A recreation of a real engagement
What follows recreates what we walked into at a 47 store automotive group, and what changed thirty days later. The people on screen are actors and the client is not named. The sequence, the numbers, and the quotes are exactly what happened. If you run finance for a multi location business, you have met both of these mornings.
Day 0 / The incoming CFO
Over $1,000,000 sat unreconciled, and nobody could say how much was truly missing.
They had just taken the seat. Deposits were months deep across cash, cards, and a customer financing program, and every answer had to be requested from one person and waited on. Not one of those dollars was known to be lost. That was exactly the problem.
Day 0 / The controller, before anyone else arrives
“It’s just copy and paste. Yeah.”
Their own words, describing their morning. Thirty nine stores of daily cash drawer totals, matched by hand against five exports that never agree on a date, every single day. The rules that made them agree were written down nowhere except in one person's head.
Keep scrollingWatch a single day close itself
The answer, end to end
Cash, cards, and customer financing reach the bank on three different clocks. The exports arrive, the rules run, and the day resolves. Everything after this is what that actually takes.
01 / What the CFO inherits
Not at month end. Today, before lunch. Until every store day is tied to the deposit that proves it, your cash position is an estimate, the close drags, and nobody can separate what is genuinely missing from what is simply still in transit.
02 / What the controller does
Daily cash drawer totals, bank activity, card batches, the financing report, and invoice level tender. None of them agree on dates. Do the rules that make them agree exist anywhere except in one person's head?
03 / What it costs you
Hours every morning, a coverage plan whenever they take leave, and a thirty tab workbook stacked on top at month end. If they resigned tomorrow, would the rules walk out with them?
“It’s just copy and paste. Yeah.”
04 / What we built
That is what we built. Your team drags the same exports onto a page. The engine matches every store day against the bank on its real funding calendar, applies the weekend and cutoff rules for you, and never guesses. Nothing is ripped out, and nothing new has to be integrated.
05 / What you get
Because what arrives instead is one short list: the stores where money is genuinely missing, each row explained and, wherever the record allows, pinned to the sale behind it. Everything else is already closed and evidenced.
How it works
Six stops from the exports landing in your controller's inbox to the short list on your screen. This is the whole method, with nothing left out.
Stop one / The question
It sounds like a question with a quick answer. It is not. The money left those stores as cash, on cards, and through a consumer financing program, and each of those travels a different road to the account, on a different timetable.
Stop two / Five sets of books
Daily cash drawer totals. Bank activity. Card batches. The financing report. Invoice level tender. Five files, five formats, five different clocks, and not one of them is wrong. They simply do not line up on the same day.
Stop three / The rules nobody wrote down
One location banks on its own. The financing program pays a single lump sum for everyone, net of fees. Holidays move all of it. These rules live in somebody's head, and they are the reason most mismatches are not mismatches at all.
Stop four / The bridge is a person
It takes hours, it is never quite finished, and it quietly makes one person the only one who understands how any of it works. When that person leaves, the knowledge leaves with them.
Stop five / The engine
Every store, every day, matched against the bank on its real funding calendar. Exact math and stated rules, and never a guess. When a row cannot be explained honestly, it says so plainly instead of inventing an answer.
Stop six / The short list
Each unmatched row arrives with its reason attached and, wherever the record allows, pinned to the exact sale behind it. Items escalate on their own at day three and day five, so nothing needs chasing and nothing quietly ages.
Before the second cup of coffee the day is closed, the exceptions are named, and the person who used to carry all of it has their morning back.
Map your reconciliationBuilt alongside the accounting team that does the work. Their rules became the engine, session by session, screenshot by screenshot. More than forty releases shipped in two weeks.
Matched to the bank on a realistic window of banking days, with a grace amount, branch depositor handling, and each store's quirks encoded. Sundays and bank holidays are excluded from the count automatically.
Matched on their true funding calendar. Weekend bundles, nightly cutoff splits, late funders and next day netting are all handled automatically and explained in plain English on the row itself.
One company lump sum decomposed into per store expectations with exact fee math, then cross checked against the processor's own daily report, which predicts the bank credit to the penny one banking day out.
A shortage gets pinned to the single invoice that caused it, so the store makes one phone call. A sale rung on the wrong payment type gets flagged with the matching offset suggested. Two processor funding shortfalls, $607.58 and $862.19, were caught to the penny and documented.
Link, offset, write off, unpair, correct a source number. Every action records who did it and when, and every action can be undone. Month end write offs become a clean printable report instead of archaeology.
Items that sit unresolved move up on their own, to the team at day three and to leadership at day five. The company president sees exactly one thing: the stores where money is genuinely missing.
Recreation / Days 1 to 14
Nobody handed this team a product and left. The controller walked us through their actual morning, screen by screen, on a shared screen. Every rule they had been applying by hand became a rule in the software while they watched, which is why it fits how this business really moves money rather than how software usually assumes it does.
