Understanding VAI Metrics

Plain‑language explanation of Hold, Speed, Reach and the combined Visual Attention Index (VAI)

✓ VAI Composite Score ✓ Time-Based Analysis ✓ Attention Analytics ✓ Statistical Insights
AI attention prediction heatmap
Original image before AI prediction
<
>

Interactive Demo: Drag the slider to see how VAI metrics transform raw images into attention insights

Visual Attention Index (VAI)

VAI is our comprehensive composite metric that combines Hold, Speed, and Reach measurements to provide a single, actionable score representing the overall attention-grabbing and retention power of any visual element

VAI = Balanced Blend of Hold, Speed & Reach

A single score summarizing: How long people look (Hold), how quickly they first notice (Speed), and how many people notice at all (Reach).

What Each Metric Means

⏱️ Hold

Average time (dwell) people keep their eyes on an area once they look at it. Higher Hold = people spend longer actually looking there.

⚡ Speed

Average time it takes people to first notice an area. People who never look are counted as “slow” (they add a large time), so slow discovery and being missed both lower Speed.

👁️ Reach

Percentage of participants who ever looked at the area at least once. 100% Reach means everyone noticed it; 50% means only half did.

📊 VAI

Single combined attention score balancing holding power (Hold), notice speed (Speed) and visibility coverage (Reach). Use it to rank elements at a glance.

Understanding Attention Components

Scores are comparative. A higher score simply means an element wins that aspect versus others in the same image. Focus on rankings and gaps, not the absolute number. A Hold of 22% can be great if everything else is below 15%.

Comparative Scoring System

Attention scores are relative metrics designed for comparison between elements (AOIs) within the same analysis. They range from 0% to 100%, where higher numbers mean better performance relative to the others.

Key principle: The element with the highest score in a component performs best in that aspect—its absolute percentage alone is not the point. A Hold score of 25% may be excellent if everything else is lower, meaning it keeps attention longer than any other element.

Interpretation tip: Think of each metric as a race. The top score is the winner; the distance to the next scores shows competitive advantage. Focus on ranking and score gaps, not whether a number “looks big enough.”

Example: If three headlines have Hold scores of 25%, 18%, and 12%, the first headline clearly retains attention best—even though 25% might sound modest out of context.

VAI

Overall winner. Use to spot your natural focal points and low performers needing redesign.

Speed

Early capture. High = noticed quickly. Helps validate intended viewing order.

Reach

Audience coverage. High = most people will at least glance at it.

Hold

Retention depth. High = people stay and process content.

Interpreting Time Ranges

Each time range is cumulative (includes everything before it). Use them to separate instant grab from sustained engagement:

  • 0–500ms: Snap impact / what pops first
  • 0–1000ms: First-second discovery & hierarchy check
  • 0–3000ms: Short-term engagement & scanning pattern
  • 0–5000ms: Medium evaluation – secondary content emerges
  • 0–10000ms: Full exploration – late discoveries & depth

Core Attention Metrics (End‑User View)

⏱️

Hold

Average Dwell Time

How long, on average, people keep their eyes on this area when they look at it. Longer Hold = deeper processing or interest.

What a Higher Hold Suggests:

  • People are actually reading or inspecting it
  • The content sustains attention after being found
  • Good for detailed claims, copy or product shots

Interpretation:

  • High Values: Engaging, compelling content
  • Low Values: Quickly scanned or overlooked
  • Package Design: Logo, product image, key claims
  • Web Design: Content engagement, call-to-action effectiveness

Time Analysis:

500ms 1000ms 3000ms 5000ms 10000ms

Analyze attention evolution over time

Speed

How Fast It Gets Noticed

Average time until first look. People who never look count as very slow, so missing the area drags Speed down. Higher Speed = earlier discovery.

