Emotional messages written by AI hurt word-of-mouth by up to 24% and make people see the company as dishonest.
People are up to 47.8% more forgiving of AI’s mistakes when the AI uses self-deprecating humor (e.g. “My non-brain just froze”) vs no humor.
How to design an ecommerce AI chatbot your customers will love, using insights from our Wharton Blueprint for Effective AI Chatbots.
Optimizing your website for AI buying agents can drastically increase the likelihood they choose your product (e.g. tweaking a product title increased choice by 80 percentage points).
Research on 164m sales across 973 ecommerce websites finds that AI traffic drives more revenue per session than paid social, but less than any other channel. This gap is smaller for complex products.
People are up to 29% less willing to buy products designed by AI (e.g. a chair, a snack flavor), compared to those designed by humans.
People are up to 30% more likely to stop using a service if it’s delivered by AI or robots (vs a human) when warmth is important to satisfaction (e.g. hospitality).
How to combine an automated phone tree with human agents to cut costs while delivering a great experience.
People with a limited understanding of how AI works are up to 31% more likely to use it because they see it as magical.
Customers were up to 49.7% more satisfied, and more likely to buy when AI chatbots use interjections (e.g. wow, aw, oh!) because they felt better understood.
People are 18.5% less likely to act unethically (e.g. try to commit fraud) with AI chatbots that have human-like features (e.g. a name and avatar, speak less formally).
People dislike AI chatbots (vs humans), unless you frame it right - or pass on the savings.
