Spark is your AI analyst in Alchemer Dashboard. Ask questions about your data in plain English from a dedicated place in the navigation, get charts and written answers back, and pick up past conversations right where you left off.
Spark requires the Advanced Dashboard add-on. If your account doesn't have it, you won't see Spark in the navigation. To learn more about access, see Alchemer Dashboard Permissions Overview.
Open the Spark chat space
Select Spark in the top navigation. It's the first item, just to the left of Overview. The chat space opens with a new chat, ready for your first question.
You can still ask quick questions from the Spark prompt bar at the top of the Overview page. The chat space adds a full-page view and keeps your conversations so you can come back to them.
What's on the page
| Area | What it does |
|---|---|
| My Conversations panel | Holds your chats on the left side of the page. Use the collapse icon to hide the panel and give answers more room. |
| New Chat | Starts a fresh conversation. |
| Chats | Lists your past chats, with the most recent activity at the top. |
| Get Started with Spark | Opens the Getting Started with Spark article. |
| Question bar | Sits at the bottom of the page. Choose a data source, pick an answer mode, and type your question here. |
Ask a question
- Select New Chat.
- Choose a data source from the picker in the question bar. Select All data models to let Spark look across your data, or choose a single data source. When you choose a single data source, select Preview data to see sample rows.
- Choose Quick Search or Deep Analysis.
- Type your question in the Enter your question box, then select the send button.
While Spark works, it shows short progress updates, such as "Searching for a data model" and "Visualizing data," so you can follow along.
Good to know: Spark is an AI system, so review its answers before you act on them. The notice under the question bar is a reminder of this.
Quick Search or Deep Analysis
| Answer mode | Use it when |
|---|---|
| Quick Search | You want a fast answer to a specific question, like a single metric or a trend over time. |
| Deep Analysis | You want Spark to dig deeper into a broader question with a multi-step analysis. |
Start over
Spark treats each new question in a chat as a follow-up to the ones before it. To ask something unrelated, select Reset in the question bar or start a New Chat.
Read an answer
Each answer includes:
- Show work: Expand it to see the steps Spark took to reach the answer.
- A written summary of what Spark found.
- A chart, in most cases. Tokens across the top of the chart show how Spark built the query, such as the measure, any filters, and the time grouping.
- Key insights, and sometimes a suggested follow-up question.
Work with the chart
| Option | What it does |
|---|---|
| Chart and table views | Switch between the chart and the rows behind it. The card shows how many data points or rows are displayed. |
| Expand icon | Opens the chart in a larger view. |
| Pin | Pins the chart to a dashboard. |
| Save | Saves the chart so you can use it again. |
| Download | Downloads the chart. |
| Edit | Opens the chart so you can change it. For details, see Getting Started with Spark. |
Hover over a bar or point to see its values. Under each answer, you can:
- Select thumbs up or thumbs down to tell Spark whether the answer was helpful.
- Select Copy response to copy the answer so you can paste it elsewhere.
- Select Double check response to ask Spark to verify its answer. Spark sends "Please double check the last response" and runs the analysis again.
You can also reply to a specific part of an earlier answer. The quoted text appears above your new question, so it's clear what you're asking about.
Pick up a past conversation
Spark saves your chats and names each one from your first question. To reopen a chat, select it in the Chats list. Spark loads the history and shows a This is a past conversation. banner. Keep asking questions to continue where you left off.
Hover over a chat to see its full title and a ... menu, where you can rename or delete it.
Good to know: Chats from the earlier Spark pop-up aren't carried over to the chat space, and chats can't be shared by link.
When Spark can't answer
- The chart can't be drawn. Sometimes a chart type doesn't fit the data, for example a map when location values are empty. The card shows a message, and the table view still shows the data.
- Spark can't answer the question. Spark tells you in the chat and suggests what to try next, such as asking a more specific question, naming a data source, or asking it to try again.
Tips for better answers
- Be specific. Name the measure, the time period, and how to group it, for example "Show average NPS by month for the last six months."
- Choose a single data source when you know where the answer lives.
- Build on an answer with a follow-up question instead of starting over.
- Make sure your data is ready for Spark. See Spark Dataset and Model Readiness.
- Learn how to edit charts and coach Spark in Getting Started with Spark.
FAQ
Do I need Advanced Dashboard to use Spark?
Yes. Spark requires the Advanced Dashboard add-on. Without it, the Spark item doesn't appear in the navigation.
Can I still use Spark from the Overview page?
Yes. The Spark prompt bar on the Overview page works as before. The chat space gives you a full-page view and saves your conversations.
Can I build a whole dashboard with Spark?
Yes, with SparkViz in the dashboard editor. Describe the dashboard you want and SparkViz builds it for you. See Getting Started with SparkViz.
Can I share a chat?
Not by link. To share a result, copy the response, pin the chart to a dashboard, or download it.
