AI-ready: A setup guide for PowerMetrics

Published 2026-04-22

Summary: The difference between a generic AI response and a genuinely useful one almost always comes down to the same thing: context. This guide shows you how to add this all-important layer to your PowerMetrics account.

A Brilliant New Analyst

Think of the PowerMetrics AI Assistant as a brilliant new analyst — fast, tireless, and capable of synthesizing your entire metric catalog in seconds. But, like any analyst, it’s only as effective as the context you give it.

Walk them into a catalog of metrics with confusing names (like MRR_v2_FINAL), inconsistent dimension labels, and no descriptions? They'll do their best, but it probably won't be great.

Walk them into a catalog where metrics are clearly-named (like, Monthly Recurring Revenue) and meaningfully-defined, dimensions are consistently labelled, and tags tell you which metrics matter most? They become extraordinary.

It’s all down to AI readiness. This isn’t a technical configuration. It’s a discipline everyone can follow to add context, depth, and relationships to your catalog.

Before diving into the actions you should take to get your account ready for AI, let’s look at the PowerMetrics AI Assistant, how it works and why you can rely on it for secure data access and insightful exploration.

What the PowerMetrics AI Assistant Can Do

Uncover insights — go beyond what happened to why it happened. Ask "Why did CAC spike in March?" or "What's driving the Vancouver revenue increase?" The Assistant will surface patterns from across your catalog in seconds.

Create visualizations — describe what you want in plain language and get charts and dashboard previews instantly, without manually configuring metric IDs or dimensions.

Get help — ask how to do anything in PowerMetrics and get immediate guidance and coaching without leaving the interface.

Explore your catalog — find the right metric across a large account without knowing its exact name (especially helpful for large accounts with many metrics). Ask "What retention metrics do we have?" and get a curated, descriptive set of options.

How it Ensures Secure Access

The Assistant operates on permission parity: it can’t see or query anything that the user doesn't have access to. It’s not a back door to sensitive data — it's a natural language interface that works within your existing permissions.

Your data stays within PowerMetrics. It’s never sent to external services and is never used to train public models.

By default, only admins have access to the AI Assistant. If desired, they can extend access to editors and viewers in the account settings section.

Four Essential Actions for AI-Readiness

1 — Names and Descriptions

The most effective thing you can do to improve AI (and human) consumption is to give your metrics descriptive names. A metric called MRR is workable. A metric named Monthly Recurring Revenue with a description of "Recurring revenue from active subscriptions. Includes expansion. Excludes trial-to-paid conversions." is a superpower.

The name tells the AI what the metric is. The description tells it the logic — what's in, what's out, how it behaves. That logic is what allows the AI to explain your data, not just report it.

Start here. Add meaningful descriptions to your 10 most-used metrics. With this added layer of context, these few metrics will deliver more AI value than a hundred name-only metrics.

2 — Building Metadata: Tags, Certification, and Favourites

Metadata includes everything that describes a metric without being the metric itself. For example, its name and description (discussed in step #1) but, also, its tags, dimensions, owner, and intended audience. Robust metadata helps the AI prioritize the metrics you and your team use the most.

Adding tags to metrics (and dashboards and data feeds) organizes assets in your account into groups, making it easier for the AI to find related items. For example, if you ask "What are our most important growth metrics?", the AI will focus on what's tagged #BoardReady or #Growth — and deprioritize what's tagged #Draft or #Deprecated. The same logic applies to certifying and favouriting your most used, trusted metrics.

The best-practices of deprecating or removing old/broken/stale metrics

Unused, outdated, or duplicate metrics create noise that makes it harder for the AI to find the right answer. When the AI finds five metrics that could all represent "churn," it has to ask for clarification or guess — neither is a great user experience.

Tag metrics that are intentionally experimental, stale, old or in-progress so the AI (and users) know to be cautious (or explicitly prompt the AI in the personalizations to ignore certain tags)

Think of tags, certification, and favourites as a table of contents for the AI. Without this added layer, everything looks equally important. With it, the AI knows exactly what to look for.

Add depth: Apply tags to related metrics, dashboards, and data feeds. You can ask the AI Assistant to help you identify themes, make suggestions, and even apply the tags. Certify validated metrics and favourite the ones you refer to the most.