What a Higher Speed Suggests:

  • Strong immediate visual pull
  • Clear hierarchy / placement
  • Good for logos, CTAs, headlines

Interpretation:

  • High Values: Immediately attention-grabbing
  • Low Values: Noticed later in exploration
  • Package Design: Logo recognition, visual hierarchy
  • Web Design: Button prominence, headline impact

Speed Categories:

Fast (0-1s): Immediate impact
Medium (1-3s): Quick discovery
Slow (3s+): Later exploration
👁️

Reach

How Many People Noticed It

Percentage of participants who looked at the area at least once. Higher Reach = broader visibility.

What a Higher Reach Suggests:

  • Placed where eyes naturally travel
  • Visually clear and distinguishable
  • Effective for broad awareness goals

Interpretation:

  • High Values: Broadly visible to most viewers
  • Low Values: Missed by significant portion
  • Package Design: Brand visibility, message delivery
  • Web Design: Navigation discoverability, content reach

Visibility Benchmarks:

90%+ Excellent visibility - Universal notice
70-90% Good visibility - Majority notice
<70% Needs optimization - Many miss it

Technical Specifications

Statistical Measures

For any selected Area of Interest (AOI), we provide:

  • Average: Mean cell attention score within AOI
  • Maximum: Highest cell attention score in AOI
  • Minimum: Lowest cell attention score in AOI
  • Standard Deviation: Attention distribution variability
  • AOI Score: Metric-specific calculation:
    • Speed & Reach: Average cell value × 100%
    • Hold: Percentage of total attention captured
    • VAI: Average of Speed, Reach, and Hold AOI scores

Time Range Analysis

500ms: Initial visual capture 1000ms: First second exploration 3000ms: Short-term engagement 5000ms: Medium-term patterns 10000ms: Comprehensive viewing

Each time range provides cumulative attention data from 0ms to the specified limit.

Industry Applications & Use Cases

📦 Product Packaging

VAI: Overall package shelf effectiveness
Hold: Brand name engagement and message processing time
Speed: Logo recognition speed and visual hierarchy
Reach: Key information visibility coverage

Optimization Goals:

  • Logo Speed > 0.7 (fast recognition)
  • Brand name Hold > 0.5 (sustained engagement)
  • Key claims Reach > 80% (broad visibility)

🌐 Web Design & UX

VAI: Page element importance and effectiveness
Hold: Content engagement depth and reading time
Speed: Call-to-action prominence and navigation speed
Reach: Content discoverability and accessibility

Optimization Goals:

  • CTA buttons Speed > 0.8 (immediate notice)
  • Key content Hold > 0.6 (reading engagement)
  • Navigation Reach > 90% (universal access)

📢 Advertising & Marketing

VAI: Ad component performance and overall impact
Hold: Message retention potential and engagement
Speed: Headline impact and initial attention capture
Reach: Brand visibility and message delivery coverage

Optimization Goals:

  • Headlines Speed > 0.75 (instant impact)
  • Brand logo Reach > 85% (recognition)
  • Key message Hold > 0.4 (processing time)

🛍️ E-commerce & Retail (soon)

VAI: Product listing effectiveness and conversion potential
Hold: Product image engagement and detail examination
Speed: Price and offer visibility in search results
Reach: Product information and reviews discoverability

Optimization Goals:

  • Product images Hold > 0.6 (exploration)
  • Price/offers Speed > 0.7 (quick find)
  • Buy buttons VAI > 0.8 (conversion focus)

A/B Testing & Optimization Workflow

1

Baseline Analysis

Generate VAI metrics for current design

2

Identify Issues

Find low-performing elements using individual metrics

3

Design Variations

Create alternatives targeting specific metrics

4

Compare Results

Analyze improvements across all attention metrics

Ready to Apply These Metrics to Your Designs?

Discover how VAI, Hold, Speed, and Reach can transform your visual optimization process with data-driven insights

📊

Statistical Analysis

Get detailed metrics with avg, min, max, std dev for every AOI

🎯

Time-Range Analysis

Track attention evolution from 500ms to 10 seconds

🔬

Advanced Analytics

Based on comprehensive attention modeling and analysis

Book a demo to see our web tool for instant attention predictions, or learn about API integration for automated workflows