Getting answers with the PowerMetrics AI Assistant (Beta)

PowerMetrics is designed to give you an in-depth, contextual understanding of your data. Our AI Assistant turns data analysis into conversational insights, helping you identify current and future trends, perform historical comparisons, apply best practices, and take effective action.

Use the AI Assistant to:

• Get actionable answers. Practical observations, eye-opening visualizations, and helpful summaries that inform next steps.
• Reveal a holistic view of your account, by exposing the relationships between your metrics, data feeds, dashboards, and users.
• Create visualizations and add them to dashboards for collaboration or to Explorer for further investigation.
• Iteratively build dashboards you can preview before adding them to your account.
• Manage asset tags to keep everything organized.

This article includes:

• Secure, reliable AI-assistance
• Where do the answers come from?
• Account-level settings
• Preparing your account for the AI Assistant
• Chatting with the AI Assistant
  • Personalizing your chat experience
  • Starting a new chat
  • Managing chats
    • Opening chats
    • Renaming chats
    • Deleting chats
• Interacting with the AI Assistant
  • Working with chat-generated visualizations
  • Working with chat-generated dashboards
  • Managing your asset tags
• Personalization examples
• Prompt tips and suggestions

Note: By default, all users can access the AI Assistant. Admins can modify this setting in Settings > AI Assistant (PowerMetrics only) > Account Access to enable or disable access for various user types in the account.

Secure, reliable AI-assistance

The PowerMetrics AI Assistant provides:

• Governed access: The AI Assistant respects the access settings as configured in your PowerMetrics account, ensuring everyone sees only what they’re supposed to see.
• Read-only protection: The AI Assistant can’t edit, delete, or overwrite your data, metrics, or dashboards.
• Trusted data: By exposing relationships between assets in your account, the integrated knowledge graph provides an in-depth, holistic understanding of your data that naturally results in more reliable, relevant answers.
• Privacy: No data from within your metrics is sent to any external service. All data processing, queries, and analysis is performed within the PowerMetrics platform.
• Reliability: The AI Assistant only refers to data in your account so, while it may sometimes misinterpret your questions, its answers are always based on your own, trusted data.

Where do the answers come from?

The AI Assistant is designed to give trustworthy, contextual responses to your questions. When you enter a prompt, the AI engine looks for relevant answers in reliable sources, including:

• Your PowerMetrics account. This information deep-dive is powered by our MCP Server and Knowledge Graph. Together, they give you a complete view of your data, including the relationships between your assets (users, metrics, data feeds, and dashboards) – enabling answers to questions like: “Compare this month’s MRR to the same month last year”, “What operands are included in this calculated metric?”, “Who’s the owner of this marketing dashboard?”, and “Show me a list of data feeds that have no associated metrics”.
• The PowerMetrics Knowledge Base. By connecting to our internal documentation, the Assistant helps you navigate and understand the product to get the most of your PowerMetrics experience.

Account-level settings

Account admins control user-access to the AI Assistant. They can also enter a system prompt to improve Assistant responses, aligning them with the tone and needs of the company.

To set account-level options:

1. Click the button in the left navigation sidebar and select Settings > General.
2. Under AI Assistant (PowerMetrics Only):
   • At Account Access, choose between Enabled for all users (the default setting), Admins only, Admins and Editors, or Disabled for all users.
   • At System Prompt, enter information that will help the AI Assistant understand your business, brand, and response preferences. Note: You can enter up to 5000 characters.

Preparing your account for the AI Assistant

There are a few things you can do to help the AI Assistant return accurate, meaningful answers. Like any AI tool, the Assistant works better when it has context. It’s much easier for it to return the best possible answers when:

• Assets in your account are well-defined and organized (consistent naming, clear descriptions, asset tags)
• It’s obvious which metrics serve as your single-source-of-truth (SSOT) (favourites, certification)

Let’s look at some practical steps to get your account AI-ready. Bonus: These steps don’t only improve AI Assistant responses, they also make the self-serve analytics experience better for everyone.

Defining and organizing assets

It’s important to clearly name and define the metrics, dimensions, data feeds, and dashboards in your account. This doesn’t only make it easier for the AI to gather related information, it also helps your team understand the data they’re working with. In addition, applying tags to assets helps keep everything organized and searchable, a boon for users and the AI Assistant.

