Saving Face was losing money. Nobody could find where.

Fragmented systems, and blindspots nobody could see

Saving Face was not a practice sitting on its hands. Three Austin locations. A real EMR. They'd brought in a new bookkeeper and a fractional CFO. They'd already gone through the exercise of asking how do we get better at this, and they'd changed real things.

And the numbers still said money was leaking.

Here's the part I think gets skipped in every case study I've ever read: they weren't blind. They had blindspots they couldn't see. Those are different. Blind means you know you can't see. A blindspot is a thing you're certain you have handled, and you don't, and it costs you money anyway.

The systems were real. They just didn't connect.

This is the honest state of most practices, including good ones. You build a system for one problem. Then another for the next problem. A spreadsheet here. A shared doc there. A process that lives in one person's head because she's been here six years and she just knows.

None of it talks. And everything that would have actually told them what was happening sat still in a spreadsheet while the business kept moving.

The EMR was never going to do this

An EMR books appointments and charts patients. That's what it's built for and it's good at it. Inventory is a tab bolted onto the side.

It won't show you what's expiring. It won't show you what's actually out of stock across three locations right now. It won't tell you what a unit cost you, or whether the number it's showing you is even right. Saving Face had the EMR and still had to go count things by hand to know anything.

What they didn't know they were struggling with

• Where product was actually going. It left a room and didn't get logged. Nobody was stealing anything — there was just nothing to check a count against.
• What it truly cost. Vial price divided by units. That was the number every decision ran on, and it's wrong for four separate reasons.
• Product they'd been promised and never got. A rep commits samples as part of a deal. Nobody tracks the promise. Nobody knows to go ask. You don't chase what you don't remember you're owed.
• Whether what was ordered actually showed up. Invoice arrives, boxes arrive, somebody puts it away. Did all of it get put away? Did all of it get entered? Nobody could answer that.
• Which provider was using how much. Not to police anybody — to know. Two providers doing the same treatment can be twenty units apart and nobody would ever see it.
• Credits sitting there unclaimed. A memo lands and nobody mentions it. Not the rep, not the vendor.
• The entire back office. Licence expirations, prescriptive authority agreements, insurance renewals, lease terms, who's trained on what. All real obligations, none of them in anything that would raise its hand before a deadline.

Every one of those is normal. That's the whole point. Normal is why they sat there.

Stopped trying to fix the business. Picked one thing they could count.

This is the best decision in the story and it had nothing to do with software.

Not "fix operations." Not "get better at inventory." One category, in one room. Six toxin products — Botox, Daxxify, Dysport, Jeuveau, Letybo, Xeomin — on a ledger, counted at the end of every day, with a name attached to each count.

Toxin because it's the most expensive thing moving through the rooms and because it's measured in units, and units can be counted. If you can't get your arms around the one thing that's genuinely countable, you have no business trying to fix anything harder.

The reason they started there: they'd already hand-counted the room once and found the shelf and the records about 1,500 units apart. Not a rounding error — in a category where a unit is real money. That count is the reason Stcki exists at all.

And here's the finding that actually mattered — not the number. Some of that gap was tracking error. Some of it wasn't. Nobody could tell which. A vial used and not logged, a count rushed at close, units entered where vials were meant, product that walked. Four different problems, one identical symptom, and no data to separate them.

That's the trap. You start hunting a culprit when what you have is a missing system. Your team feels accused, you feel helpless, nothing resolves, and it happens again next month.

From six products in one room it went to every location, then every room, then everything with a cost of goods attached to it, then the parts of the practice that were never inventory at all.

Operational efficiency

This is the one owners feel every day and can never put on a line item. Nobody invoices "the team stopped running around." It showed up first, it compounds, and it's still what the staff actually talks about.

They know what's in every room without walking into it

Three locations, seventeen rooms, and anybody can answer what's on that shelf right now from wherever they're standing. That sounds small. It is not small. It's the difference between preparing the right amount once and preparing twice, and between a patient waiting while somebody drives product across town and that never happening.

