Data you decide with
A mid-sized company can run the way the big ones do. What a large company has is not better people: it is that the number arrives on its own, says where it came from, and goes unchallenged in the meeting. Your company already holds the data, in its ERP, its ledger and its stock records. And Monday's decision still gets made with whatever somebody managed to assemble on Friday.
Here you pick one concrete decision, one of the ones that repeats, and you get the five to eight figures that hold it up, read straight from your own system. Each one with its definition written down: which table it comes from, what it filters, what it leaves out and what date it cuts at.
At the handover session every figure is reconciled in front of you against your ERP's own report, or against the count your people already do by hand. If one does not match and the difference cannot be explained in writing, I fix it at no cost.
One-decision dashboard: USD 950, two weeks, about 6 hours of your people's time. Management dashboard: USD 2,400, four to five weeks.
It is not that data is missing. There is too much of it, and every system tells a different version of the same week. The cost shows up in no account, because it is paid in decisions postponed.
Somebody rebuilds the same file by hand every week, pasting together what they exported from three screens. By the time it reaches the meeting, the week it describes is over, and the person who built it is the one who knows the business best.
Sales does not agree with accounting. The meeting is spent deciding which of the two is right, and the decision slips another week. Usually both are right: they measure different things, with different cut-offs, and nobody ever wrote down which measures what.
The indicator gives the outcome and not the cause. Without the breakdown underneath it, the meeting ends in opinions and the correction gets made on instinct.
Delivered a year ago, good-looking, and showing the same number for two weeks. Nobody knows whether the business stopped or the load is broken, which is exactly why nobody dares decide with it.
Launch offer: the first three to commission the dashboard get three months of the monthly watch at no cost. It is not a discount: the price does not move, and once the three slots are gone it stays where it is. Before you talk to me, do this arithmetic with your own numbers. Think of the decision that gets made today without the figure it needs: how much money is on the table each time it is made, and how many times a year it is made. If the result, worked out conservatively, does not comfortably clear the price of the dashboard, I will say so and offer you the smaller step or none. Those numbers are yours and they are published nowhere.
Two steps. The first one settles a single decision and tells you whether this changes anything for you; the second takes the same method to the management table.
You pick a decision that is made blind today and that repeats: which customer gets credit released, which product gets restocked this week, which job is eating the margin. You get the five to eight figures behind it, read straight from the ERP.
One sheet per figure: which table it comes from, what it filters, what it leaves out and what date it cuts at. It is the document that stops someone calculating the same thing differently next month and producing the second number.
An hour and a half with whoever decides, not with whoever administers the system. We read the dashboard together, reconcile figure by figure against your ERP, and write down what happens when each one moves out of range.
The same for three or four decisions, which is as much as a management table can hold before people quietly stop looking at it.
One definition per measure, not one query per chart. That is the difference between a dashboard you can extend and a pile of reports that contradict each other as soon as somebody adds the fifth.
Automatic refresh at a stated hour, who sees what according to their role, and the definitions signed off by the people who use each figure to decide.
It runs on what you already have: your database and your ERP, with no new licence to buy. If your company already pays for a dashboard tool, the data model and the definitions serve you just the same, because they are yours and do not depend on the tool; and if you also want the final screen to live inside that tool, say so on the first call and I will tell you plainly whether I can keep it standing or not.
A dashboard with thirty figures changes no decision. It gets looked at in week one, admired in week two, and by week three nobody opens it, because watching thirty things at once is watching none of them. That is why the entry step covers a single decision: the one that repeats and that is made on gut feel today.
The underlying rule comes from two places. From Lean Analytics, by Alistair Croll and Benjamin Yoskovitz: one figure leads at a time, and the rest explain it. And from Thinking in Systems, by Donella Meadows, two warnings that get ignored a lot: an indicator that arrives late corrects nothing, because by the time it speaks the damage is done; and improving the indicator is not improving the system, because people learn to move the number without moving the business. Between them, those two decide what goes on the dashboard and what stays off it.
And something a sales page rarely admits: what is sold here is not chart design. The charts are deliberately plain, because the work is in the figure matching its source and in the decision being takeable while looking at it. If what you want is a striking presentation for a board, there are people who do that better than I do, and I will tell you so on the first call.
Saying it before quoting avoids the awkward conversation in week four, once money has changed hands.
If the figures do not come out of the source clean, a dashboard does not fix that: it hides it. That is a different job with its own page, data that arrives clean. It shows up in the first few days and gets said before quoting, not after.
Said in advance
It is the most requested thing and the least useful. A dashboard that shows everything points at nothing, and you end up paying for a catalogue of figures nobody decides with.
Hence one decision at step 1
A model that predicts has to be kept standing as the data moves, and without that upkeep it turns into a liability that decides badly with a scientific face. If it turns out to be needed, it is quoted separately and with its caveats stated.
Not in this price
What you get runs on your database and on what is already in production. If you decide to buy a tool, your company contracts and pays for it, and I tell you before it turns up on an invoice.
Never in my name
What lives in a notebook, in the warehouse supervisor's own sheet or in somebody's head cannot be read. It has to be captured first, and that is an earlier and different job.
Capture comes first
An importer and distributor running Odoo with eight years of history. No data comes from a real client: the company and its figures are made up, and it is the same one that appears elsewhere on this site.
The decision chosen: which customer gets credit released this month. It is made several times a week, it is made by the sales manager with the receivables in his head, and when he gets it wrong the company finds out ninety days later.
The six figures on the dashboard: average days to pay over the last twelve months, overdue balance by age bracket, the term the contract in force sets, approved limit against current balance, the customer's margin for the year, and orders held waiting for approval.
