Athena 101 · Lesson 17 · Answer from governed numbers
Transcript
Transcript
Get a straight answer to a business question, from numbers your whole team defines the same way. Here’s what you’ll learn.Ask three people what freight cost means, and you’ll get three answers. An approved model gives everyone one. Type @ and the model’s name to add it to your question. Priya, Sam’s analyst, defined freight cost and budget once, and deployed them. Open Metrics to read them. Sam doesn’t build anything. Every number he gets comes from the same definitions.You get the reason behind a number, not only the number. Ask the question you actually have, which is usually a why. Here, why Midwest freight is over budget this month, by carrier and by charge type. Athena reads the numbers from the approved model, split the way you asked. The answer says where the overrun comes from. Here it’s detention, mostly from two carriers. Every claim carries a citation back to the numbers behind it.Before you repeat a number, check it. It takes one click. Hover a citation to see where it came from. The chip says Cube, the layer the approved numbers come through. Click it. The model opens and reruns the same query against current data, so you see the rows behind the answer. The badge shows whose access was used: your own connection, or a shared workspace credential, if that’s how your firm is set up.An answer in chat changes nothing by itself. Put the decision where your team already looks. Sam asks Athena to add it to the weekly exceptions memo: who acts, what they do, and by when. The decision lands at the end of the memo, and the citation travels with it. Anyone reading the memo can check the number the same way Sam did.If the question comes back every week, give it a dashboard. Ask for one from the same model. It reads the same approved numbers, so it matches the memo. When a dashboard reads straight from the model, Athena checks each viewer’s access as it loads the data.Quick check. A number in the answer looks wrong. What do you do first? Click its citation. The model reruns the query, so you see the rows.Here’s how a field sales team uses this. Every week, reps ask the same model why a territory is behind budget. Each answer has citations, so a rep can check a number before the conversation. The action they agree goes straight into their weekly note.Now try it with a model your team has deployed. When it works, you’ll see a query card, then citation chips you can click.
Before you start
Upload these files from the Northwind starter kit: The sample model, Northwind Freight Costs.xlsx, and Weekly exceptions memo.docx in Lesson 17. Ask your admin and analyst for a working approved data model and connection to practice live queries. Otherwise, use the workbook and sample definitions to discuss the decision and update the memo. The sample model alone cannot answer numerical questions; live queries and dashboards are watch-only without the working model.Steps
1
Start with approved definitions
Mention the Northwind Freight Costs model. Read its approved definitions so everyone uses the same meaning for freight cost and budget.
2
Ask the business question
Ask why freight is over budget, split by carrier and charge type. Read the explanation and check that each claim links to its evidence.
3
Check the source
Open a citation to rerun the question against current data and inspect the supporting rows. Check whose access was used for the answer.
4
Record the decision
Ask Athena to add the action, owner, and deadline to the weekly memo. Check that the citation stays with the decision.
5
Make a small dashboard
Ask for a dashboard from the same approved model when the question repeats. Check it against the memo; each viewer’s access applies as the data loads.
Try it
@[your approved model] Why is [cost] above or below target this month? Break it down by [category]. Cite each number.Look for a question answered from the model, with citations you can open.