PrymeIntelligence

Customer support·Template

Retention offer matrix

A retention offer matrix for AI support agents: offers by reason for leaving and situation, caps, eligibility and fairness rules, with a filled sample.

Customers who want to leave should get offers based on facts, not on how hard they push. A retention offer matrix sets what may be offered by reason for leaving and situation: the offer, its cap, eligibility, frequency, and whether an AI agent may make it. If the customer still wants to leave, they leave.

Why put retention offers in a matrix?

Retention is where support teams have traditionally used the most discretion, and where outcomes vary most between customers. Two people with the same contract and the same complaint can leave the same call with very different deals. An agent following vague guidance would reproduce that variation at scale, or apply the most generous offer to everyone.

A matrix fixes both. Offers depend on stated, recorded facts, and the same facts produce the same offer. That is the argument in retention offers need limits, not judgement: the judgement goes into designing the matrix, and the agent applies it.

When should the agent use it?

When a customer says they want to cancel, downgrade or leave. The agent first asks why, records the reason, then checks eligibility and offers what the matrix allows, once. Whether an AI agent can offer a discount to keep a customer covers the conditions, and the step-by-step process is in how to make retention offers within set limits with an AI agent.

How do you fill in each field?

  • Reason for leaving. A fixed list: price, service problems, moving, found a better deal, no longer needed, financial difficulty, other.
  • Customer situation. Tenure, contract status, value band. Keep it to facts in your systems.
  • Offer. Discount, plan change, added feature, waived fee. One offer per cell, or a short ordered list.
  • Maximum value and duration. The cap, such as a discount of 10 a month for six months.
  • Eligibility. For example, not offered a retention deal in the past year.
  • Agent depth. Depth 3 to make the offer within the cell; depth 2 to draft for a person.
  • Fairness rules. Applied to every cell: one offer, no pressure, cancellation completed on request.

Retention offers are not the same as goodwill. A goodwill gesture apologises for a failure; a retention offer is a commercial incentive. Keep them apart, with the goodwill gesture matrix covering the first.

The retention offer matrix

Reason for leaving Situation Offer Max value and duration Eligibility Agent depth If declined
Process cancellation

Fairness rules (all cells): make one offer; do not repeat it after a clear no; never suggest cancellation is difficult; complete cancellation in the same conversation where the product allows.

A filled sample: a mobile network, out-of-contract customers

Reason Situation Offer Max value and duration Eligibility Agent depth If declined
Price Tenure over 2 years Move to current SIM-only tariff with same allowance Price of that tariff Not on a retention deal in past 12 months 3 Process cancellation
Price Tenure under 2 years Discount on current plan 5 a month for 6 months As above 3 Process cancellation
Found a better deal elsewhere Any Match plan from current range, if one is within 5 a month Current range only As above 3 Process cancellation
Service problems Any No offer; route to fix the problem Not applicable Not applicable Escalate to technical support Process cancellation if customer still wishes
Moving abroad Any No offer Not applicable Not applicable 3 Process cancellation
Financial difficulty Any No sales offer; lower-cost plan information; free advice signposting Not applicable Not applicable 2 Process cancellation or hold for payment support

Three cells carry the ethics of the matrix. Service problems get a fix, not a discount, because a discount on a broken service fixes nothing. Moving abroad gets no offer, because there is nothing to retain. Financial difficulty gets no sales offer at all, and a person reviews the conversation, because pushing a customer in difficulty to stay on any plan risks harm.

What should the agent record?

The reason given, the offer made, the customer's answer, and, if they left, the time cancellation was completed. Those records let you check that offers were consistent and that no customer was obstructed. The wider stance is in the AI support agent that knows when to say no.

Questions people ask

Should the agent make a second offer if the first is declined?

No. A second offer after a clear no turns retention into pressure. If the customer asks whether anything else is available, the agent can answer from the matrix.

Can different customers get different offers?

Yes, when the matrix says so, based on recorded facts such as tenure. Not because one customer pushed harder. That consistency is the matrix's purpose.

Is this related to data retention?

No. A retention offer is about keeping a customer. Data retention concerns how long you keep personal data, a separate topic covered in whether an AI agent can delete data when retention periods end.

Decide what your agents may do. Prove it. Enforce it.

Pryme Intelligence certifies each agent on your own cases, checks every action against five bounds at the moment it happens, and files every verdict.