Henderson Farm Storemoved off a registerthat could not tell themwhat anything cost.

They opened in January 2024 on a card reader that carried no inventory and no cost data. They now run the store, the farm and a 45-vendor consignment program on one system, and they can price a cut of beef from what it actually cost them.

Monrovia, IndianaFarm store and farmOn Odoo since June 2025Engagement ongoing

What the old register could not do.

Henderson Farm Store opened in January 2024 and ran its first eighteen months on Clover. It took money reliably. Past that it was blank.

No cost data of any kind.: Nothing in the system knew what a product had cost, so nothing in the system could say what the store made on it.
No inventory.: Stock levels lived in whoever had walked the shelves most recently.
Around 99% of transactions anonymous.: Eighteen months of sales, and almost no way to know who had bought anything twice.
One company.: The store and the farm sold to each other and neither set of books saw it.
Accounting outside the register.: Every question that needed sales and costs in the same sentence was a manual reconciliation.

We never measured what the old system cost them, in money or in hours, so this page does not claim a saving on either. What it can show is the capability difference, and what the register did after the switch.

The switch, and the numbers that survive it

We stood the new system up in December 2024 and spent six months getting it ready. In the first week of June 2025 the store rang its last week on the old one and its first on the new one. June came in at $21,655 across both.

+27%

Trailing twelve months, before against after. $266.5K on the old register from June 2024 to May 2025, $339.5K on Odoo from July 2025 to June 2026.

Open days a month, up from 13. The store went from Tuesday, Friday and Saturday to Tuesday through Saturday.

634

Loyalty members who have actually earned points, against a register where about 99% of sales were anonymous.

9,221

Sales rung through the new system and paid, as of 26 August 2026.

Most of that growth came from two deliberate decisions, not from organic lift: the store opened twice as many days and moved to a more premium mix. On the days it was already open, like for like, revenue per day moved 2.6%.

The two systems do not count the same way. The old one counted each item; the new one counts each sale. So the only fair comparisons across the switch are what the store took per open day and per item sold, and those are the only two we have used.

Henderson Farm Store, ThriveWell client engagement. Figures read from their production system on 26 August 2026 and dated as of that day.

Seven weeks, and the parts no ERP ships with.

Between 11 February and 4 April 2026 the system got the pieces that are specific to this business. Odoo does not ship any of them, and none of them is a setting.

A consignment portal the vendors run themselves: 45 local vendors carry stock in the store. They sign in, see what of theirs is on the shelf, request a restock and submit new products for review. A restock request builds the draft purchase order on its own.
A catalogue page that maintains itself: The public list of what is in the store rebuilds itself off the shelf catalogue every four hours. 1,283 items, grouped and current, with nobody editing a page.
Profit and loss per brand: Every brand in the store gets a weekly snapshot pulling sales, purchase cost, scrap and shrinkage together. It exists because answering "is this vendor profitable" used to mean four manual pulls and an afternoon.
Milk sold from the farm to the store: The farm sells raw milk to the store daily. A week of those transfers becomes one invoice on Saturday, and both sets of books see it instead of neither.
Shelf labels that fit the sheet they buy: A label we built for them, six to a page, on the exact sheet stock the store already buys, in their colours. It took seven releases in two days, three of them just to get their logo to render.
A nightly close that cannot be forgotten: A register left open across two calendar days produced a $286.56 cash discrepancy on 3 April 2026. The job that closes it out at 11pm every night was created the next day.

Two things we built that are not Odoo at all.

Twice the right answer was that the main system was the wrong place to put it. Both times we said so and built the other thing.

A breeding tracker for five species, built for one gloved hand

Kayla identifies her animals by tattoo and ear tag, read by eye, standing in a field. We looked at five herd apps before writing anything, and none of them fit: the simple ones only handle cattle, and the ones that handle five species are full farm accounting packages. So we built her a tracker that works with no phone signal, installs from a link with nothing to buy, and takes no typing at all on the path she uses every day.

The plan, the build, her review of it, the fixes that came out of her review, and a 13-page printed guide all landed on 19 June 2026. One day.

It is live and in her hands. Her sign-off is still open, and every breeding interval in it is marked as waiting on a vet to confirm, which is where they stay until one does.

An assistant the owners can text, so they never have to open the system

Adam and Kayla do not use the main system day to day and have no wish to learn it. They message an assistant on their phones instead and ask in plain English. It reads the business, files a photo of a receipt onto the right record, and hands anything it should not be answering back to us. It can only do a short list of things, and anything that would change or delete a record stops and asks first. The Henderson team named him.

We tested it against a fixed set of questions, three runs each. With the routine we wrote for its weekly reporting it got every one right. Without it, three in four.

That test is the only controlled measurement in this engagement. Everything else on this page was observed. That one was checked.

What is still open.

A case study that stops at the wins is an advertisement. This engagement is ongoing, and these are the open items as of 26 August 2026.

A costing defect we found, and have not yet fixed.Every sale was recording its cost twice. We reproduced it on a copy of the system, proved the fix nine different ways, and then found that the obvious correction had already been tried once in January and had quietly undone itself within six months. It is still not applied, because we will not make a change to a live set of books on a fix we cannot demonstrate first. Until it is, the store reads its margin off the sales lines rather than the accounts, and we have told them so in writing.
Two systems the owners asked for, and have not needed yet.They asked for a way to settle livestock sales and a way to run vendor events, and we built both to what they described. One holds a single record. The other has not run a live event. Both work, and both are waiting for the season that needs them. It is here because delivered and used are not the same thing, and only one of them is worth counting.

A reader deciding whether to hire us is owed the same view of this engagement that we have. The defect is ours to fix. The two systems are nobody's failure, and we would still rather you knew.

If your register cannot tell you what anything costs, start there.

The free operations review is a conversation about what you actually have and what it would take to put it on one system. If it is not a fit, we will say so.

Get Your Free Operations Review

One operator, two to three engagements at a time. If the timing is wrong, we will say so.