Case study, since September 2026
From no reporting to a daily scorecard in two weeks
I built sales reporting from scratch for Liquidity Services’ live-selling team. Since then the work has grown into show merchandising, inventory and oversell control, analysis and content support, across about 25 TikTok Shop and 40+ Whatnot shows a week.
- 2 wks
from first build to daily reports on both platforms
- 787
live shows tracked, 307 on TikTok and 480 on Whatnot
- 5,183
TikTok products with full sales history, 17,867 product-level sales rows
- 508
bad listings removed after auditing about 65,000 Whatnot listings
The challenge
Liquidity Services sells through daily live shows on TikTok Shop and Whatnot. When I joined, there was no reporting process. Sales lived inside each platform’s dashboard, and a shared Google Sheet was filled in by hand when someone had time.
No daily view
Nobody could see how yesterday’s shows did without logging into each platform and adding it up.
Error-prone numbers
Hand-typed figures in the shared sheet drifted from the platforms’ own numbers.
No host view
There was no way to tie each show’s sales to the host who ran it.
Merchandising by memory
Producers built product lists without a clear view of what actually sold.
What I built
A custom reporting system turns the platforms’ data into ready-to-send reports every morning. I review every email before it goes out.
- Daily
Results email
Sales, orders and average order against target, top show, best seller and host performance, for TikTok and Whatnot. Built to read on a phone.
- Live
Sales dashboard
Every show in one view: sales, items sold, new followers, host and moderator.
- Monday
Weekly deck
One deck per platform, 10 slides for TikTok and 7 for Whatnot. Each opens with what happened, why, and what to change next week.
- Weekly
Show scorecard
Every show ranked per hour on air, next to its own result last week, with a flag for shows that stay near the bottom several weeks running.
- Daily and weekly
Host scorecards
Performance per hour for each host, matched from the real staff schedule so hosts are compared on equal time.
- Daily
Top sellers and underperformers
Top 10 products, zero-sales items, low-conversion items and repeat non-sellers.
- Weekly
Week-over-week drivers
Breaks every change in weekly sales into its causes: number of shows, orders per show and order size.
- Weekly
When shows sell
A day-and-time grid, so the schedule can be built around the slots that pay.
- TikTok
Viewers to buyers
Where each show loses people: not getting watched, not clicking, or clicking without buying.
- Whatnot
Buyers and giveaways
First-time versus returning buyers by show and host, and whether giveaways bring in new buyers.
Why not just use the platform dashboards?
TikTok and Whatnot show each show’s numbers, but only one show and one platform at a time. The work is in connecting them.
| Question | Platform dashboard | This system |
|---|---|---|
| How did every show do yesterday, on both platforms? | Separate logins, one show at a time | One email each morning |
| Which host sells the most per hour? | Not tracked; dashboards don’t know who hosted | Host scorecards from the staff schedule |
| Which products keep selling, and which never do? | One show’s products at a time | Sales history across every show since day one |
| What should go on tomorrow’s show? | Nothing | Lists built from proven sellers |
| Did we hit our average order and items-per-hour targets? | No targets | Every report measured against your targets |
| Do the numbers in our own sheets match the platform? | Nothing | Mistakes flagged against platform data |
Who it’s for: a solo seller running one show a day is fine with the dashboards. This is built for teams running several shows a day with multiple hosts.
It also catches what a manual process misses
A missing week
Shows from one week were missing from the historical files. They were recovered and backfilled.
Typos in the shared tracker
Mistyped sales figures were flagged automatically against the platform data.
Late-settling orders
TikTok keeps revising a show’s totals for a few days after it airs. Every refresh re-checks recent days instead of trusting the first number.
Wrong host rows
The shared tracker often still held last week’s copied schedule. Hosts now come from the staff scheduling app, matched to each show by time.
“In two weeks we went from no reporting at all to a daily scorecard for TikTok and Whatnot. I didn’t think that was possible that fast.”
Then the work grew past reporting
Once the numbers were trusted, the team started using them to decide what goes on air, what gets posted and what to do when results move.
- TikTok, 2 days ahead
Show merchandising
I build the product list for every TikTok show. Top sellers only: an item makes the list because it has sold on that same show before, or is selling well across all shows. The lists were tuned on producer feedback until they could be used as delivered.
- On request
Analysis
When results changed and nobody knew why, I compared before and after per hour on air, by traffic source, and was clear about what the data couldn’t separate. When the team suspects something, it gets checked against the data first.
- Weekly
Content picks
A short list of what to post and feature each week, pulled from what sold in the shows, plus promo posts, captions and video prompts for hosts.
- On request
Promo picks
Ranks a promo list against sales history: push, hold, cut price or test.
Inventory and oversell control
Keeping about 65,000 Whatnot listings in line with real stock
Liquidity Services lists the same inventory on several marketplaces at once. When a unit sells in one place while it still shows as available in another, the second sale becomes a cancelled order. I audited every Whatnot listing against the store’s inventory, cleaned up what didn’t match, and traced where oversells come from.
- ~65,000
active Whatnot listings compared against the store’s inventory
- 508
bad listings removed across two passes, each one approved first
- 53,796
Whatnot orders analyzed over four weeks
- 3
separate causes of oversells identified
The audit
Every active listing checked against the inventory system. It surfaced listings showing stock that didn’t exist, items listed twice, quantity mismatches, and in-stock items missing from Whatnot.
The cleanup
I built the removal lists and got them approved before anything was touched. Afterward I verified that none of the removed listings came back and the rest matched.
The re-check that paid off
The first pass kept the oldest copy of each duplicate. The re-check showed the oldest copies were the ones no longer updating, so the rule was reversed. That is what the re-check is for.
A weekly check
A read-only check now runs every Monday and produces a short list of new problem listings for approval. Nothing is removed without sign-off.
Why oversells happen
Individual oversells were traced minute by minute across three systems. Three causes: a unit sells on another marketplace first, a cancelled order returns a unit to stock and it sells again, and stock split across two locations hides a shortage.
Fixes tested before rollout
A proposed fix was tried on a single order first. The test showed it wasn’t safe on its own, so it never ran at volume, and a step-by-step procedure was written for the safe version.
What I can set up for you
The system is already built. It works from the exports you already download from the platforms you sell on, and setup takes about a week.
- Reporting
Know your numbers
Daily results email, Monday deck, product performance, and what to change next week. Show and host scorecards for live teams.
- Reporting + merchandising
Numbers and what to feature
Everything in Reporting, plus the product list for every show, built from proven sellers.
TikTok Shop only
- Full operations support
The whole operation
Everything above, plus inventory and oversell control, analysis on request, and content picks and social support.