# Visual Ad Creative Principles: Composition, Contrast and Text for Feed Ads

> Five visual principles for feed ads, backed by eye-tracking and perception research: one focal point, measurable contrast, few words set large, the offer as a shape, and purposeful faces. Plus a squint test and a guide to diagnosing weak ads.
- **Author**: Marxx
- **Published**: 2026-10-24
- **Category**: Creative strategy
- **URL**: https://marxx.ai/posts/visual-ad-creative-principles

---

**What makes a feed ad work visually, and which tools are best for designing it?** Five principles do most of the work: one focal point, measurable contrast, few words set large, an offer you can see as a shape, and a person or hand where it helps. Any tool can apply them. Canva and Adobe Express are fastest, Figma suits design systems, and Ads Manager's preview checks the result.

## Feed ads are seen in a permanent squint

Designers have an old trick for checking a layout: step back and squint. Squinting blurs the detail, so only the big shapes, the strongest contrasts and the largest words remain. If the design still works, the structure is strong.

Feed ads are always seen this way. They're small, fast, and glanced at by someone who's half-focused on something else. Research on mobile feeds suggests people can register content in a fraction of a second, and scroll faster on phones than on desktop (see [our mockup guide](/blog/facebook-ad-mockup-tools) for the evidence).

So the question isn't "is this ad beautiful?" It's "**what's left of this ad after a glance?**" This guide answers that in four parts:

1. What eye-tracking and perception research says about how people look at ads.
2. Five principles, each with the mechanism behind it, when it fails, and a ten-second check.
3. What long-running ads in the Marxx library look like through a blur.
4. How to diagnose a weak static ad, and which tools to use.

Three long-running ads shown sharp and blurred: Wonderla 99 days, Withnell 164 days, Porcellia 135 days

---

## Part 1: What the research says about how people look at ads

### Pictures, text and brand each do a different job

The largest study of its kind is still one of the most useful. Rik Pieters and Michel Wedel used infrared eye-tracking to follow more than 3,600 consumers looking at **1,363 print ads** in 65 magazines ([University of Michigan, 2004](https://news.umich.edu/print-ad-pictures-less-is-more)).

Pieters and Wedel's findings: the picture grabs attention at any size; more text space means more attention; a bigger brand doesn't win more attention but routes the eye

- **The picture grabs attention whatever its size.** Making the image bigger didn't increase attention to the ad as a whole.
- **Text earns attention in proportion to its space.** Each 1% increase in text space raised total attention by about 0.15%.
- **A bigger logo didn't win more attention.** The brand element mainly helped steer attention around the rest of the ad.

**What this means for feed ads:** The image is your hook, and you don't need to make it enormous for it to work. Text is where size pays off, so if words matter, give them room. A giant logo is mostly wasted space.

**Limitations:** These were print magazine ads in the Netherlands, viewed at reading pace, not phone ads in a fast feed. The *direction* of the findings is useful. The exact numbers won't transfer.

### Easy to process feels better

Psychologists Rolf Reber, Norbert Schwarz and Piotr Winkielman reviewed the evidence on **processing fluency**: how easily the brain takes something in. Their conclusion was that the easier something is to perceive, the more people tend to like it, and the more true and familiar it can feel ([Reber, Schwarz & Winkielman, 2004](https://pages.ucsd.edu/~pwinkielman/reber-schwarz-winkielman-beauty-PSPR-2004.pdf)).

Processing fluency: easy to see leads to feels familiar and true, which leads to feels more likeable

**Why this matters for ads:** Clutter doesn't just make an ad harder to read. It can make the brand feel less trustworthy. Simple, high-contrast, focused ads benefit twice: they're noticed, *and* they feel better.

**When it fails:** Fluency isn't the same as boring. A slightly surprising image can be what stops the scroll. The goal is a surprising idea, presented simply.

> **The quotable version:** In a feed, simple isn't a style choice. It's the price of being seen, and of being believed.