3 — Composable Calculated Metrics

When a calculated metric, like LTV:CAC Ratio, is built from documented base metrics, the AI can trace the lineage and explain the math. When it's a hard-coded number with no documented source, the AI can report it but can't explain or diagnose it. Composable metrics give the AI a chain of reasoning, not just a value.

Define base metrics: It’s important to include a description for every metric, of course. But, when it comes to calculated metrics, make sure that description includes the metric’s formula and important details about each of its base metrics.

4 — Consistent Dimension Naming

Apply consistent dimension names across the account. For example, if two metrics are looking at data for the same dimension (i.e., they both measure “x by country”) and one metric uses Country and the other uses Geo, the AI will have difficulty connecting them. When performing cross-metric analysis and asking questions like "How does ad spend compare to signups by region?" the same concept needs to use the same name everywhere for fast, accurate results.

Rename dimension fields: When creating new metrics, use business-language, universally-understood dimension field names. The best time to do this is when you create a metric for the first time. But, you can also edit existing metrics to harmonize and align their dimension names. It’s a bit of a short-term effort, but pays off big in productivity moving forward.

Level up data-driven decision making

Make metric analysis easy for everyone.

Get Started Now

Configuring the AI for Your Account

Account-Level Settings (Admins Only)

Go to the “General Settings” section in the account admin UI to configure account-level options. Set the AI's tone and behaviour by entering specifications in the system prompt. Think of this as the AI's onboarding brief — you're telling it how to communicate before it meets your team. You can also keep the AI Assistant as admin-access only (the default setting), or enable it for admins and editors, or for all users.

Personalization (Admins, Editors, and Viewers)

Personalization shapes the AI's responses for you only — it doesn't affect anyone else. Access these settings via the three-dot menu in the AI Assistant.

Set your role, industry, and response style. Most importantly, use the free-form context field to share your key metrics, internal terminology, customer segments, and anything else that helps the AI give relevant answers. A user who writes "Our key metrics are ARR and NRR. We call our customers 'Members'." will get fundamentally different responses than one with no context set at all.

Tip for success: Treat personalization as a first-day task, not an optional setting.

AI Readiness Checklist

Your Catalog

[ ] Top 10 most-used metrics have clear, plain-language names — no abbreviations or version numbers
[ ] Each metrichas a description covering what's included, what's excluded, and where the data comes from
[ ] Primary dimensions are named identically across all metrics that share them
[ ] Calculated metrics reference documented base metrics — the formula is traceable
[ ] Draft, legacy, or experimental metrics are tagged for deprioritization
[ ] Active metrics are tagged by team (#Sales, #Marketing) or audience (#BoardReady)
[ ] Data feeds are refreshing on a schedule your team understands

Your AI Settings

[ ] Account-level tone and behaviour configured (admins)
[ ] AI Assistant access confirmed for the right roles in General Settings (admins)
[ ] Personalization completed: role, industry, response style, business context (everyone)
[ ] Business context field includes key metrics, terminology, and relevant segments

Pro-Tips: Let the AI Help You Get AI Ready

The AI Assistant can do much of this work for you. These prompts are the fastest way to start.

Start with a full audit
"Audit and analyze the assets in my account. Identify any quality gaps and suggest ways to improve my setup."

Draft missing descriptions
"List all metrics in the [Marketing] category that are missing a description. Based on their names, suggest a one-sentence description for each."

Surface and tag low-quality metrics
"Identify low-quality metrics and suggest tags to classify them."

Check dimension alignment
"Compare the dimensions for [New Signups] and [Ad Spend]. Do they use the same name for geographic location? If not, what should I rename them to?"

Bootstrap your tagging system
"Review my top 10 most-used metrics. Suggest 3–5 tags for each that would help a new user understand their purpose."

⚠ Data Freshness: The AI's answers are only as current as your data. If your data feed refreshes weekly, asking "How are we doing today?" returns data that may be days old. Ask the AI "When was this data last updated?" if recency matters.

Closing Thoughts

The PowerMetrics AI Assistant is ready. The question is whether your catalog is ready to meet it.

Everything in our 4-part concept series — choosing the right metrics, defining them precisely, building them composably, documenting their relationships — was always building toward this. A catalog that any person, or any AI, can walk into and immediately understand.