Customer

You do not need a complete customer graph to take the next action on this case.

Companies fund 360-degree views, personalization engines, and journey maps, then the agent or seller still opens five systems to see what happened last week. Intelligence that does not travel with the case is unused.

50%

marketing output gains in studies of structured, measurable work.

Stanford HAI, AI Index 20261
Adnan Boz
Adnan Boz

How I see it

A 360 view that does not change a case is a warehouse project

Customer intelligence is useful when it changes a decision on a live case: who to contact, what to offer, what broke last time, whether this guest or account is at risk, and what policy applies.

The expensive rebuild is familiar. A seller, a coordinator, or a service agent reconstructs history from CRM, tickets, PMS, orders, and email. That reconstruction is the first surface, not a new CDP program.

Personalization and lifetime-value scores are easy to present. They become operating leverage only when they appear inside the working list and the case, with an owner who will act when the signal is wrong.

Start with one action: a retention save, a service escalation, an account brief, or a campaign exclusion. Use the fields you already trust. Do not wait for a complete identity graph.

Where the cost sits

Intelligence on a case

minutes of reconstruction × volume of that case type × loaded cost, plus the miss cost of the action you failed to take

Use one action. Lifetime value is a later term. Reconstruction is the term you can count now.

Is this your problem?

Five signals the cost is already real.

  1. Which customer action would change this month if the signal were in the case?

    If you cannot name it, you want a profile, not intelligence.
  2. Where does the person already work the case?

    The signal has to land there. A new portal will not be opened.
  3. Which source do people already trust: CRM, tickets, orders, or PMS?

    Start there. Do not wait to reconcile every system.
  4. Would a saved view on existing fields produce most of the value?

    If yes, do not fund a model to avoid a list.
  5. Who is accountable when the signal causes a bad contact?

    Unowned intelligence becomes optional, then ignored.

What done looks like

The result, not the category.

A solution is finished when a named result no longer waits on reconstruction.

01

Reconstruct the case once

History, policy, and system state are assembled before the agent or the person answers.

Done when

The first response does not start with let me look that up.

Not done if

A bot greets the customer and then transfers an empty case.

02

Close it in the system

The ticket, the note, and the next action are written back. Deflection is not closure.

Done when

The case is done in the system of record.

Not done if

Containment goes up and reopen rates stay the same.

03

Keep a human on the exception

Authority, refunds, and messy history stay at a named control point.

Done when

Routine cases complete and exceptions have an owner.

Not done if

Every case still needs a person, or none do.

What usually fails

How solution shopping wastes a quarter.

01

Building the graph before naming the action

A complete profile with no working list is a reporting project.

02

Personalization with no offer owner

A recommended next product that nobody is staffed to sell is decoration.

03

Treating marketing data as the service record

The last complaint, exception, or missed delivery is often the signal that matters, and it does not live in the campaign tool.

This is not the problem if

Do not start with a category purchase.

  • The request is a CDP or journey program with no case owner.
  • Nobody will change a working list even if the signal is right.
  • Identity across systems cannot be joined for the case type you named.
  • A better note in the existing CRM would remove most of the hunt.
Adnan Boz

A useful next step

Bring one expensive problem. Get guided into production.

We guide the implementation, go deep on the technical path, and stay hands-on through operations — or tell you when a simpler answer is better.

Discuss an AI opportunity