---

## Part 2: Five principles, with the reason behind each

Five principles: one focal point, big contrast, few words and large, the offer is a shape, faces and hands

### Principle 1: One focal point

**What it is:** One element that the eye clearly lands on first.

**Why it works:** Attention is competitive. When three elements have equal visual weight (similar size, contrast and colour), none of them wins quickly, and in a glance "not quickly" means "not at all".

Hypothetical before and after: the before version has a logo, photo, offer and feature list competing; the after version makes the offer dominant with the photo and logo smaller

**When it fails:** A single focal point that's the *wrong* thing. A beautiful lifestyle photo can win the glance while the offer, which is the actual message, loses.

**Ten-second check:** Blur the ad. Point to the first thing you notice. Is it the thing you most need people to notice?

### Principle 2: Contrast you can measure

**What it is:** The difference in brightness between text and its background.

**Why it works:** Brightness contrast survives blur, small size and poor screens. Colour differences alone often don't.

**How to measure it:** Web accessibility guidance (WCAG) uses a **contrast ratio**. It asks for at least **4.5:1** for normal text and **3:1** for large text ([W3C](https://www.w3.org/WAI/WCAG21/Understanding/contrast-minimum.html)). These standards weren't written for ads, but they're a useful, objective floor.

Contrast ratios: grey on grey 1.8 to 1 and white on orange 1.9 to 1 fail; white on navy 18.9 to 1 and navy on white 17.2 to 1 pass

**The counterexample that surprises people:** White text on a bright orange background looks bold and punchy in a design review. Measured, the orange in our example gives a ratio of only about **1.9:1**, roughly the same as grey on grey. The colour *feels* strong; the brightness difference is weak.

**Ten-second check:** Turn the ad black and white. If the headline sinks into the background, add a solid block behind it or darken the colour.

### Principle 3: Few words, set large

**What it is:** A short main line, big enough to read at arm's length on a phone.

**Why it works:** In Pieters and Wedel's study, text earned attention in proportion to its space. Fewer words means each one can be bigger. People glance at feed ads; they don't read them.

How text shrinks on a phone: 8 words readable, 15 words risky, 30 words should move to the caption

**What the platform says:** Meta removed its old "20% text" rule in 2020, so text-heavy images are no longer penalised in delivery. But it still advises keeping text short, clear and concise ([Social Media Today, 2020](https://www.socialmediatoday.com/news/facebooks-removing-its-restrictions-on-text-content-in-facebook-ad-images/)).

**When it fails:** Some products need words, such as a specific claim, a price or a list of what's included. Then keep *one* line large and move the rest to the caption.

**Ten-second check:** Shrink the ad to phone size. If you have to zoom in to read the main line, it's too small or too long.

### Principle 4: Make the offer a shape

**What it is:** Designing the offer so it's recognisable *before* it's read: a big number, a bold block, a price in a badge.

**Why it works:** The eye sees shapes before words. "Buy 3, Get 1 Free" in a big white block registers as "a deal" from its shape alone. "Under $20,000" is a number-shaped object.

**Ten-second check:** Blur the ad. Can you still tell *where* the offer is, even if you can't read it?

### Principle 5: Faces and hands, used with purpose

**What it is:** A person, or a hand holding or using the product, in the first frame.

**Why it's commonly recommended:** People are highly tuned to notice faces, and a hand using a product shows scale and use at a glance. Research on gaze cueing in banner ads ([Palcu et al., 2017](https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2017.00881/full)) also suggests that where a face *looks* can direct where viewers look.

**When it fails:** A face can steal attention from the offer. Pieters and Wedel found the picture grabs attention regardless of its size, so a striking face can win the glance and leave the message unseen. Use a face that looks at, or holds, the product. Avoid one that just smiles at the camera beside it.

**Ten-second check:** Is there a person or hand in the first frame? If there is, does it point attention *towards* the offer or away from it?

