Is a Picture Really Worth a Thousand Words? – On the Role of Images in E-commerce

Is a Picture Really Worth a Thousand Words? – On the Role of Images in E-commerce

Wei Di, Neel Sundaresan, Robinson Piramuthu, Anurag Bhardwaj eBay Research Labs, eBay Inc., San

Jose, CA

http://labs.ebay.com

Image: Significant Variation in C2C market

2

Convey unique VISUAL information

A reflection of TRUST-WORTHY marketplace

Changing landscape of e-commerce

C2C Market: Variation in Images

Standard merchandising imagesMajority in Individual Seller dominated Market

Goal

3

Image Factors: Display Format Quantity Image Quality

E-market: Category User Behavior Conversion …

Systematic Detailed Study on:

Difference across Categories, Sellers, Buyers? Difference across Display format, Quality, Quantity? Always improve the Conversion Rate? Compared to Other Selling Variables?

Overview Study: Quantity

4

- Conversion rate doubled twice when using two images as compared to one. - Using more than two photos do not help.

Image Quantity vs. Conversion Rate

Overview Study: Quality vs. User

5

Buyer – Do they care?

A: most frequent buyerE: least frequent buyer

Experienced buyers - Less sensitivity Inexperienced buyers - More concern about the quality of image.

Transaction data

Case Study: Resale – Profiting from Better Pictures

6

Strong correlation between increment in photo count and profit rate, not image quality.

Images are copied from the initial sale, and modified by cropping and scaling.

Image Quantity vs. Image Quality

Case Study: Better Image Leads to More Watch?

7

Positive correlation of “Watch” per item vs. Image Quality

Image Quality vs. User’s behavior (attention)

Apparel based category (left) more impacted by image quality than category

(right) of more unique product.

Action Model: A Regression Explanation

8

Table 2 Logistic Regression Results for “Clothing, Shoes & Accessories”.

Jointly understanding image impact with other selling dimensions

Three factors (Format, Quantity, Quality) exhibit significant impact on products that rely more on visual.

Conclusion & Future Work

9

First systematic study jointly considering multiple key e-commerce

dimensions.

Positive evidence - images increase buyer‘s attention, trust, conversion rate,

but vary in it’s own dimensions (display, quantity and quality).

Image quality matters more for products rely more on visual information, but

shows ambiguous results on other categories.

Future work includes content based image quality to understand the impact

of image aesthetics on attracting buyer‘s attention.

10

Thanks