Build1 distinct publisher3 min readUpdated
Nine broken IIF files produced nine identical success dialogs on QuickBooks Mac. Two of those runs posted nothing, and one still created accounts of the wrong type.
The Engineer · Build desk
Compiled by The EngineerSomething wrong?How this is made
The dialog is the entire result channel, and it does not vary with the outcome. Nine deliberately broken files produced it nine times [1], and two of those runs posted no transactions at all [2], so the message cannot separate a write from a no-op [16]. There is nothing narrower underneath it: no per-record status, no count of rows accepted. That is the author's generalisation, and it holds beyond this product. When a format carries no per-record result channel, the one remaining place to report fires whether or not anything worked, and a partial write underneath it is invisible [13].
The single run that did report a problem is worth reading closely. The unbalanced file raised a warning telling the user to check that the amounts in the detail area on the form match the amount at the top, and there is no form on screen: one OK button, and it advances to the success message [6]. Failure notice and success notice sit on the same click, and the click destroys the only copy of the first one [6]. QuickBooks for Windows 2019 and later is widely described as writing a line by line import report with numbered error codes that people quote on Intuit's forums [7]. None of these imports on QuickBooks Mac Plus 2024 produced one, and the application's own log held internal identifiers and nothing else [8].
The company file started with exactly one account, and every other account the files named was missing on purpose. QuickBooks created all of them without asking and typed every one as a bank account [9]. Not from the name, not even for names QuickBooks itself uses, and not from how the file used the account [10]. The receivable file is where that gets expensive: it wrote no transactions and left two new accounts behind [12]. The source's own arithmetic does not fully close, since the opening says one failing run left three records while the later attribution puts two of the three with the receivable file [3][12], so the figure to hold is the load-bearing one, two accounts created by a run that posted nothing [17].
A wrong account name gets fixed, because the account turns up somewhere visible and somebody goes and corrects it. A wrong type stays put. Expenses sit in something that looks like a bank account, the profit and loss report is missing them, and every total involved is wrong until someone notices [11].
The converter had left IIF's !ACCNT type section out on purpose, reasoning that creating accounts in somebody's books is worse than a bad import you can restore from backup [15]. The test removes the premise, because the accounts get created either way and the only open variable is who picks the type. The same measurement caught a reassurance on the converter's own pages, that Uncategorized Expense and Uncategorized Income ship with QuickBooks, in a blank company file that had neither [14]. A user told only that an import failed will assume nothing happened [18], and here that assumption is wrong by two accounts [17].
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
One of the two files that wrote no transactions still left three new records behind in the database.
Nine test IIF files all produced the same QuickBooks dialog: "Your data has been imported."
Two of the three orphan accounts were created by the receivable file, which is one of the two files that wrote no transactions; that run rejected the transaction, kept the side effects, and ended on the message saying the data had been imported.
Two of the nine test files wrote no transactions at all.
The author's general claim: if a format has no per record result channel, there is exactly one place left to tell the user what happened, it fires whether or not anything worked, and a partial write underneath it is invisible.
The author maintains a browser based converter that turns bank CSV files into accounting formats, one of which is IIF, a tab separated format QuickBooks Desktop has used since the 1990s and which has a reputation for failing quietly.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
First-hand reproducible test, one build, one tester
The core findings are backed by a deliberately designed nine-file matrix with counted outcomes and verbatim dialog text, which is stronger than anecdote. It is weakened by a single tester, a single product build (QuickBooks Mac Plus 2024), a throwaway company file rather than a realistic chart of accounts, no published artifacts or screenshots in the supplied text, and no independent reproduction. One relayed element, the Windows import report with numbered codes, is untested second-hand description.
One shipped tool change, no usage data
There is concrete deployment evidence at exactly one point: the author's converter now emits !ACCNT account types in every file, and two documentation pages were corrected. Beyond that the supplied source gives no user counts, download figures, third-party adoption of the practice, or any indication that other IIF producers or Intuit changed anything, so ecosystem uptake cannot be scored higher.
Modestly framed, one generalization outruns the test
The post is unusually restrained: it scopes findings to the tested build, publishes its own documentation error, and states what it could not verify. The small positive gap comes from generalizing a single Mac importer's behavior into a rule about any format lacking a per-record result channel, and from contrasting against Windows reporting behavior that was never exercised.
Maintainer of the linked commercial converter, disclosed
The author writes about a product they maintain and closes by pointing readers to qbofile.com, so there is a promotional incentive in publishing a rigorous-looking measurement post. That is offset by the disclosure being explicit and by the piece admitting an unmeasured claim previously shipped on its own pages, which cuts against pure marketing. No vendor sponsorship, funding, or third-party payment is indicated in the supplied source.
Credible single source, no corroboration
The mechanism is specific and internally consistent, and the counts are stated precisely, which supports moderate confidence in the described behavior on the tested build. But the cluster contains exactly one publisher and one author with a commercial stake, no vendor comment, no independent replication, and no evidence about other QuickBooks versions, so confidence in any broader statement stays near the middle.
product
Rillet's $100M reads as proof mid-market ERP is rip-and-replace, mostly at the cheap end1 distinct publisher
leadership
ClickHouse buys Langfuse, turning a neutral tracing layer into someone's roadmap1 distinct publisher
build
Rillet raised $100M in 48 hours. The audit cycle is the unpriced part.2 distinct publishers
build
A parser you can read beats one you trust: hand-rolling .xlsx to keep a CSP promise1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · August 23, 2026