The worst way to design a main image is to imagine one. The best way is to find images that already work, understand why, and adapt them.

This phase produces **3–5 reference images** — real listings that look close to what you want — each with a written note on exactly what you are taking from it.

[Watch the walkthrough](https://kb.trackiq.com/kb-assets/kbvideo-refimages-explainer-ed7f6eb2.mp4)

*Prefer to watch? One hundred and twenty-five seconds on the five sources ranked by usefulness, what an image has to do in a fraction of a second, and why three directions beat three variations.*

## **Copy, adjust, paste**

The principle, stated plainly: **you are not inventing a composition, you are adapting a proven one.**

A reference image is a photograph of an existing Amazon product that resembles the result you want. Alongside it you write what specifically you are borrowing — the angle, the background, the way a benefit is called out, the lifestyle context, the way scale is communicated.

**Without that note the reference is useless**, because a designer will copy the wrong thing. "Like this one" is not a brief.

![Where your clicks come from, keyword by keyword](/kb-assets/shot-b28-brandterms-view-a83aae.png)

## **Where to look**

Five sources, in roughly descending order of usefulness.

### **1. Direct competitors on your main keyword**

Search the term you most want to rank for and study the images taking the clicks. These are directly relevant, because they are competing for the same shopper in the same context.

**Look at the ones performing well, not just the ones ranking well.** Brand Analytics shows which listings receive the most clicks for a keyword, which is a better signal than position alone.

### **2. Direct competitors in your subcategory**

Broader than one keyword, and it surfaces conventions you may be breaking without realising — if every listing in the category shows the product in use and yours is on white, that is worth knowing.

### **3. Adjacent categories**

The most productive source, and the one people skip.

**Look at products that are similar in kind but not competitors.** Selling a toothbrush? Study electric toothbrushes. A reading pillow? Study pregnancy pillows.

These categories have solved the same visual problems — showing scale, showing texture, showing use — without being your direct competition, so borrowing from them makes you distinctive rather than derivative.

### **4. Design and craft platforms**

Search your main keyword on visual and handmade marketplaces. The photography conventions there are different from Amazon's, and importing one is an easy way to stand out in a grid of identical white backgrounds.

### **5. Your own reviews**

Not a source of compositions, but of *subjects*. What customers repeatedly praise or complain about tells you what the image needs to communicate.

## **What you are looking for**

An image has to do work in a fraction of a second. Aim for reference images that show one of:

**The main benefit**, visible without reading.

**How the product makes someone feel** — freshness, warmth, calm, cleanliness. Emotional register is doing more work than most sellers credit.

**The key feature**, especially anything that distinguishes you from the listings either side of you.

**Scale**, if size is a common question or complaint.

**And the single most valuable subject: whatever your competitors' 1–3 star reviews complain about.** If the category's recurring complaint is that the product is smaller than expected, an image that settles size immediately is worth more than any aesthetic improvement.

## **Building the collection**

**Collect every candidate**, without filtering as you go. Judgement comes later; the goal now is volume and range.

**For each one, write:** what you are borrowing, and why you think it would work for your product.

**Then narrow to three.** You want three distinct directions rather than three variations of one idea — if all three are the same composition with different backgrounds, your test will only tell you about backgrounds.

**Aim for genuinely different approaches:** one that leads on the benefit, one that leads on use or lifestyle, one that leads on a feature or on scale. Then you learn something regardless of which wins.

## **Check where your traffic actually comes from**

One practical decision belongs at the end of this phase, because it changes what the designer produces.

**Look at whether your sessions are mostly mobile or desktop**, and design for the majority.

The two crop differently and are read at very different sizes. An image with fine detail or small text may work on desktop and be illegible on a phone. See [**Understanding Business Reports**](https://kb.trackiq.com/hc/knowledge-base/articles/understanding-business-reports).

## **Common mistakes**

**Designing from imagination** rather than from what already works.

**Collecting references without notes**, so nobody can tell what you meant.

**Three variations of one idea**, which wastes the test.

**Only looking at direct competitors**, and producing something that blends into the category.

**Ignoring the review complaints**, which are the clearest brief you will ever get.

## **Related**

- [**Why the Main Image Decides Your CTR**](https://kb.trackiq.com/hc/knowledge-base/articles/why-the-main-image-decides-your-ctr) — whether this project is worth running.
- [**Writing an Image Brief**](https://kb.trackiq.com/hc/knowledge-base/articles/writing-an-image-brief) — turning these references into instructions.
- [**Testing a New Main Image**](https://kb.trackiq.com/hc/knowledge-base/articles/testing-a-new-main-image) — proving which one wins.