A live dashboard from QuickBooks Desktop, without a connector subscription
Every answer on page one is a connector to buy, plus a dashboard you then build and maintain. Here is the other route. The numbers leave QuickBooks once a week into a database that is checked against QuickBooks before it publishes. The dashboard is a standing report in your inbox, plus any question you care to ask.
What we do about it
We take your data out of QuickBooks as reports, put it in a database, and connect it to your Claude or ChatGPT. QuickBooks stays as it is.
The first step is small on purpose. We model one year of your sales data and get on a live call where you ask and your own numbers answer. You give us the three questions you most want answered, and we get you those answers or figure out how we can deliver them. You keep the database, connected to your Claude or ChatGPT.
The first step is $950, credited in full toward your setup if you go ahead. If you don’t, we delete the project and your data doesn’t stay with us. If it earns its place, we keep it current every week and send the reports you pick.
Nothing we build can write to your books.
What a dashboard is for
An owner opens one to answer three or four questions. How are we doing against last year. Who owes us. What is slipping. What sold. A dashboard answers those and no others. The day the question changes, someone rebuilds a tile.
On QuickBooks Desktop the hard part is the plumbing: a driver on the QuickBooks machine, a subscription, the file open when the refresh runs, and last month’s numbers on screen when any of that breaks.
The route without the connector
The standing reports are defined once, from the questions you actually ask, and arrive by email every week. That is the dashboard most owners actually use. The same database is connected to your Claude or ChatGPT, so the question the tile does not have is a sentence away. Every refresh is checked against QuickBooks’ own totals before it goes out.
If you already use Power BI, the database is plain Postgres and Power BI reads it with nothing to buy.
Two views from a fictional company
Last week against the same week last year, and who is paying slowly. Both answered from the database as it stood at the last refresh.
How do the last four Julys compare, and what did August do after each one?
| Year | July | The August after | August against July |
|---|---|---|---|
| 2023 | $1,122,837 | $1,006,837 | −10% |
| 2024 | $1,303,342 | $1,360,858 | +4% |
| 2025 | $1,150,705 | $998,326 | −13% |
| 2026 | $1,161,860 * | — | — |
* 2026 is through July 24, the last day in the data, and is already past 2025’s whole July of $1,150,705 with a week still to go. The three finished years show no August pattern; this is not a forecast. It is four years side by side, in one sentence.
How that was computed
select substr(date, 1, 4) as yr,
cast(sum(case when substr(date, 6, 2) = '07' then amount end) as int) as july,
cast(sum(case when substr(date, 6, 2) = '07'
and cast(substr(date, 9, 2) as int) <= 24 then amount end) as int) as july_thru_24,
cast(sum(case when substr(date, 6, 2) = '08' then amount end) as int) as august,
round(100.0 * sum(case when substr(date, 6, 2) = '08' then amount end)
/ sum(case when substr(date, 6, 2) = '07' then amount end) - 100) as aug_vs_july_pct
from v_sales_lines
where substr(date, 6, 2) in ('07', '08')
group by yr
order by yrWho is paying slower than their terms, and how much of our money is that?
The eight biggest customers by what they were invoiced in the last twelve months, with how long each one actually takes to pay:
| Customer | Invoices | Average days to pay | Invoiced, 12 months | Terms |
|---|---|---|---|---|
| Norlund Homes | 94 | 41 | $1,224,017 | Net 30 |
| Kestrel Ridge Homes | 71 | 35 | $863,472 | Net 30 |
| Bergstrom Construction Group | 48 | 38 | $606,420 | Net 30 |
| Stonebeck Homes | 48 | 45 | $539,600 | Net 30 |
| Halvard Custom Homes | 42 | 31 | $425,705 | Net 30 |
| Pinewick Home Center | 38 | 35 | $255,049 | Net 30 |
| Alder Falls Hardware & Lumber | 39 | 35 | $236,863 | 2% 10 Net 30 |
| Norhaven Lumber Co | 38 | 33 | $226,120 | Net 30 |
Norlund Homes, the biggest customer on the books, averages 41 days on Net 30. QuickBooks has the invoice date and the payment date and does not keep the days between them. The database works it out for every invoice.
How that was computed
select e.canonical_name as customer, count(*) as invoices,
round(avg(s.days_to_pay)) as avg_days_to_pay,
round(sum(t.total)) as invoiced_last_12_months,
e.terms
from invoice_settlement s
join txns t on t.id = s.txn_id
join entities e on e.id = t.entity_id
where t.date > '2025-07-24'
group by e.canonical_name, e.terms
order by invoiced_last_12_months desc
limit 8Northgale Building Products is a fictional company. Every name, figure and supplier on this page is made up, built to behave like a real three-and-a-half-year QuickBooks Desktop file so we can show real screens without showing anyone’s real numbers. Data as of the 2026-07-27 refresh; the last invoice in the file is dated 2026-07-24.
Short answers
- Can I connect QuickBooks Desktop to Power BI without a connector?Not directly. The built-in routes need a driver on the QuickBooks machine. A database built from exports is something Power BI reads with no connector.
- How current is the data?As of the last weekly refresh. A Monday report that ties to the books beats a live tile that might not.