Data on its own rarely tells you what to do. Knowing that a customer placed an order in March, another in June, and a third in September is just three facts sitting in a row. But the moment you turn those three facts into a single number, total spend this year, or average days between orders, you suddenly know something useful. You can act on it.
That act of turning raw data into a meaningful number is the job of a calculated insight.
What a calculated insight is
A calculated insight is a metric that Data Cloud computes across your data. Instead of looking at one record at a time, it summarizes many records into something you can use, like a single value attached to a customer.
Common examples make this concrete:
- Lifetime value — the total a customer has spent with you across all their orders.
- Engagement score — a number that reflects how often someone opens emails, visits the site, or interacts with your brand.
- Average order value — total spend divided by number of orders.
- Days since last purchase — how long it has been since they bought anything.
None of these live in your source systems as a ready-made field. They are calculated by looking across many rows of data and rolling them up into one clear measure.
Why the unified profile makes this powerful
Here is the part that matters most. By the time you build calculated insights, your data has already traveled the full journey: connected through data streams, harmonized into a shared model, and unified through identity resolution into one profile per person.
That last step changes everything. Without unification, a customer’s spend would be split across a “John in e-commerce” and a “J. Smith in Salesforce,” and any total you calculated would be wrong, only counting half their orders. Once those records are unified into a single profile, a calculated insight can finally sum all of John’s purchases, no matter which system they came from.
A metric is only as trustworthy as the profile it sits on. Calculated insights become genuinely reliable only after identity resolution has gathered each person’s scattered records into one.
This is why lifetime value computed in Data Cloud means something. It is not a partial picture from one system. It is the whole person, measured honestly.
Where these metrics go to work
A calculated insight is not just a number you admire. It feeds the steps that follow, and this is where it earns its keep.
Smarter segments
Segments group people by shared traits so you can reach them with relevance. A calculated insight gives you far better traits to group by. Instead of a crude “everyone who bought once,” you can build “customers with lifetime value over a threshold who have not purchased in ninety days.” That is a segment worth acting on, and it only exists because the underlying metrics were calculated first.
Grounding for AI
When an Agentforce agent helps a customer, it is far more useful if it knows that this person is a high-value, long-loyal customer rather than a brand-new visitor. Calculated insights provide exactly those summarizing facts. They give the agent context that no single raw record could.
In both cases, the insight is the bridge between raw history and a confident decision.
How they are computed
You do not calculate these by hand. You define an insight once, describing what to measure and how, and Data Cloud runs it across your data and keeps the result current as new data arrives. Many insights refresh on a schedule, so a customer’s lifetime value quietly updates as new orders flow in.
The thinking work is in choosing which metrics matter to your business. A subscription company might care deeply about engagement and churn risk. A retailer might care about average order value and recency. The platform does the arithmetic; you bring the judgment about what is worth measuring.
A gentle reminder about meaning
Years of teaching taught me that a grade is not the point; it is a shorthand that points at real understanding underneath. A calculated insight is the same. Lifetime value is not the goal. It is a clear signpost toward a real relationship with a real person, and toward the next helpful thing you can do for them.
So when you define your first insight, ask the human question first. What would I most want to know about this customer if they walked through the door? Then let Data Cloud calculate the number that answers it.
Start small. One or two insights that genuinely change how you treat a customer are worth more than a dozen metrics nobody uses. Build the ones that lead to action, and let them sharpen every segment and every AI conversation that comes after.
Your next step
See where your new metrics make the biggest difference:
- Your First Segment in Data Cloud — put calculated insights to work grouping the right people.
- Identity Resolution for Beginners — the unification that makes these metrics trustworthy.
- Browse more in the Data Cloud category.