How to overcome perceptual biases with UI color choices?

Hey everyone,

We're a small SaaS team, and we just launched a new analytics dashboard feature. Our main goal is to make the data visualizations and key calls-to-action (CTAs) instantly understandable and engaging, driving users to explore deeper insights. We really want our users to have a smooth, intuitive experience from the get-go.

Despite careful design and following common color theory principles, our initial A/B tests for button colors (e.g., 'Upgrade Now', 'Generate Report') and data highlight colors are showing inconsistent and often counter-intuitive results. Users aren't converting as expected, and feedback suggests some visual elements aren't conveying the intended urgency or importance. We suspect we're running into strong perceptual biases that our current color scheme isn't addressing effectively, which is really impacting their visual perception of key actions.

Here's what we've tried so far:

• Tested standard red/green/blue for success/danger/neutral CTAs.

• Experimented with various shades and tints, focusing on brand-aligned palettes.

• Used tools like Coolors and Adobe Color to ensure harmony and contrast ratios.

• Conducted internal user testing with small groups, but unfortunately, results didn't scale to our live A/B tests.

And here's where things got tricky (unexpected results):

• A vibrant green 'Upgrade' button performed worse than a muted blue one, which was totally contrary to common marketing advice we've always heard.

• Using red to highlight 'critical alerts' in data sometimes led to users ignoring them completely, possibly due to 'banner blindness' or an unfortunate association with errors rather than urgency.

• Certain data visualization colors meant to show positive trends were perceived as negative by some user segments, indicating cultural or learned perceptual biases that are affecting their data interpretation.

So, I have a few specific questions and would really appreciate any help or insights:

• Are there specific frameworks or methodologies for selecting UI colors that account for cognitive biases beyond basic color theory?

• How do experienced marketers/designers approach A/B testing colors when initial results are contradictory and confusing?

• Any recommended resources, tools, or expert advice on leveraging 'cognitive color' principles to actively counteract undesirable perceptual biases in a SaaS UI and improve overall visual perception for better UX?

• What's your experience with cultural differences in color perception, especially for a global user base? This is something we're increasingly thinking about.

Really keen to hear from anyone who's tackled similar issues or has some solid advice. Thanks in advance!