whatnotThe Voice of your Customers Analyzed
This dashboard is live. Use the filters above (platform, year, theme) or click any theme bar, and every stat, chart, and review recomputes instantly. Go explore.
A job application, in dashboard form

This is my job application: hire me for your open Data Scientist roles.

I'm Jacob Cacheria. Every number on this page comes from a data pipeline I built: 34,982 public written reviews web scraped of Whatnot's iOS and Android App Store reviews (Dec 2019 to Jul 2026), collected with Python via public app-store APIs, audited to zero data-quality defects, classified into complaint themes, and rendered with hand-built SVG.

Skills in play: Python and pandas ETL, web scraping, advanced SQL, data-quality auditing, KPI and metric design, A/B testing and causal-inference thinking, data visualization, and communicating insights to technical and non-technical audiences.

The short version: problems in your data, and my fixes

Product insights ↗

What I foundApril 2026 ratings dipped to 3.43★ (37% negative), and your users named the cause: an 835-upvote review describes a UI update that blocks the stream view during shows.

What I would doPut review themes in release QA as guardrail metrics, so the next regression dies in staged rollout instead of your reviews.

Why it helpsIt turns a lagging signal into a leading one: a bad release is caught at partial rollout instead of dragging the App Store rating down for a month, protecting new-user downloads and the roadmap's credibility.

With your data, I could alsoShipping just became the #1 complaint (10.6% of reviews, up 1.1pp) while scams recede. I'd join complaint rates to order-level delivery times and costs, size the fulfillment fix, and re-rank product priorities by measured user pain.

Customer Experience insights ↗

What I foundYour support team replies same-day (median) but covers only 12.7% of reviews, and 303 users edited their review after receiving a reply. Replies visibly move sentiment.

What I would doMake reply coverage, lag, and post-reply re-rating owned CX KPIs, so you can see the effect each reply has on the customer.

Why it helpsIt lets CX prove its own ROI and route effort where it moves the needle: once you know what a reply is worth by theme, you can shift reply capacity toward the complaints where it recovers the most rating and retention.

With your data, I could alsoThe share of reviews mentioning being "banned" climbed from 1.9% to 2.5% of all reviews year over year (2025 to 2026). With ticket data I'd forecast the appeals queue before it spikes, set agent-team priorities, and define an appeals-experience metric.

Finance & Payments insights ↗

What I foundRefund pressure is climbing: the share of reviews asking for their "money back" is up about a quarter year over year (0.7% to 0.8% of all reviews), and 2026's #2-upvoted complaint (230 upvotes) is not seeing shipping costs before purchase.

What I would doTrack refund language as a leading indicator of refund cost, and test upfront price transparency.

Why it helpsIt gives finance an early-warning gauge on a real cost line, so refunds get forecast and intervened on before they land, and earlier shipping-cost disclosure lifts both checkout conversion and contribution margin.

With your data, I could alsoCheckout complaints are rare (1.9% of reviews) but 83% negative. With payments data I'd build failure-severity dashboards tied to GMV at risk: the actionable KPI your finance and payments partners review weekly.

Rating momentum

Monthly average rating and review volume · low-volume months omitted
Avg rating (left) Review volume (right)

What people talk about

Share of reviews mentioning each theme · click a bar to filter
Trust cluster Mostly praise Other

Where Whatnot replies

Developer-reply rate by theme · grey line = overall rate in view

In their words

Top 5 most-upvoted reviews in the current view

Let's start working together

Everything on this page came from limited public data, only the reviews anyone can read. With your internal data (orders, support tickets, payments, and session and experiment logs) I could turn these signals into owned KPIs, live dashboards, and product changes I can actually test. There's far more to find once I'm on the team.

Method. 34,982 public written reviews of Whatnot's iOS and Android apps (Dec 2019 to Jul 2026), with upvote counts and developer replies, collected via public app-store APIs. Written reviews are the vocal minority: Whatnot holds ~896K App Store and ~290K Google Play star ratings (~4.7 avg); only these ~35K users wrote text (~3.5 avg). This measures what motivated users say, not average satisfaction. Theme tags are a keyword first pass. Built by Jacob Cacheria.