Skip to content

Decide what to connect first

Connect the one system that answers most of the questions your business keeps asking, and leave the rest for later.

  • Permissions: anyone who’s signed in can connect a system from Connect in the rail. Uploading a file into a shared workspace takes a Creator or Admin role there. Outside onboarding, only admins can load the sample company.
  • Data you need: none yet. You need your questions, and a rough idea of which systems your team works in.
  • Credits this uses: uploading files and syncs you start yourself cost nothing. A scheduled sync uses a credit each time it syncs something, and each message to the Librarian uses one.

It’s tempting to connect everything at once. Don’t. Each system brings its own tables, names and gaps, and one is much easier to check than six.

Start with the questions, not the systems. Write down the handful your business asks every week or month: the ones that come up in meetings, or that someone spends an afternoon in spreadsheets to answer.

If you went through onboarding, you already have a start. Open Library, then the Questions tab. Anchor questions have a question-mark icon, and the narrower ones under them say “refining question”. To add one, ask the Librarian, for example “Add a question: which customers pay late?“

Next to each question, write the systems that hold the numbers today. You don’t need table names. “The online store” or “the finance system” is enough.

Here’s how that looks for five of Curio’s anchor questions:

QuestionWhere the data lives
What is our revenue by channel (eCommerce, retail, B2B)?Shopify, Square and NetSuite
What were total eCommerce sales last quarter?Shopify
What is our current finished-goods inventory by warehouse?NetSuite
Which B2B accounts also purchase through eCommerce?Salesforce and Shopify
What is our on-time delivery rate by carrier?ShipStation

Count how often each system appears. The one with the most questions goes first. For Curio that’s Shopify: three of the five questions need it, and one needs nothing else.

What about a tie? Pick the system behind the question people argue about most, or the one that’s quicker to connect. A sign-in connector opens the provider’s sign-in window, while a database needs its connection details, which you may have to ask someone for.

A question that spans systems, like Curio’s revenue by channel, is only fully answered once all of them are in.

The difference shows up next month.

Upload a fileConnect a system
Best forA one-off export, or a spreadsheet someone keeps by handData that changes, and questions you’ll ask again
Staying currentOnly when you upload a new version with Replace file…Refresh by hand, or on a schedule
Setting it upDrag the file onto the pageSign in, or enter connection details
CreditsNoneNone for syncs you start. A scheduled sync uses 1

Either way, Querri answers from a copy, so answers are only as fresh as the last upload or sync. CSV, Excel, JSON and Parquet files become data you can ask about, while PDFs, images and documents join the Library as files.

Connectors overview lists what you can connect, and File Uploads covers files.

Curio, the sample company, comes with data from the kinds of systems a business runs on: finance and accounting, a CRM, an online store, point of sale, ad platforms and shipping. Load it to see what connected data turns into before you connect your own.

  • During onboarding, choose Explore with sample data on “Let’s bring your data in.”, then A ready made company.
  • After onboarding, an admin can install it from a workspace’s Demo data tab in Settings.

If you’re an admin, give Curio a workspace of its own, so its collections, questions and KPIs don’t sit alongside yours. Explore the sample company has the steps.

First, switch to the workspace the data belongs in. A new connector goes into the workspace you’re in if you’re a member, and into your private workspace otherwise.

Click Connect in the rail to open Add data. Find your system under Connect a data source, or add your file under Upload from your computer.

Once the new data settles, Querri marks the questions it can now answer. Go back to the Questions tab and read the badges:

BadgeWhat it tells you
ANSWERABLEThe data is there. Ask it
PARTIALSome of the data is there. Your list from step 2 shows what’s still missing
UNANSWERABLEThe data isn’t there yet. Connect another source, or refine the question
No badgeNot checked yet. The Librarian works it out as it answers

Then ask one of your ANSWERABLE questions in Librarian chat and look at what comes back.

  • The connector is listed under Connectors in the Library’s browse panel, or your file is on the Files tab as a Dataset.
  • More of your questions read ANSWERABLE or PARTIAL than before.
  • One of them gets an answer you’d stand behind. See Check an answer before you share it.

The connector says it’s connected, but no data has loaded yet. Click the connector under Connectors in the Library, then Load data now. Only the connector’s owner can do this.

Your system isn’t in the list. Some sign-in connectors only appear once Querri has set them up for your environment. If yours still isn’t there, Don’t see what you’re looking for? sends a request. Meanwhile, export the data to a file and upload that.

A file ended up in your private workspace. You can’t add files to the workspace you were in. Ask a workspace admin for a Creator role, then upload it again from that workspace.

The collection page says No view yet. That’s about views, not data, and it doesn’t mean your data can’t answer the question. Check the badge on the Questions tab instead.

An Inbox card says Can’t answer this yet. Querri found a question your data can’t answer yet. Accepting the card records a fact. It doesn’t build a view or connect anything, so treat it as a pointer to the next system.