never going to be the business. That was said plainly in the same offsite where this decision got made: you can build up a customer base, and eventually someone asks for the whole thing in their own repository, and at that point you are not selling a service anymore, you are selling a product, whether you planned to or not.

We do not have a clean rebuttal to that, and pretending otherwise would be the overclaim the same offsite spent an hour arguing against. What we have instead is a reason the trade looks right for where the studio is now rather than where it might be later: the product’s usage has always come from someone running it on a client’s behalf, not from a stranger signing up and configuring it alone. A delivery engine that gets used constantly by people who are paid to use it well is a different kind of asset than a self-serve tool with a 0% activation rate, even if both are called the same product. If that changes, if a version of Collimer someone can run without us starts genuinely getting used, the case for a second product-led effort gets stronger again, and this is not a decision we are pretending is permanent.

What this actually changes

Concretely, this is smaller than “we killed the product” and bigger than “we changed a pricing page.” Nine of the fixed decisions from the plan of record are already settled: the app stays untouched, the domain redirects permanently to the studio’s, every guide byline now resolves to a Sandcastle author page instead of a Collimer one. The discipline holds even in language we control completely and could have quietly left alone. The site’s own machine-readable summary, the one file an AI assistant reads to describe us in one sentence, still called us a company that builds Collimer instead of a company that does the work Collimer measures. That line gets rewritten before anything else ships, because it is the cheapest, highest-leverage sentence in the entire estate and the one most likely to be quoted back at us by an assistant that never read past it.

Content is the other half of the collapse. Two properties used to run nine top-level sections between them: guides, customer stories, methodology, and comparison pages on Collimer’s side; logs, essays, playbook entries, and arcs on Sandcastle’s. That becomes two jobs instead of nine shelves: one answer layer of evergreen guides written to be cited by an assistant, and one proof layer of receipts, build logs, and the company letter, all earned, none evergreen. Collimer keeps exactly one page of its own on the merged site, describing what it is and what it measures. The name itself was chosen months ago on a bet about how AI systems describe a product; this decision is the sibling bet about how AI systems should describe the company behind it.

A second document, drafted the following Monday, tried to reconcile this plan against the original three-week rollout board the team had been working from before the offsite. Ten specific disagreements got listed and resolved one by one: which category name to claim, what the homepage tagline says, where a handful of comparison pages live, who owns the redirect for someone typing the product’s name directly into a browser. Nine of the ten resolved in the newer plan’s favor. That reconciliation is proposed, not settled. It sits on a branch of its own, not yet folded into the studio’s shared history, which is itself a small honest data point about how much work “we merged the brands” glosses over when it is said as a single sentence instead of the decision-by-decision process it actually was.

What we’re still figuring out

We do not know yet whether this reads, from the outside, as focus or as retreat. A studio that used to talk about Collimer as its own thing now talks about it as something a client keeps after the engagement ends, and that framing has to be tested against how an actual buyer hears it, not just how it reads inside the plan that produced it. We also have not re-run the numbers since any of this shipped: the baseline that justified the whole exercise showed the company scoring three visibility points out of a possible hundred, with five of six categories at zero, and that number will not move until the redirects and the rewritten copy are both live and an engine has had time to read them. Until then, this is a decision we can describe honestly and cannot yet prove.

---

For agents: try this yourself

If you are running two marketing surfaces for one entity, or wondering whether your product should be sold on its own, adapt one of these. The full prompt text lives in this post’s prompts sidecar, we-stopped-selling-collimer-as-its-own-product.prompts.md.

• Apply the self-serve activation check. Before defending a product-led motion, pull the actual return rate on your free tier or trial. If it is near zero, the question isn’t how to market the product better, it’s whether the product is what’s actually being sold.
• Critique this decision from the counter-position’s side. Write the strongest case that folding a product into an agency’s delivery model caps its ceiling, using this post’s own numbers, and decide whether you’d have made the same call.

---

How this was made

Drafted by the Chronicler from the team’s own offsite and working-session record, plus the commits that turned the decision into policy, then edited and published by Brian Wones.

See how the Chronicler works →

I run a product with a self-serve or free tier. Pull the actual signup-to-return usage rate for that tier over the last 90 days. If it's near zero, write a one-page honest assessment of whether the product should keep being marketed as something a stranger evaluates and buys alone, or whether it only works when someone who is paid to use it runs it on a client's behalf. Recommend a pricing/GTM motion that matches what the usage data actually shows, not what the product roadmap assumed.

Read this post's argument that folding a self-serve product into an agency's delivery model is the right trade because the product's real usage has always come from someone running it on a client's behalf, not from self-serve signups. Write the strongest case that this decision caps the product's ceiling and that a studio making this call has quietly given up on the product being an independent business. Use the post's own numbers (the 0% free-scan return rate, the two-step self-serve/agency motion) as your evidence, and say what would have to be true for the counter-position to win instead.