---

## Part 3: What long-running ads look like through a blur

We took ads that brands kept paying for, blurred them, and looked at what survived. Days live are from the Marxx ad library, October 2026.

| Ad | What survives the blur | Principles at work | Days live |
|---|---|---|---|
| **Wonderla**, "Buy 3, Get 1 Free" | A big swim ring, bright sky, and a white block where the offer sits | One focal point; offer as a shape; high contrast | 99 days, 35-44 copies each |
| **Withnell Hyundai**, "9 Used Cars Under $20,000!" | A single bold chrome shape on black in the opening frame | One focal point; extreme contrast | 164 days |
| **Porcellia**, "I grew my D2C brand to 30L..." | A heavy black text block on flat purple | Offer as a shape (the number); high contrast | 135 days |

**How to read this honestly:** These ads lasting doesn't *prove* the principles caused it. They might have lasted because of the offer, the audience or the budget. But they're consistent with the research in Part 1, and you'll rarely find a long-running static that fails all five checks.

Marxx creative board grouping long-running ads by hook, angle and format

---

## Part 4: Diagnose before you redesign

When a static ad underperforms, the instinct is to redesign everything. A better approach is to find which part failed. Each symptom points to a different cause.

Diagnose a weak static ad: no one stops means no focal point; stop with no action means the offer isn't visible; clicks with no sales means the visual and page don't match; weak in Reels means text under the overlays; still weak means the idea, not the design

| Symptom | Likely cause | Check |
|---|---|---|
| **No one stops** (low engagement for the spend) | No clear focal point | Blur test |
| **People stop but don't act** | The offer isn't visible as a shape | Can you find the offer when blurred? |
| **Clicks but no sales** | The visual promises something the page doesn't | Compare the ad and the landing page side by side |
| **Fine in feed, weak in Reels** | Key text under the app's overlays | Ads Manager preview in every placement |
| **Passes every check, still weak** | The idea, not the design | Test a new hook or offer, not a new font |

**The last row matters most.** Good design makes a good idea visible. It can't rescue a weak one. If an ad passes all five checks and still fails, stop redesigning and test a different message.

### Worked example: Fixing a cluttered static

*Hypothetical example.*

**The ad:** A skincare brand's static. A lifestyle photo of a woman at a window, the logo top left, "Clinically proven hydration for 72 hours. Dermatologist tested. Vegan. Free shipping over INR 999" in small white text over the photo, and a "Shop now" button.

**The diagnosis, principle by principle:**

1. **Focal point:** Blurred, the photo wins. The product is barely visible.
2. **Contrast:** White text over a bright window measures well under 3:1.
3. **Words:** 17 words, all the same size.
4. **Offer as a shape:** There isn't one. "72 hours" is buried mid-sentence.
5. **Face:** She's looking out of the window, away from everything.

**The fix:**

| Before | After |
|---|---|
| Lifestyle photo as the focal point | A hand holding the product, large |
| 17 words, same size | "72-hour hydration" huge; everything else in the caption |
| White text on a bright photo | Dark text on a solid light block (over 7:1) |
| No visible offer | "72h" as a bold badge |
| Face looking away | No face; the hand shows use and scale |

**How to know it worked:** Run both with the same budget and audience. Judge on cost per result. For a static, also compare the click-through rate *alongside* the conversion rate, because a clearer ad should attract fewer but better-qualified clicks.

