What Drives Clicks? Insights from 58’s Feed Card Experiments

The product homepage feed largely shapes users' perception of the page experience; the article explores how to design a feed that captures users' attention.

During a recent redesign of 58’s homepage feed, a series of experiments were conducted to identify factors influencing click-through rates, especially for recruitment cards which showed a significant drop.

Why did recruitment clicks decline? Four design factors were identified: image quality, content richness, card height, and text style.

Using a double‑diamond model, the team pinpointed key user concerns through prior research, ranking them as follows: 1) job authenticity (title and requirements), 2) benefit superiority (salary, perks, environment), 3) commuting convenience, 4) company reliability.

Image Experiment: “Real” Beats “Imagined”

Four variants were tested:

Version 1 – Fixed large‑banner image.

Version 2 – Small avatar image with overlaid title and salary.

Version 3 – Key fields covered with background color.

Version 4 – Real storefront image.

Results showed that users preferred authentic storefront images, achieving higher UV‑CTR and connection‑to‑click ratios compared to the other versions.

Benefit Experiment: “Fine” Beats “Vague”

The second phase focused on benefit information, testing the impact of benefit tags and recommendation copy.

Adding recommendation text and benefit tags significantly improved click performance, confirming that users pay more attention to salary and perk details than anticipated.

Position Experiment: “Intuitive” Beats “Vague”

The third phase examined how displaying job location influences clicks. Two approaches were tested: textual position cues (subtitle‑style) and map thumbnail images.

Preliminary data is still being collected, but the hypothesis is that a clear map background will outperform a simple textual description.

Feed Card Experiment Database

All experiment records, inferences, and conclusions have been compiled into a shared Feed Card Experiment Database, serving as a knowledge base for future design‑product collaborations.

Fixed‑Cycle Iteration Mechanism

The team established a three‑week iterative cycle: designers propose solutions, product provides data feedback, and the team evaluates whether experiment goals were met. If results diverge from expectations, new visual or interaction ideas are tested in the next cycle, ensuring continuous, agile improvement.

Beyond the data, the article reflects on the fluid nature of design, emphasizing that both self and design are constantly evolving, and advocates building mechanisms that allow designs to be continuously tested and refined.