Days 1 to 14 / The working sessions
Their rules became the engine while they watched.
Live software on day one, then more than forty numbered releases in two weeks. The weekend bundling, the evening cutoff roll, and the financing fee math each came from the controller walking us through a real morning on a screenshare. When something did not match how the work is actually done, it changed that same day.
Every item below came from the same 47 store automotive group. Each one started as a question their accounting team asked us, and each answer is running in production today. Do any of them sound like your close?
Cash reaches the bank through an ATM that records a card number and no store name, so deposits could not be assigned without someone recognizing them. We mapped every ATM card to its store, so a deposit now lands against the right location on its own. Locations that deposit at a branch and hold no card are handled as their own case rather than quietly going missing.
Weekend card settlements arrive bundled into a single Monday deposit, so a Friday shortage used to surface as a Saturday problem, or not at all. The bundle is now split back per day using each day's own authorization totals, which means a Friday variance shows up on Friday. On live data this exposed a shortfall on exactly the day it happened.
Anything rung after the evening cutoff funds the following banking day, which reads as missing money until it is not. Using the timestamped authorization export, the engine proves transaction by transaction which sales rolled forward, and says so in plain English on the row instead of leaving the team to guess.
A single deposit covering every location, net of fees that change by promotion, with one location funding separately on its own account. It is now decomposed per store with exact fee math and cross checked against the processor's own daily funding report, which predicts the bank credit to the penny one banking day ahead. Two real shortfalls, $607.58 and $862.19, were caught and documented this way.
A clawback lands long after the day it belongs to, which is how a clean month quietly reopens. Every case is matched back to its original sale, carries its own sign so it nets against the correct day rather than doubling the error, and sits in a dispute queue with its reason, status, and original sale date attached.
A blank note is indistinguishable from a lookup that never ran, which is why the same question kept coming back. Now every variance carries a note either way, including the ones that cannot be explained: either the report covers that day and nothing matched, or the report does not reach that day yet and needs uploading. On live data that is 99 card variances, 99 noted, and zero blank.
Reconciliation fails in two different ways depending on where you sit. This was built with both seats at the table, working session by session with the people who actually close the day.
You get asked for a cash position, a close date, and a figure for what is genuinely missing. Right now all three rest on one person's spreadsheet and one person's memory.
You already know the rules. You apply them by hand across five exports before anyone else is awake, and the backlog grows on any day you cannot get to it.
Recreation / Day 30
Same company, same two roles, thirty days later. The day is already closed before anyone arrives, and the conversation about what is left takes minutes instead of the whole morning.
Day 30 / The same morning, rebuilt
“So we can just breathe.”
The controller again, on what changed for the team. Unreconciled deposits fell from over $1,000,000 to $435,000 in a single day of the team using it. The close stopped waiting on one person being available, and what is left each morning is a short list that arrives already explained.
Send us how the money moves in your business and we will tell you exactly where the blind spots are. Nothing to install, nothing to rip out, and no obligation.
Want the detail first? Read the full case study, or the version written for CFOs and for controllers.
The same engagement, written four more ways. Whoever has to be convinced next, there is a page that speaks to their part of the problem rather than yours.
The entire engagement start to finish: what was found in the first working session, what shipped in the first day, and what the numbers looked like thirty days later. Read this if you want the evidence before the pitch.
Written for CFOs and owners of multi location businesses. Blind spots between systems, key person risk, month end archaeology, and the growth tax of adding another drawer to count by hand with every new location.
Written for controllers and accounting managers. What it means when the process only works because you personally make it work, and what changes when exceptions arrive already explained and your judgment stays in charge.
Written for the person who actually does the reconciliation each morning, and whose work stays invisible until the day they take off. Includes a one page writeup they can forward up the chain without having to make the case themselves.
No. The engine sits on top of the systems you already own, including a point of sale from another era and an accounting server that predates the cloud. Nothing is ripped out and nothing is installed on your machines. In this build the entire thing ran behind zero trust access for a company with no in house IT team.
The matching is deterministic. It is exact math against written rules that your team defines, so the numbers are never a model's guess. AI is why the tool could be built and refined this quickly, and it powers the assistant that answers plain English questions about your data. Every judgment call stays with a person.
It says so. Unexplained items are never quietly matched or absorbed. They go into a review list with whatever evidence exists, including the nearest candidate deposits and, where the data supports it, the exact invoice that accounts for the difference. A person decides, and the decision is recorded with their name and the time.
In this engagement the kickoff working session was on a Tuesday and the team logged into a live secured dashboard the next day. Working software comes first and the rules get encoded with your team from there. More than forty numbered releases shipped in the following two weeks, most of them the same day they were requested.
The math is the same at two locations as it is at forty seven. The question is whether someone on your team is spending a meaningful part of every morning answering it by hand, and whether the rules for doing so live only in that person's head. If both are true, the size of the company is not really the issue.