How one row reads. Distribuidora del Norte: the framework contract sets sixty days, they pay on average at fifty-two, they have two overdue invoices for USD 4,180 with the oldest at thirty-eight days, and their margin for the year sits above the average of the book. The decision takes a minute, and whoever takes it knows exactly what they are looking at.
One entry from the definitions. Average days to pay: payment date minus issue date, over invoices already paid in the last twelve months, excluding credit notes and cancelled invoices, cut at the last day of the closed month. It comes from the invoice and payment tables in your Odoo, with no manual step in between.
And the reconciliation, which is what makes the dashboard usable. At the handover session that figure is compared in front of you against the receivables ageing report your ERP already ships with. If they differ, the difference is explained in writing or it is corrected. That moment is the product: from then on nobody asks again where the number came from.
The real fear is not regretting it before you start. It is paying for a good-looking dashboard whose numbers nobody dares use. The guarantee is written against that fear, and it can be claimed because it can be checked.
If you cancel within fourteen days of payment and before receiving the first dashboard, I return the deposit in full.
At the handover session every figure is reconciled against your ERP's own report, or against the count your people already do by hand. If one does not match and the difference cannot be explained in writing, I fix it at no cost.
If within the following thirty days a figure turns up that does not match its source, I fix it at no cost. If I cannot make it match, I refund that step in full.
For this to apply: the definition of each figure is confirmed in writing by whoever uses it to decide, within three business days of receiving it. It is the only way "it does not match" means the same thing to both of us.
A dashboard is delivered in a twenty-minute demo and falls over in month three, when somebody compares one figure with the ledger. Six questions separate serious work from a good-looking screen.
Ask for the sheet: which table it comes from, what it filters, what it leaves out and what date it cuts at. If the definition lives inside the chart rather than in a document, the day someone challenges it there will be nothing to defend it with, and the dashboard stops being used.
The right answer is a report your company already has and already trusts. If the proposal is to check the dashboard against the dashboard, there is no check.
A dashboard built on a live system breaks quietly: a new ledger journal, a changed field, a module installed, and the figure keeps showing, only now it is wrong. Ask who watches for that, how often, and how you get told.
It is the technical question that saves the most money. If every chart carries its own query, dashboard number five will contradict number two, and nobody will know which one to fix.
Nobody can define a figure without the person who uses it to decide. Ask for the number in hours per person. If you are told almost none, the definitions will be invented, and you will find that out in the first board meeting.
A serious supplier will ask who keeps that model standing once the data shifts, because an unmaintained forecast decides badly with an air of precision. If it is offered up front without that conversation, you are being sold the expensive part before the useful one.
Two steps and a monthly watch. The first step is bought on its own, with no commitment to the second, and it is the one that tells you whether this changes anything.
One decision chosen with you, the five to eight figures behind it read from the ERP, the written definition of each one, and the reading and reconciliation session with whoever decides.
Three or four decisions with the same method, the data model behind them with one definition per measure, automatic refresh, access by role and the definitions signed off.
Every month each figure in scope is reconciled against its source again, what changed is reported in writing with the reason, and one indicator is added or retired. It also covers the loads in data engineering: one subscription for both. Paid a year ahead, USD 1,900.
About six hours spread over two weeks: an hour and a half of whoever decides to choose the decision, two hours of whoever builds the spreadsheet today to hand over the current definitions, an hour of IT or your ERP provider for read-only access, and an hour and a half of reading and reconciliation.
About twelve hours spread over five weeks, with the same people. It is written here rather than in the small print because a dashboard built without the hours of whoever decides comes out wrong, and I would rather you knew before paying.
Prices in United States dollars; under my tax regime in Ecuador, RIMPE Negocio Popular, my services are invoiced with VAT at 0 %, so the figure is the final one. Half on starting and half at the handover session. Guarantee on figures that match: full refund of the deposit within fourteen days and before the first dashboard; every figure reconciled in front of you at handover; and if within thirty days a figure does not match its source, it is fixed at no cost or that step is refunded in full. I take one engagement at a time: on the call I tell you the date of the next opening rather than promise one I cannot keep. Full conditions in the proposal and in the refund policy.
No. What is needed is that the figures exist today in a system that can be queried, and that you give me read-only access. With Odoo I know the ground and the work moves faster; with another ERP it is read through its API or through read-only views.
It shows in the first few days and I tell you straight away. Sometimes it is settled inside the same job, with a definition that excludes what does not belong. When the source itself is broken, the dashboard is not the first move: the data x-ray is, and I say so before quoting.
Because it lets you check, with little of your money, whether this changes how your company decides. If it does not, you find out in two weeks for USD 950, rather than in five months with a project on your back.
Partly, yes, and when the native report is enough I say so and sell you nothing. What the native report does not do is cross accounting with sales and with stock into a single figure, or leave written down what each one excludes, which is exactly what stops two departments arguing over the number.
You can, and sometimes it makes sense, when there is already a set of definitions everyone accepts. If there is not, starting at the top usually ends in four badly defined decisions instead of one well defined.
Your people. The data model, the definitions and the refresh are documented and sit inside your infrastructure, with no piece tied to me. The monthly watch is a convenience, not a key to the door.
Yes. The work is remote, the sessions are by video call and the hours are arranged around your time zone; I am based in Quito, which is the same time as New York for half the year and an hour behind it for the other half. Prices are the same ones on this page.
10 /
Tell me which decision gets made today without the figure it needs, and where that figure would come from if it existed. I reply with a proposal within one business day.
contacto@andresbrinez.comIf you would rather take a free first step, the maturity diagnostic tells you in fifteen minutes what stage your company is at.
Six years building systems for companies in three countries: Colombia, Ecuador and Australia · Member of the Odoo Community Association