The days actually get closed

8,278 confirmed room close-outs. Seventeen people — front desk, medical assistants, injectors — logged inventory in the last sixty days alone. Not a policy anybody enforces — a sixty-second count at the end of the day with a name on it. When something is off, it's today's problem, with today's people, in today's room. You fix it in five minutes instead of losing an hour on Friday reconstructing a week and blaming somebody.

The team went from resisting it to running it

Rollout was not smooth. Nobody asks for another thing to do at the end of a long day, and the people who'd been doing it their own way for years had the least reason to want it.

"We weren't sure this was going to work at first. Now I can't get through a day without it."

— Hannah Winkler · Saving Face

The part that says more than the quote: Stcki became the source of truth she reconciles the EMR against. Not the other way around.

That's not a compliment anybody set out to earn. It's arithmetic. End-of-day runs in Stcki every day, so Stcki is current and specific — this room, this lot, this person, today. When the EMR and the counts disagree, the counts are the ones that got taken at the shelf.

And it matters more than it sounds like it does. A system the staff resents gets half-used, and half-used data is worse than no data, because you'll trust it anyway.

Expiration stopped being a discovery

Product used to age out with an identical vial sitting next to it — not from over-buying, but because somebody reached for a vial that wasn't the oldest on the shelf. Oldest expiration goes first; everybody knows that. But when nothing tells you which one that is, it comes down to whoever's at the fridge at 4:40 on a Thursday.

Right now the system flags 11 lots expiring inside 120 days and 2 that already lapsed. That's not a clean scoreboard and I won't pretend it is. It's thirteen items that are visible instead of discovered.

The alerts go where the team already is

Nobody logs into another dashboard to find out something needs attention. Expiring lots, stock running short, a count that didn't happen — it lands in Slack, in the channel the team is already sitting in all day.

That's a small architectural decision that decides whether a system gets used. Information nobody sees is the same as information nobody has.

Provider-level visibility that didn't exist before

Across 3,420 treatment sessions and 12 providers, the practice can now see units per treatment by provider — who's consistently heavier, who's lighter, and whether a treatment is being under-dosed relative to how it was sold.

That's not about policing injectors. It's a clinical and financial conversation that was simply unavailable before, because the data didn't exist in any form anybody could look at.

An AI reads the practice every morning

This is the part that changed what Stcki even is.

Once the day-to-day is structured — what came in, what moved, what got used, by whom, in which room — you have something an AI can actually reason over. Not a chat window bolted onto a dashboard. The practice's own operating data, in a shape a model can read.

Saving Face has had a generated brief every single day for 101 days straight, covering what's off, what's about to expire, what's running short, what a count didn't explain. Nobody assembles it. It's there before anyone walks in.

And the team talks back to it. 497 conversations and 2,368 messages across nine people asking questions of their own data — 184 of those conversations in the last thirty days alone. That number is going up, not down, which is the only adoption metric I actually trust.

Underneath it, 10,981 operational signals have been raised and routed since it turned on.

Automations that removed the job instead of speeding it up

The ones that matter didn't make an existing task faster. They deleted it. Reorder points that fire before anyone is short. Expiry watch that surfaces a lot months out. Credential and insurance renewals that raise their hand ahead of a deadline. Invoice reconciliation that runs itself.

New ones ship continuously, and they all start the same way — somebody says this takes us too long every week, and a few days later nobody does it anymore.

Work that used to take days

And once the operational side was on rails, the same approach went at everything else that had been living in somebody's head: credentials and licences on expiry watch, insurance policies and renewals, leases, prescriptive authority agreements, equipment, and which person is explicitly cleared on which service.

Saving money on product and inventory

Counts stopped being wildly off

Toxin only, so it's apples to apples: in February an average count line was 18.3 units off. Now it's 3.8 — on nearly double the lines counted.

