Every freelancer and agency knows this email. "Looks great overall! A few small things: can the logo be bigger, the blue feels off to me, and something about the middle section isn't working. Also see my notes in the attached screenshot." The screenshot is a photo of a monitor, taken with a phone, at an angle.
None of this means the client is bad at giving feedback. It means the medium is bad at carrying it. Feedback about a visual thing wants to live on the visual thing, and email strips the location off every comment, leaving you to guess which blue, which section, which of the four buttons.
Why the email thread fails every time
Three separate problems compound. First, location loss: "the text near the top" costs you a clarifying round trip per comment. Second, version drift: by round three, the client is commenting on a screenshot of a version you have already changed. Third, scatter: some notes come by email, some in a text, one crucial one verbally in a call, and reconciling them into a to-do list is a job in itself.
The fix is structural, not behavioral. You will not train clients into writing better emails. You can change where the feedback lands.
Put the comments on the page
The pattern that works: the design lives at a link, reviewers click the exact spot they mean, and a pinned comment appears there. The comment carries its own coordinates. "This headline" means the headline the pin is sitting on.
Purpose-built review tools like Pastel, Markup and BugHerd all do versions of this for live websites, and they are good, particularly for ongoing QA on staging sites. For the earlier phase, when what you have is a design or a page draft rather than a deployed site, the same pattern works with less machinery: paste or upload the HTML, send one link, and reviewers pin comments straight onto the page. Nobody creates an account. Nobody installs a browser extension. The entire client-side instruction is "click where you mean."
The details that decide whether clients actually use it
No login, seriously
Every account-creation step cuts participation roughly in half, and the person you lose is usually the decision maker, who was always going to spend ninety seconds on this. The review link has to open into the work directly.
It should look like you
A review page wrapped in someone else's brand quietly undermines the work being reviewed. Your logo and your colors on the page keep the frame professional. This is free on commentable, because we think of it as table stakes rather than an upsell.
Private by default
Unreleased designs are confidential almost by definition. A public URL that happens to be obscure is not access control. Use a link that is encrypted and, when the stakes justify it, password protected.
An expiry date is a feature
A review link that dies when the project wraps is one less loose end floating around inboxes forever. Set it to outlive the feedback window by a week and forget about it.
A loop that closes in days instead of weeks
- Publish the draft to a private link with comments on.
- Send it with two or three narrow questions, and a deadline. "Anything you flag by Thursday makes it into this round."
- Comments arrive pinned to the things they are about. No archaeology.
- Fix, update the same link, reply to each pin as resolved.
That last step matters more than it looks. When a client sees their comments answered in place, they trust the process, and round two gets faster instead of slower. The thread of record is the page itself.
And if the draft came out of Claude, as more and more first drafts do, the loop starts even earlier: our guide to sharing Claude artifacts covers going from artifact to reviewable link without touching a code editor.