QuickBooks data warehouse cost and the metered API reads behind it
8 min read Finance The Adapters team
Field mapping auto-plugged · tap a port to rewire
Plug a source port into
Transform on this cable
JSON in
JSON out
5 sample records ready
A QuickBooks data warehouse costs three things: the connector, the warehouse, and since November 2025 the QuickBooks API reads themselves, because Intuit now meters every call that takes data out. The pattern you sync with decides the second number far more than your data volume does. An hourly full reload of ten entities spends about 36,000 metered reads a month per company. Hourly change data capture spends about 720.
Key takeaways
- Reads are metered. Intuit's CorePlus class, which covers reads, queries and reports, has been metered since 3 November 2025.
- Pattern beats volume. Reloading unchanged data costs 50 times the reads of change data capture.
- The free tier blocks. Builder includes 500,000 CorePlus credits a month per workspace, then stops calls until the 1st.
- Warehouse cost is the small line. Accounting data is tiny next to event data, if the tables are partitioned.
How much does a QuickBooks data warehouse cost?
Less than most teams fear on storage and more than they expect on everything around it. Five lines make up the real monthly cost, and only one of them is the warehouse. The table shows the unit each line is billed in and what makes it grow. We print no third-party prices here, because vendor price pages change faster than this article will, but the units are what you model.
| Cost line | Billing unit | What makes it grow |
|---|---|---|
| Connector or pipeline | Flat subscription, or rows, events or volume moved | With invoice count and how often tables are reloaded |
| Intuit CorePlus reads | Credits per successful read call, included up to a tier allowance | With reload frequency, not with how much changed |
| BigQuery storage | Bytes stored per month | Slowly, accounting data is small |
| BigQuery queries | Bytes scanned on demand, or reserved slots | With unpartitioned tables and dashboards that refresh often |
| Staff time | Hours spent reconciling and fixing loads | With every silent failure that reaches month end |
For a typical small business the BigQuery side is a rounding error. Three years of invoices, bills and journal entries for a company doing a few thousand transactions a month is measured in megabytes. Queries get expensive only when a dashboard scans the full history every few minutes, which partitioning by document date prevents. The lines that move are the connector and the metered reads, and both are driven by the same choice: how the tool decides what to read.
Does Intuit charge for QuickBooks API calls?
For reads, yes. Intuit's App Partner Program splits QuickBooks Online calls into two classes. Core calls put data in, such as creating an invoice or updating a customer, and are unmetered. CorePlus calls take data out, such as reading accounts, running queries and fetching reports, and are metered. Intuit decides the class by endpoint and HTTP method, counts only successful 2xx calls in production, and began metering on 3 November 2025 for partners in the US, Canada outside Quebec, the UK and Australia. The definitions are in Intuit's API classification article.
A warehouse sync is a read job from end to end. Every page of invoices, every change data capture call and every report pull is CorePlus traffic. You do not pay Intuit directly as the QuickBooks customer. The app or connector you authorize does, from its own allowance, and it either absorbs that, passes it on, or runs out.
What happens when the free API allowance runs out?
On the free Builder tier, Intuit includes 500,000 CorePlus credits a month per developer workspace and blocks CorePlus calls above that until the allowance resets on the 1st. Paid tiers include more and bill the overage instead of blocking. The allowance is shared across all production apps in the workspace, so a small vendor that runs every customer through one app on Builder has one pool for everyone. When it empties, your sync stalls with nothing wrong on your side, and the symptom you see is a dashboard frozen on the 23rd.
A worked example with real numbers
Take one QuickBooks Online company and ten entities a finance model needs: invoices, payments, credit memos, bills, bill payments, journal entries, items, customers, vendors and accounts. Assume 5,000 records per entity on average, read at the 1,000 record maximum page size, so a full reload is 5 pages per entity and 50 read calls per run. A change data capture run covers several entities in one call and returns up to 1,000 changed objects, so for a business that changes fewer than 1,000 records between runs it is one read. Over a 30 day month:
| Sync pattern | Runs a month | CorePlus reads a month | Verdict |
|---|---|---|---|
| Full reload every 15 minutes | 2,880 | 144,000 | Burns credits and the 10 connection budget for data that did not change |
| Full reload hourly | 720 | 36,000 | Common default in simple tools, and 50 times the reads it needs |
| Full reload daily | 30 | 1,500 | Cheap, but yesterday's numbers and still blind to deletes |
| Change data capture every 15 minutes | 2,880 | 2,880 | Intraday cash and AR views at a fraction of the reads |
| Change data capture hourly | 720 | 720 | Right for most finance dashboards and close tracking |
Now scale it the way a vendor does. A 500,000 credit Builder allowance covers about three companies on a 15 minute full reload, or about 690 companies on hourly change data capture. That is the difference between a tool that quietly stops in the last week of the month and one that never notices the meter. The one-time backfill is small by comparison: 50,000 historical records at 1,000 per page is 50 reads, or 500 if a script leaves the page size at the default 100. The backfill cannot use change data capture at all, because Intuit rejects a changedSince date older than its fixed retention window, so history always comes from paged queries.
Can QuickBooks be used as a data warehouse?
No. QuickBooks Online is a ledger with reports, and its reports answer accounting questions about one company. It cannot join revenue to product usage, ad spend or CRM stages, and the API reads that feed external reports are now metered. Teams move QuickBooks into BigQuery or Snowflake precisely to ask questions the ledger was never designed for, then keep QuickBooks as the system of record that the warehouse reconciles to every month.
Which warehouse is cheapest for QuickBooks data?
At accounting volumes, whichever one your company already pays for. BigQuery bills on bytes scanned, so partitioned transaction tables keep a month-end query to a few megabytes. Snowflake bills on warehouse compute time, so the cost follows how long the warehouse stays running. Postgres on a small managed instance is often enough for one company's books. The connector and read pattern matter more than the destination. The options for each are compared on QuickBooks to BigQuery integration tools, on QuickBooks to Snowflake integration tools and in BigQuery ETL tools.
Which BigQuery column types should QuickBooks amounts use?
NUMERIC for every amount. Google documents it as 38 digits of precision with 9 after the decimal point, so cents are exact. FLOAT64 is binary floating point, and a sum over a year of invoice lines drifts by cents that never match your profit and loss. Store document dates as DATE and partition by them. Keep the parent account reference on the accounts table, or the chart of accounts flattens and nothing rolls up. Google's standard SQL data types reference has the exact ranges.
Four questions that expose the real cost before you sign
Ask the vendor whether it reloads tables or reads change data capture after the backfill, and how many read calls one run makes for your entity list. Ask which Intuit partner tier it holds, because that decides whether it can be blocked late in the month. Ask what lands for the invoice line table, column by column, and what type amounts use. Ask how deleted and voided transactions are marked. A vendor that answers all four in writing has built for finance. One that answers with a feature list has not. The concurrency side of the same question, the 10 connection cap per company, is covered in QuickBooks API rate limits.
Where Adapters fits
Adapters maps each QuickBooks entity onto a typed BigQuery table, backfills with paged queries at full page size, then reads change data capture only, with retries, alerts and per-record logs. The price is flat by plan with no per-row charge, so a busy month does not change the bill. If you also need QuickBooks connected to the rest of your stack, start with QuickBooks integration tools, open the QuickBooks to BigQuery connector to see the mapping, and check plans on pricing.
A QuickBooks warehouse that reads only what changed
Map the entities once, declare exact decimals, backfill at full page size, then sync changes only. Flat plans from $49 a month, with retries, alerts and per-record logs.
The live demo needs no card, and Starter is $49 a month.