Free, no-login, no-tracking, and why

Most "free" things on the web aren't really free. The polished UI and zero-dollar price tag usually mean you are the product: your behavior, your account, your attention sold to whoever pays. The bargain isn't hidden; it's just rarely named out loud, because naming it would make it harder to accept.

The actual cost of the principle, in ad dollars

Display advertising on a niche editorial site like this one runs roughly $2 to $5 RPM (revenue per thousand page views) when the ads are contextual: matched to the page's topic, no profile of the reader required. Behaviorally targeted advertising, the kind that follows a reader across sites and matches ads to a tracked profile, runs 1.5x to 2x higher on the same inventory. As of April 2026, that gap has narrowed since the 2021 privacy-rule changes (iOS App Tracking Transparency, third-party-cookie deprecation) but remains material.

The decision to stay contextual-only on icandothat costs roughly half our potential ad revenue at parity reader counts. We're trading the easier revenue path for a quieter one, and we're saying so out loud because the math is what makes the principle credible. A free, no-login, no-tracking tool that didn't cost the operator anything to be that way wouldn't be much of a principle. It'd be a marketing line.

The reason behavioral pays more is mechanical, not magical. Behavioral ad networks bid on a per-impression basis using a profile that includes which sites the reader visited recently, what they searched for, what they clicked, and (in many networks) demographic inferences from those signals. An advertiser will pay more for an impression where the bidder can prove the reader is in-market for the advertised product. Take the profile away and the impression reverts to what the page itself can offer: topic, content, and the assumption that someone reading about used-electronics resale might be interested in something resale-adjacent. That's contextual. It's the older model. It's also the one that doesn't require knowing who's reading.

What no-account costs you, what it costs us

The principle isn't free, and it isn't free for the reader either. Both columns of the trade-off, side by side:

The non-negotiables, named

• Free. Fully usable without payment. No paid tier with the good features locked behind a card, no "Pro" mode, no escalating prompts to pay. The tool has to work as-is for someone who never spends a cent. Rules out: subscription revenue, paid feature gates.
• No required login. No account to create, no email harvested, no signup form between you and the tool. The session you use today isn't tied to a profile of you, because there is no profile. Optional sign-in may come later as a user benefit; it would never be a gate. Rules out: email marketing, profiles across visits, easy personalization.
• No identity-linked tracking. No profile of you tied across visits, no ad retargeting based on what you uploaded last week. Cookie consent is your control point: the Reject button on first visit denies ad and analytics cookies (the tool still works, just without the analytics signal we'd otherwise see in aggregate); the Accept button enables contextual ad cookies and aggregate analytics, neither of which builds a profile of you across sites. Either way, no third-party retargeting cookie is set, regardless of which button you pick. Rules out: the advertising model that depends on a detailed profile of each reader.
• Anonymous storage, six-month retention, only to improve our own tools. Your inputs (photos, descriptions, damage notes) live on Cloudflare storage tied to an anonymous session ID, not to you. We keep them for up to six months, then delete them. During that window we use them for one thing: making the tools better. Never sold, never shared outside the AI providers already in the request loop, never linked to a person (no account exists to link to). The active session itself expires an hour after your last action.

When the AI might be wrong

AI isn't perfect, and we won't pretend it is. When the tool can tell its identification or price suggestion is shaky, it surfaces that uncertainty. But no model alive catches every false positive or false negative, and we're not going to guess at confidence we don't have. Treat the suggestions as a starting point, not a verdict. If the answer feels off, trust that instinct over the tool.

If we ever change any of this, what would change and what wouldn't

Forward commitments are stronger than principles, because principles read as aspirations and commitments read as promises. If icandothat ever introduces optional sign-in (the most likely change), here's the explicit list of what it would and wouldn't do:

• Would: save your past listings, remember your preferred marketplace, let you export your listing history.
• Would NOT: become required to use the tool. The free, no-login path stays as the default and the full-feature default.
• Would NOT: enable cross-site tracking. The principle is the principle whether or not an account exists.
• Would NOT: enable email marketing without an explicit second opt-in (separate from the sign-in itself).
• Would NOT: change the no-data-sold pledge. Selling user data isn't on the table whether you're signed in or not.

If any of those four “would NOT” lines ever flip, that's the moment to call us on it. The dated commitments matter more than the values writeup; values can drift quietly and commitments can't.

What this means for the person at the keyboard

What you don't have to do: sign up, hand over an email, install an app, accept a privacy policy you didn't read, watch your data get sold to someone who'll target ads at you next week. What you can do anyway: use the tool, get an answer, leave. A future you might choose to sign in, to save favorites or track sales history, and that would be a benefit, not a gate.

The version of the web where AI tools serve the person at the keyboard isn't impossible. It's just unpopular among the people building most AI tools.

For the broader picture (what AI was supposed to be and what got broken), the companion piece is delivering what was promised. For the concrete data-handling specifics, read privacy and your data, in plain English.