Variances did not go to zero and they won't. Counts still disagree with the system regularly, because that's a live practice with real people in it. Anybody claiming zero is either lying or not counting. What changed is the size, and size is what costs money.

They found out what product actually costs

Four things move the real number between the invoice and the treatment room: the discount printed on the invoice, the credit memo that lands a quarter later that nobody mentions, loyalty dollars sitting in a portal, and free product. Saving Face had been counting one of the four.

Across the 92 products purchased in the last twelve months, weighted by spend, true cost per unit came back 18.4% under invoice price. The practice paid for 4,495 vials and took in 5,164.

Here's the part that surprises people. That 18.4% did not come from negotiating better prices. On the same products year over year, what the practice pays per vial is basically flat — up about 2%. Every point of it came from counting what happens after the invoice. That money was already theirs.

Promised product stopped disappearing

This one was never a money problem. It was a memory problem.

A rep commits samples as part of a deal. Six weeks later nobody remembers the commitment, so nobody asks. 133 promises are now tracked. 107 were delivered in full. 20 came in short. 6 are still open.

Twenty short deliveries. Under the old way, not one of them gets caught — and guaranteed product the practice negotiated for simply evaporates.

Ordered, invoiced, shipped, received — and it all matches

This is the newest piece and it closes the loop. You ordered it. They invoiced it. They shipped it. Somebody received it into Stcki. Do all four agree?

For Saving Face right now, every purchased item on an invoice reconciles to something received. Zero unaccounted for. Before this existed, boxes arrived, somebody put them away, and whether all of it got entered was a matter of faith.

Knowing which product to reach for

With a real cost on every line the practice can rank its own products by what each one actually returns. Almost the entire filler shelf rings the same to the patient. What comes back varies by a lot depending on which syringe gets used.

The patient experiences nothing different. Nudge the mix a few points toward the better end of the same shelf and it's meaningful money every month — without selling one more syringe and without touching what anybody is charged. Going into the busiest season, Saving Face bought deep on purpose on the categories it could prove were carrying it, instead of guessing.

Doing more with less

Here's where it lands. The practice's CFO, working off closed books, has taken 4% off cost of goods in the last three months. Over that same stretch, revenue is ahead of last year.

More treatments going out the door, less cost of product behind them. That is the entire objective, and it's the one line on this page Stcki did not calculate — Stcki sees purchasing, not the P&L. What it can show you is the mechanism underneath it.

Revenue intelligence — the next build

Everything above is operations and product cost. Two of the three, and I'm not going to dress it up as three.

Revenue intelligence — what converts, which provider drives which outcome, which services carry a practice and which ones only look like they do — isn't built yet. Not at Saving Face, not in the product.

It's being built now, and the reason it goes last is the same reason Saving Face started with six toxin products in one room: you don't get to analyze what you can't yet measure.

Where it doesn't hold up

• Counting still isn't at 100%. Rooms close out a bit over half the days they should. The counts that come in are good. Not all of them come in.
• Variances didn't go to zero. Roughly four in ten toxin count lines still show some variance. What dropped is the size — 18.3 units to 3.8. If someone shows you a zero, ask how many lines they counted.
• 18.4% is a gap, not a saving. It's the distance between invoice price and true cost over twelve months. It is not a year-over-year cost reduction and this page doesn't claim one — invoice prices on the same products are up about 2%.
• The 4% is the CFO's number off the P&L, not Stcki's. Stcki sees purchasing, not revenue. Don't let anybody hand you those two as the same variable.
• Ten deals is ten deals. Strong start, not a year of data. One of the ten landed above their own cost and it's netted into the figure, not quietly dropped.
• Not everything can carry a unit price, on purpose. Some items are consumed inside a service sold whole — provider time, machine time, a consumable — often as a package of three or four or six at a price that moves with the package. There's no honest per-unit number to assign.
• Some automations are specific to this practice. Built around how Saving Face runs. Another practice gets the same problem solved a different way, and that takes time.