• Using a clear, well-understood name for each metric helps the Assistant provide fast, relevant responses to your analytics questions. Ambiguity causes confusion and misalignment—prevent it by purposefully naming all of your metrics. The same goes for dashboards and data feeds.
• Don’t forget about dimensions—they need clear, consistent naming too. We recommend you name dimensions carefully when building your metrics, however, it’s never too late to clean them up! Time spent editing metrics to align dimension names is well-worth it.
  Here’s an example of why consistent dimension naming matters: Let’s say you have three different base metrics, all with a dimension for “city”. In the first metric, you call that dimension “city”, in the second, you call it “store_location”, and in the third, you call it “town”. The PowerMetrics system and the AI Assistant see these differently-named dimensions as three separate things. If you build a calculated metric that combines these base metrics, it inherits the dimensions, but, since they’re named differently, it can't be filtered by “city” because the differently-named dimensions can’t act as a shared lens.
  Tip: Need help finding inconsistencies? Ask the AI Assistant to “Review the dimensions being used across my metrics. Are there any overlaps, naming inconsistencies, or gaps I can improve?”
  For more examples and best practices, check out our Metric Naming Conventions Guide.
• When building or editing metrics and data feeds, consider adding a brief description, for example, what a metric measures, who it’s for, or how it’s calculated. This is a great way to provide extra context to the AI Assistant and to your team. When building assets, remember, what might seem obvious to you now may not be clear to others later, including your AI!
  Tip: Include synonyms or business logic in your metric descriptions, for example, “This is our primary Revenue metric, also referred to as 'Top Line' or 'Gross Sales'. Calculated as Price * Quantity."
• Add asset tags to metrics, dashboards, and data feeds to make it easier for the AI to connect related items and to help users quickly find and interact with groups of related content. Learn more about asset tags.
  Examples: Ask the AI to only return information for metrics that have the tag “Marketing Key Metrics”. Or, ask it to ignore all “in-development” metrics that have the tag “draft”.
  Tip: Ask the AI Assistant to help you identify themes, make suggestions, and even apply tags to your assets–learn more managing tags with the AI Assistant.

Signifying SSOT metrics

Tell the AI Assistant (and your team members) which metrics matter most:

• When you set metrics as favourites, the AI Assistant’s responses prioritize them over other similar metrics. Learn how to set metric favourites.
• Certify a metric to indicate it as a verified source of truth. This lets the AI Assistant know to return answers for it over other similar, but uncertified, metrics. As a bonus, it also shows business users which metrics have been validated and approved for use. Learn about metric certification.

Chatting with the AI Assistant

Working with the AI Assistant is a collaborative process where business users interact with the AI to iteratively derive insights from their company’s data.

Personalizing your chat experience

As a first step, we recommend you personalize your chat settings. Not only does this ensure responses are styled based on your preferences, it also helps the AI deliver better answers by learning more about your role, industry, and goals.

Need some guidance? Check out these personalization examples.

Note: This section describes selecting individual preferences. Your account admin applies account-wide AI Assistant settings.

To personalize your chat experience:

1. In the left navigation sidebar, click AI Assistant.
2. In the AI Assistant window, in the upper-right corner, click the 3-dot menu and select Personalization.
3. Under Your role, enter relevant information.
4. Under Your industry, enter relevant information.
5. Under Response style, use the drop-down menu to choose a writing style.
6. Under Additional details, enter any information you think would help the AI Assistant understand who you are and what kind of information/help you’re looking for. Note: You can enter up to 5000 characters.
7. Click Save.
   Note: The format and content of future chats rely on the information you enter here. If you change your personalization entries later, your new entries don’t build on your previous ones, they replace them.

Starting a new chat

Chats build iteratively based on your prompts and the AI’s responses. To change the context to discuss a new topic, start a new chat.

Not sure where to begin? Select a suggested prompt to get started. (See some examples below.)

Note: The number of prompts/user/day and prompts/account/day is subject to fair use limits.

To start a new chat:

1. In the left navigation sidebar, click AI Assistant.
2. To start a new chat:
   • In the AI Assistant window, enter a prompt.
   • In an open chat, click the New chat button or click the 3-dot menu and select +New chat. Enter a prompt.
     Tip: You can pause a chat midstream by clicking the Stop button and restart it by clicking the Regenerate button. After getting an answer, you can click to rerun your question.

Managing chats

After receiving a response, a chat can be opened for further investigation, renamed, or deleted.

Note: Chat management actions are often performed in the Chat history page. To get there, click AI Assistant > Recent > View all or click the 3-dot menu in the upper-left corner of the AI Assistant window and select Chat history.

To open chats:

1. To open a single, recent chat, select it from the list under AI Assistant > Recent. (This list can include up to five chats.)
2. To open a chat from a list of all your chats, in the Chat history page, click the name of the chat you want to open. Note: You can also click the 3-dot menu for a chat to open it in a new tab.

To rename chats:

1. In and open chat or in the Chat history page, click the 3-dot menu for the chat and select Rename.
2. Enter a new name for the chat and click Save.

To delete chats:

1. To delete a single, open chat:
   • Click the 3-dot menu in the upper-left corner of the window and select Delete chat.
   • In the pop-up dialog, click Delete again to confirm deletion.
2. To delete a single or multiple chats:
   • In the Chat history page, select the checkbox beside the chat(s) you want to delete, then click the Delete button.
   • In the pop-up dialog, click Delete again to confirm deletion.

Interacting with the AI Assistant

Chat with the AI Assistant to learn everything there is to know about your data. When you discover something interesting, it’s easy to dig deeper or share your findings with others as metric visualizations and dashboards. The AI Assistant can also automate tag management, saving you time and keeping everything in your account organized and accessible.

