Product

The advisor, and the analysis it stands on.

The landing page makes three claims: StoreVitals tells you why your shop changed, tells you whether your fix worked, and answers questions from your real numbers. This page is the depth behind each, then the measurement layer underneath, which is there to make the advisor right, not to be the product.

Part one

The advisor

Four screens that turn the shop's own data into a cause, a fix, a verdict, and an answer.

Diagnosis

Why revenue moved

Every change in revenue is decomposed, listing by listing, into price, volume and mix: parts that add up to the change exactly, with a bucket for listings that entered or left. Detectors then look for a cause and rank what they find by dollar impact.

Each finding is labelled measured or modelled, and says which half is which.
A finding on too few orders says so instead of guessing; a detector that couldn't run is listed, not silently skipped.
Detectors cover a fading top earner, stale listings, price resistance, a rating drop, ads that stopped, a channel drying up, and the shop's own recent edits.

Diagnosis: revenue down 15% versus baseline, biggest driver conversion; a 'What moved' chart by driver and the revenue bridge.

Actions

What to do, ranked by dollars

Every recommendation carries a dollar estimate, a confidence, and an effort level. They're split by how fast Etsy can show the effect: this week for levers like price, photos and shipping, and next two weeks for title and tag changes that wait on Etsy search re-indexing.

Mark done, snooze, or dismiss. Done actions are linked to the listing edit that followed, so the verdict comes back to the action.
"Draft a fix" writes the title, tags or description for you, validated against Etsy's limits, previewed, never auto-published.
A loss is an action too: "Put back the change that cost you" ranks beside every win.

Actions: a ranked bar chart of expected impact, then 'This week' and 'Next two weeks' columns, each action with a dollar estimate, confidence and effort, plus Mark done / Snooze / Dismiss.

Changes

Did it work?

Every edit to a listing (price, title, tags, photos, description) is recorded, and once enough time has passed it gets a verdict: Helped Hurt or Too early to tell. The comparison is against the rest of your shop over the same window, so a slow week isn't blamed on one edit.

Sessions, orders and conversion each get their own control-adjusted verdict; the headline is the one that matters for that kind of change.
A win on one listing becomes a suggestion for the ones without it, marked as modelled upside, not promised.
Recording starts the day you connect, on every plan. History that begins at upgrade time is worthless.

Actions: an action is marked done; then the Changes screen, where each edit on each listing carries a verdict: Helped, Hurt, Too early to tell.

Mark an action done, then see every edit's verdict on Changes. Recorded on the sample shop.

Ask

Answers from your own numbers

Ask a question in plain words. The assistant can only choose from a fixed set of lookups over your ledger and traffic, the same figures on your screens, and it lists which ones it used. It cannot write a query, and it cannot make a number up.

When the data can't settle a question, the answer says so and says what would.
Works per listing too: open any listing and ask about that one.

The model writes sentences; it never produces a figure. Every number in an answer was computed before the model saw it.

Ask panel: an answer naming the weakest listing with its revenue, orders and momentum, what the data can't confirm, and the sources used.

Part two

The measurement underneath

The advisor is only as good as the numbers it reads. These are those numbers: computed once, from your ledger and your traffic, and shown to you as plainly as they're shown to the detectors.

Money

True profit

Revenue, minus the fees Etsy actually charged (read from the payment ledger, not a fee calculator), minus your cost of goods. Where revenue comes from, where the fees go, and which listings carry the shop.

Per-listing profit uses the fees Etsy attributed to that listing. Shop-level fees are shown as their own line, never smeared across listings by revenue share.
Cost of goods is yours to enter, or import a supplier invoice and the arithmetic is done for you.

Money: net profit per day, true margin and Etsy fee rate, then revenue and orders by week.

Funnel

Traffic problem, or conversion problem?

Google Analytics traffic joined to Etsy orders, per listing. A listing nobody visits and a listing people visit and don't buy look identical in revenue, and need opposite fixes. The funnel tells them apart, and says which listings are which.

Traffic and engagement from GA; orders and money always from the Etsy ledger. GA never carries a dollar figure.
A conversion rate on few visits shows its range rather than a false-precise point.

Optional. Without Google Analytics, everything derived from orders, listings and fees still runs. The funnel is the part that needs traffic.

Funnel: shop-wide visitors, engaged and buyers with drop-off, and which traffic channels convert.

Sample shop with Google Analytics connected.

Customers & LTV

Who comes back

Repeat rate, lifetime value, and how each quarter's new buyers behave over time. Etsy exposes a buyer ID and nothing else, no name and no email, which is all this needs.

New versus returning, over time, and the share of revenue that's repeat.
Cohort LTV by the quarter a buyer first purchased.

Customers and LTV: customer count, repeat rate, average lifetime value, repeat revenue share; new vs returning over time; LTV by cohort.

Explore

The whole history, scrubbable

From two weeks to twelve months, with the previous period and the same period last year overlaid. Revenue by listing, category and country; visits by channel. Your own edits are marked on the timeline so you can see what happened after each one.

One download, then every period change is instant, with no waiting on a server per click.
A window the data can't fully cover is labelled as such, never quietly clamped.

Explore: the period switches from 30 days to 90 to the year; the sessions line, its overlays and the change markers redraw instantly each time.

30 days → 90 → this year, no requests in between. Sample shop with Google Analytics connected.

Trends

Momentum, and an honest thirty days

Which listings are rising, which are falling, and whether that's speeding up, on a chart sized by revenue, so the ones that matter are the ones you see.

Day-of-week shape and week-over-week movement per listing and category.
"Next 30 days" is a range, and it is an extrapolation of your own recent rate and weekly shape, not a prediction. It knows nothing about seasonality, launches, or anything you change. It says so in the product, every time.

Trends: a 'Next 30 days' range labelled as an extrapolation of the shop's own rate, then a momentum-versus-acceleration bubble chart.

See it on your shop.

We're onboarding a small group of sellers now. Join the list, or ask for a free teardown.

Join the waitlist Get a free shop teardown