---

## Which tools are best for designing ad creatives?

The principles matter far more than the software. Any of these can produce an ad that passes all five checks.

| Tool | Best for | Helps with |
|---|---|---|
| **Canva** | Fast statics, carousels and resizes from templates | Speed; change templates enough to stand out |
| **Adobe Express** | Teams sharing brand kits | Consistent colours and type |
| **Figma** | Designers building systems and many variants | Precise layout, reusable components |
| **Any contrast checker** (many are free online) | Measuring text contrast | Principle 2, objectively |
| **Meta Ads Manager preview** | Seeing real placements and overlays | Catching text hidden in Reels and Stories |
| **Marxx generation canvas** | One brief to finished ads in every placement size | Starting from hooks and formats that are lasting in your category |

*Based on each tool's public features, October 2026.*

Marxx trending tactics showing which hooks and formats are rising in a category

---

## Six mistakes, and how to spot them

| Mistake | Symptom | Fix |
|---|---|---|
| Judging contrast by eye | Bright colours that blur into mush | Measure the ratio; aim for 4.5:1 or better |
| A beautiful photo as the hero | People remember the image, not the offer | Make the offer the focal point |
| Writing the whole pitch on the image | Tiny text nobody reads | One line on the image; the rest in the caption |
| A bigger logo "for awareness" | Less room for the message | Pieters and Wedel found a bigger brand didn't win more attention |
| A face looking away | Attention leaves the ad | Use faces that look at or hold the product |
| Redesigning a weak idea | Prettier ads, same results | If it passes all five checks, change the message |

## What to take into your next creative review

- **Design for what's left after a glance.** Blur it, shrink it, and see what survives.
- **The picture grabs attention at any size; text earns it with space.** Spend your space on words that matter.
- **Measure contrast.** Bright colours often fail. White on orange can be as weak as grey on grey.
- **Simple ads are seen, and believed.** Processing fluency links ease with liking and trust.
- **Design reveals ideas; it can't replace them.** When every check passes, change the message.

## FAQ

**What makes a good visual ad for social media?**
One focal point, strong measurable contrast, a short headline set large, an offer you can see as a shape, and a face or hand that directs attention towards the product.

**Which tools are best for designing digital ad creatives?**
Canva for speed, Adobe Express for brand kits, Figma for design systems, a contrast checker for text, and Ads Manager's preview to check placements. The Marxx canvas turns one brief into finished ads in every size.

**How much text should be on a Facebook ad image?**
Meta no longer penalises text-heavy images, but it still recommends short, clear text. Keep one main line large and move the rest to the caption.

**What contrast ratio should ad text have?**
A useful floor is the accessibility standard: 4.5:1 for normal text and 3:1 for large text.

**What's the squint test?**
Blur or squint at your ad. If you can still tell what it's selling and where the offer is, it will work at feed size.

---

## Make ads that pass the squint test

**Trial the Marxx generation canvas:** Start from what's lasting in your category and get finished ads in every placement from one brief.

Marxx video node assembling an ad

**[Trial the canvas: Turn one brief into a finished ad ->](https://www.marxx.ai)**

*Related: [Best Tools for Designing Facebook Ad Mockups](/blog/facebook-ad-mockup-tools) - [How to Design High-Performing Ad Creatives](/blog/high-performing-ad-creatives) - [Instagram Carousel Ads: 12 Story Structures](/blog/instagram-carousel-ad-ideas)*

### Sources

- University of Michigan (2004). [Print ad pictures: Less is more](https://news.umich.edu/print-ad-pictures-less-is-more) (Pieters & Wedel, *Journal of Marketing*, 2004).
- Reber, R., Schwarz, N. & Winkielman, P. (2004). [Processing fluency and aesthetic pleasure](https://pages.ucsd.edu/~pwinkielman/reber-schwarz-winkielman-beauty-PSPR-2004.pdf). *Personality and Social Psychology Review*, 8(4).
- W3C. [Understanding WCAG: Contrast (minimum)](https://www.w3.org/WAI/WCAG21/Understanding/contrast-minimum.html).
- Social Media Today (2020). [Facebook removes restrictions on text in ad images](https://www.socialmediatoday.com/news/facebooks-removing-its-restrictions-on-text-content-in-facebook-ad-images/).
- Marxx ad library, October 2026. Days live and copy counts were accurate on 25 October 2026.

---
- [More Creative strategy articles](https://marxx.ai/posts/category/creative-strategy)
- [All articles](https://marxx.ai/posts)