Working with chat-generated visualizations

While chatting and building metric visualizations with the AI Assistant, if you discover something you want to investigate further, you can go to the metric’s overview, open the visualization in Explorer, or add it to a dashboard. You can also easily share your metrics with others.

Note: When creating metric visualizations alongside the AI Assistant, you can access the metrics that currently exist in your account – you can’t create new metrics.

Tip: Hover over a value in the generated visualization to access query context and calculation details (see the following example):

Access the following options from the visualization’s 3-dot menu:

• Expand the visualization to take a closer look.
• Open the visualization in Explorer for further investigation.
• Click the metric’s name to: Open the metric overview in a new tab, share the metric, or learn more about it.
• Add the visualization to a dashboard.

Note: To build and interact with visualizations for specific metrics, you must have appropriate access to the metric, for example, you must be the owner or it has to be shared with you.

Working with chat-generated dashboards

Using natural language prompts, you can work with the AI Assistant to build comprehensive dashboards, styled your way. You’ll interact with a preview, making it easy to iteratively build the dashboards you need before adding them to your account. Once added, you can edit your chat-generated dashboards just as you would any PowerMetrics dashboard.

Note: If you want to include images, you’ll do so after the dashboard has been added to your account. Learn how to add images here.

To build a dashboard:

1. In the AI Assistant window, enter a prompt.
   Tip: Start simple, for example, ask the Assistant to “Build a dashboard that shows my GA4 data”. Or, get more descriptive with a prompt like “Create a dashboard for my Q1 sales performance including total revenue, top-selling products, and regional growth”. No matter what you initially ask the AI to create, you can keep chatting and iterating to make fine adjustments or even rebuild the entire dashboard.
2. Click the Dashboard preview field to open it. (See example below.)
   
3. If you’re happy with the dashboard as-is, in the preview, click Create Dashboard.
4. If you want to make changes, the options are virtually endless. For example,
   • You can refine a dashboard’s visual appearance by asking the Assistant to modify the colour scheme, apply a dark or light theme, or change the layout or visualization types.
   • You can adjust the data being displayed by changing the date range, adding or removing metrics, adding a period comparison, or adding or changing applied filters.
   • When you're done modifying the dashboard preview, click Create Dashboard.
5. The dashboard opens in your PowerMetrics account. If you want to modify it further, click the Edit button. Learn more about editing dashboards.

Managing your asset tags

Asset tags are used to organize and categorize metrics, dashboards, and data feeds. The AI Assistant offers an automated way to manage these tags in your account. Learn more about tag management.

• Applying tags: Users can instantly apply tags to assets with the AI Assistant by entering a suitable prompt. For example, tell the Assistant to “Add the tag “Marketing Team” to the “Ad Spend”, “Campaign Reach”, and “Engaged Sessions” metrics".
• Removing tags: Similarly, you can instantly remove tags from assets by entering a prompt such as: “Remove the “Key Metric” tag from the “Emails Bounced” and “Emails Clicked” metrics".
• Replacing tags: You can replace tags with a different tag by entering a prompt like: “Replace the “2020 metrics” tag with a “Legacy metrics” tag in the “2020 Goal Conversions” and the “2020 Leads” metrics".
• Adding new tags: You can ask the AI Assistant to add new tags to your account. Your request can be specific, for example, "Add a new tag called "2026 metrics" and apply it to the "2026 Goal Conversions" and the "2026 Leads" metrics". Or, you can ask the AI to suggest tags. It will create a few recommended tags that you can review and edit before applying them to your metrics, dashboards, or data feeds. You'll be prompted to approve the addition(s), so you can cancel the action if desired.
• Deleting tags: If there are tags you no longer need, ask the Assistant to delete them, for example, enter a prompt like "Delete the "2025 Campaign" tag". You'll be prompted to approve the deletion, so you can cancel the action if desired.

Personalization examples

Here are a few examples you can refer to when entering your personalization settings.

Prompt tips and suggestions

The prompts you enter when chatting with the AI Assistant are only limited by your imagination. Here are some ideas to get you started:

• To analyze data trends, enter a prompt such as “Compare this month’s MRR to the same month last year” or “What drove the increase in customer churn last quarter?”
• For forecasting purposes, ask “Can you forecast sales for the next quarter?” or “If sign-ups continue at this rate, what will they be by the end of the year?”
• For general data exploration, ask questions like “What metrics are using the same dimensions as Revenue Growth Rate?” or “Why did cost per acquisition increase yesterday?”
• To see relationships and lineage, enter a prompt like “Show me all the metrics created by Susan” or “Which operand metrics are included in our Customer LTV calculated metric?” or “Give me a list of the data feeds that have no metrics.”
• For quality monitoring, ask the AI Assistant to “Show me metrics with data gaps” or “Tell me which metrics haven’t been updated in the last 7 days” or “Which metrics failed to refresh today?”
• To understand usage, enter a prompt like “Which dashboards are viewed most often?” or “Which metrics in the account are starred by the most users?” or “Is Average Order Value being displayed on any dashboards?”
• To see what’s being shared, ask “Which metrics are shared with the marketing team?” or “Give me a list of the dashboards that are being shared with published views.”