"Don't like the right corner of the kitchen, something with the cabinet," is a message that triggers 2 hours of clarifying questions. Which right corner, from the door side or window side? Which cabinet? What about it: color, height, location, hardware?
The alternative is an annotation on the image: the client clicks the specific spot and writes "too tall, drop it by 4 inches." Point + text and no more clarifications.
This is the difference between "talking about the picture" and "pointing at the picture." The second model cuts iterations 2 to 3 times. We break down why.
What an image annotation is
Technically, it is a marker on a pixel of the image (or on a 3D model) tied to a comment. When the client sees the mood board, they:
- Click the problem spot
- A text input field appears
- Write a short note ("too dark," "remove the cabinet," "move closer to wall")
- Save
The designer sees the same image with markers and attached comments. Each comment is a separate task with clear context.
When chat works
Text messages in chat are not "the bad tool," they just solve different problems:
- General concept: "I thought again, want to add dark accents"
- Budget questions: "How much does a sofa like this cost?"
- Timeline and logistics: "When can we meet to discuss the layout?"
- Personal situations: "I am away for a week, replies will be slow"
For such messages, an annotation on a mood board does not fit, they are not tied to a specific image.
When annotations work
Annotations are better for:
- Pointed revisions: "do not like the color here," "the table is too close here"
- Local preferences: "want a light here," "a chair here, not a cabinet"
- Specific dimensions: "too high, drop 6 inches lower"
- Doubts: "what is this thing anyway?"
Any comment that refers to a specific spot on the image is a candidate for an annotation.
Simple selection criterion
| If the revision is about... | Right tool |
|---|---|
| ...a specific point on the image | Annotation |
| ...a specific zone on the image | Annotation |
| ...an item on the image | Annotation |
| ...general atmosphere / style | Chat |
| ...budget / timeline | Chat |
| ...the brief overall | Chat + link to brief |
| ...multiple mood boards at once | Chat |
Why the mixed approach wins
Real work needs both. Studios that strictly demand all revisions only as annotations encounter clients sending "general" messages on WhatsApp bypassing the system. Studios that accept everything in chat drown in clarifying questions.
Working model:
- Pointed revisions → annotations on a specific mood board version
- General questions → chat for the project
- Final approvals → action on the platform ("Approve" button), not "ok" text
This gives the client freedom to express in a comfortable way, and the designer a structured stream of inputs instead of chaos.
What a good annotation tool should have
Not every tool fits. A good annotation system:
- Tied to a specific file version, not "the project in general"
- Has statuses (open / resolved / rejected), so the designer can mark completed
- Supports replies to annotations, short dialog on a specific point
- Works on mobile, 60% of clients view mood boards on phones
- Preserves history, annotations to V1 do not vanish when V2 appears
If the tool lacks this, you have not "revisions on the image," but "comments attached to a place." Better than chat, but not by much.
Client psychology
Annotations work better not only because of technical convenience but also because of psychology:
- Lowers the revision barrier. Easier to "mark a point and describe" than "formulate a general thought in chat"
- Lowers fear of offending. A pointed revision feels like a clarification, not criticism of the whole work
- Surfaces hidden objections. In chat the client writes "all good," in annotations there turn out to be 7 revision points. It is not "they lied," it is "they did not consider small details worth a separate message"
This psychological effect is not theory. Per our data, switching from chat to annotations increases the count of identified revisions per iteration by 30 to 50%, while the total number of iterations drops, because more revisions are caught upfront.
What to avoid
A few anti-patterns:
- Too many fields in the annotation form. If creating an annotation requires 5 fields (category, priority, tags, etc.), the client stops using it. Minimum: point + text.
- Mandatory categories. "Select revision type" is a barrier. The designer categorizes later if needed.
- No notifications. Client left annotations, designer never knew. Push, email, or at least a clear dashboard indicator.
- Cannot edit after sending. Client wrote, changed their mind, must be able to fix (but history should remain).
CTA
In Roomix annotations work per the model above: one point one comment, tied to a version, statuses, replies, history. All next to the project chat, not instead. Try it on the free plan, one project with no time limit.
See also: 7 mood board approval mistakes, studio workflow.