What Images to Load to Get Real Client Reactions (The 3-Source Method)

Load images from three sources: references shared by the client, images from your own past projects, and the inspiration you have already gathered for the project. Aim for at least 20 to 30 images per project. The variety matters more than the count. A real spread is what turns a client's gut reactions into a style direction you can actually design from.

The hardest part of a reaction-based discovery is not the client's side. It is the blank project. You open a new project, you know you want your client to react to images, and then you stare at an empty screen wondering what to put in.

That image set is the input the whole method runs on. In Glintera's Style Profiler, your client reacts to the images you load and you get a style profile back, the archetype, colors, materials, and a per-room breakdown, each part tied to the images they responded to. So your read is only ever as good as the spread you put in. Here is the method we recommend, and the value of each source.

Why variety matters (when your client reacts)

Variety earns its keep when your client is the one reacting. Reactions only tell you something when there is a range to react to.

Load ten near-identical rooms and your client will like all of them, and you will learn nothing. Load a real spread, warm and cool, busy and spare, safe and bold, and the pattern in what they stop on becomes obvious. The goal is contrast, not volume. You want enough variety that their reactions separate into a clear signal.

One exception worth knowing. If you are running Preview Style on your own instead of asking your client to rate and tag images in the Style Profiler, the goal flips. Preview Style treats every image you load as liked and reads a style from that, so you want a focused set that represents the client's taste, usually their own references, not a contrast spread (unless that's reflected in the images they sent you). Images you expect a "no" on would only muddy that read. (Here is how to choose between the two.)

The rest of this guide is for the case where your client will react. That spread comes from pulling from three different places.

The three sources

1. References shared by the client.

The client has likely sent you references already, the images they pulled to show you what they like. Start here. This is where you learn which specific parts of a room they love, and which parts they do not. Those references carry instincts they have never put into words, so they show their taste in close to its most honest form. One caveat worth holding onto: what a client pulls is shaped by their own feeds and algorithms, so it reflects a real slice of their taste, not always the full range of it, or what it could grow into once they see more. That is part of why the next two sources matter.

2. Images from your own past projects.

Your client hired you because they like what you do. Watching which of your past projects they react to tells you what to lean into, and what to leave behind. It also keeps your own range in the mix, so the profile reflects how you actually design, not just what is trending on Pinterest this month.

3. The inspiration you've already gathered for this project.

You have a sense of the client. You know the goals. You have probably started a board already. It is a direct read on the direction you are both circling, and it gives the client something concrete to react to instead of a description. Add it to round out the set.

Sharpen the set by finding the gaps

Once you've pulled from the three sources, do one more pass to make sure the set is balanced. Look at what you've loaded and ask what is missing. Is every image a living room, when the project is a whole house? Is everything warm, with nothing cool to test against? Are there styles you suspect they'll reject that you haven't given them a chance to reject?

You don't have to eyeball this alone. In Glintera, look for "Sharpen the profile" then hit Check balance. It highlights the gaps in your set, the rooms and styles you'd expect a reaction on but haven't covered yet, suggests search terms so you can go find images that fill them, and points you to matching images in your own library if you've already loaded a lot.

Then fill those gaps. A few images you expect a strong "no" on are as useful as the ones you expect a "yes" on. The reactions you get back are only as complete as the spread you put in.

How many images should you load?

Around 20 to 30 images per project is the range that works.

Enough variety that reactions form a real pattern. Not so many that it turns into homework and the client rushes the last half. The more rooms you are designing for, the more images you will likely want to add. A good floor is at least five images per room, so every space gets enough coverage for the reactions to mean something.

Why this beats asking clients to describe their style

A written style question captures the words a client already has. It cannot capture how they react when they see something real.

Clients have strong instincts about what they want. They just do not have the vocabulary to describe it before they see it. So when a form asks them to pick "modern" or describe their dream kitchen, they reach for borrowed words that mean something different to everyone. Images skip the translation. The client reacts to the thing directly, which is exactly what they do every time they look at a room. (More on that in Client Questionnaire vs. Image-Reaction Discovery.)

It is not a bad client problem. It is a process problem. Loading the right images is how you fix it.

Where Glintera fits

This is the exact step Glintera's Style Profiler was built around. You load images from these three sources, the client gets a link with no account or app required, and they rate what they love, tag what they would never live with, and add a note if they want. What comes back is a style profile, with the archetype, colors, materials, and a per-room breakdown, each part tied to the images they actually reacted to. You can also hit Preview Style to get a first read yourself before you ever send it.

Get early access to Glintera

FAQ

What images should I use for interior design client discovery?

Pull from three sources: references shared by the client, images from your own past projects, and the inspiration you have gathered for the project. Together they give a real spread of styles, which is what makes a client's reactions meaningful. Aim for 20 to 30 images total.

How many inspiration images should I send a client?

Around 20 to 30 per project. That is enough variety for reactions to form a clear pattern, without becoming a chore that makes the client rush. The more rooms you are designing for, the more images you will likely want to add. Aim for at least five images per room so every space gets real coverage.

Should I include my own past projects in the images?

Yes. Your client hired you for your work, so watching which of your past projects they react to tells you what to lean into and what to avoid. It also keeps your range in the profile, not just images pulled from the internet.

Why use image reactions instead of a style questionnaire?

Clients recognize what they want the moment they see it, even when they cannot describe it in words. A questionnaire captures the vocabulary they already have; image reactions capture how they actually respond. For style direction, reactions are more accurate. Use the questionnaire for facts and logistics.

What makes a good spread of discovery images?

Contrast. Warm and cool, spare and layered, safe and bold, across the rooms in the project. Include a few images you expect the client to reject, not just ones you expect them to love. The clearer the range, the clearer the signal you get back. This applies when your client will react to the images.

Do I need variety if I'm only using Preview Style?

No. Variety matters when your client reacts to the images, because their reactions need a range to separate signal from noise. Preview Style instead reads whatever you load as liked, so a focused set of the client's own representative references gives a cleaner read than a contrast spread. Save the wider spread for when you send the project to the client.