A major live-streaming platform · Creator analytics · 2025

Creators didn't need more numbers. They needed to know whether to go live tonight.

The creator back office only reported what creators did each day. I led research and design, interviewed creators, tested prototypes with them, and over three versions turned a report card into a go-live tool.

Role

Product Designer (freelance): interview guide, running interviews, prototype tests, IA, UI

Team

PM, in-house designer (design review)

Duration

May to July 2025

700M+

Yearly visits (platform scale)

3 versions

Iterations

4 → 1

Pages merged

Final first screen: a go-live timing card and new creator tasks, with the timing card enlarged on the right
The final version after testing. The first card answers one question: is now a good time to go live? Screens are de-identified under NDA.

The brief

“Make the creator analytics usable.”

It sounded like a cleanup job. Then I opened the old back office page by page, and the problem was the order of the information. Try finding the answer below.

Friday, 7 pm. How many more streaming hours do you need this month to hit the hours bonus?

Find it in the old back office. Use the page buttons on the right.

Old activity page
Not here

It only shows minutes streamed per day. You add up the month yourself.

Opened 1 / 4

Research

Before locking the design, I asked creators

I listed the old back office's problems and looked at two similar platforms. Then I wrote the guide and interviewed two creators, one focused on shows and one on fans. Each session started with how they decide, then moved to the prototype.

Bonus conditions come first. Have I hit the bonus? Have I hit the campaign hours?Creator B, on deciding to go live last night

Reframing

I reordered the back office around one daily decision

Every day, creators decide whether to go live tonight and for how long. That depends on how far they are from this month's goal and bonus threshold.

How I rewrote the problem

Give creators more complete dataHelp creators decide about tonight as soon as they open it
Group by data typeOrder by how they decide: goal, sources, reference

Design decisions

Three versions, each closer to how creators decide

Three phones: the May 29 first version, the June 5 tested version, the July 3 version after testing
Left to right: May 29 first version, June 5 tested version, July 3 version after testing.
May 29: tidy up

I merged four pages into one, with sections and trends.

June 5: goal first

Interviews showed creators decide by bonus thresholds and hours, so I moved this month's goal to the top and tested it.

July 3: answer “now or not”

After testing, the first card says if now is a good time to go live.

Prototype testing

The goal got noticed, and questioned

It lays out exactly the earnings and goals creators care about, with trends like a stock app.Creator B, first impression after 5 seconds
What creators saidWhat I changed
How is the goal calculated?The same creator suggested covering it in onboarding, not on screen. So the operations team explains it, and the UI keeps a help link.
I'd rather set my own goalPlanned for the next version: platform goals alongside personal ones.
I can't see how many people are onlineAdded on July 3 as the first card: competition level, fans online, audience activity.
I mostly check at month endThe default range is now 30 days.
Diamond earnings matter mostDiamond earnings moved to the top.

“The dashboard is the result. The live stream itself matters more.” Creator A

Her problems were stream quality, filters and failed photo uploads in chat. A back office can't fix those, so I didn't force them in. I wrote them up as a separate list for the PM.

After launch

Three numbers I'd watch after launch

Delivered and now being built. No launch data yet; these three numbers would tell me if it works.

Weekly opens

Has it become a daily tool?

Going live within an hour

Does checking the goal lead to a decision?

Monthly goals hit

Are more creators reaching their goal?

Reflection

I nearly built a pretty, useless analytics page

My first version grouped the data well but put it in the wrong order. I sorted by data type, while creators read in the order they decide. Next time I'll ask creators to rank what they look at first, before showing any prototype, so the prototype doesn't steer their answers.