A BPO gives you judgement at scale. It cannot give you consistency at scale.
Both of those are worth buying, and they are not the same purchase. A hundred agents will hold a difficult negotiation better than any AI. They will also deliver a hundred slightly different calls, cost the same whether the customer answered or not, and only ever have five per cent of their work reviewed.
A settlement argued down, a furious customer talked round, an exception handled with discretion. Real skill, and it does not automate away.
Every account attempted at the right hour under the right rules, every call scored, and the ones needing judgement put in front of your best people.
Read that as a division of labour rather than a contest. The mistake we see is buying one where the work needed the other: putting agents on 40,000 low-balance reminders, or putting a bot on a settlement negotiation.
Seven things a floor and a platform do differently.
| What you are comparing | A contact-centre floor | Oriserve |
|---|---|---|
| How capacity is added | Another seat, another trainer, another six weeks of ramp. | Another line of concurrency, available the same day. |
| What quality depends on | Which agent picked up, and what kind of week they are having. | The same script, tone and rulebook on call one and call one hundred thousand. |
| Coverage | Shifts. Evenings and Sundays cost more, and the customer who is only reachable at 9pm often is not reached. | Every hour the rules allow you to call, at the same cost per call. |
| QA coverage | A sample, typically under five per cent, scored by hand. | Every call recorded, transcribed and scored. |
| Where the money leaks | Volume you cannot afford to work: small balances, low-value renewals, long-tail queries. | That volume becomes serviceable, because cost per attempt collapses. |
| Judgement calls | This is the real strength. A good agent negotiates, reads the room and closes. | We route these to your people, with context, rather than pretending to replace them. |
| Commercial shape | Cost per seat or per hour, whatever the hour produced. | Attached to the outcome, against a baseline agreed up front. |
Same SLAs. A quarter of the seats.
That is the clearest evidence we have that this is a change in the ratio rather than a change in the promise. The service levels held. The human layer stayed, doing the work that needed a human. What left the floor was the repetitive volume that was never a good use of a trained agent in the first place.
The saving is not the whole prize, and it is often not even the main one. The bigger number is usually the volume you were never able to work: the small balances, the low-value renewals, the customers reachable only at inconvenient hours. When cost per attempt collapses, that book becomes serviceable, and revenue you had written off starts landing.
Do not automate these. We will say so in the meeting.
- Negotiation with real authority. A settlement, a waiver, a restructure. Anything where the answer is a decision rather than a lookup belongs to a person who can make it.
- Relationship accounts. High-value customers who expect a name they recognise. Efficiency is not what they are buying.
- Situations that need a human of record. Some regulatory and dispute conversations require an accountable individual, and no amount of audit trail substitutes for that.
- Low volume, high variance. If a process runs forty calls a month and every one is different, the engineering does not pay for itself. Keep the agent.
- Everything beyond the phone. Back office, verification, fulfilment, the wider outsourcing relationship. We do conversations; a BPO does considerably more than that.
The floor does not change on day one.
Start where the floor is losing money, not where it is winning
The unworked tail, the reminder volume, the after-hours gap. Nobody has to give up work they are good at to prove this.
Agree the baseline with the operations team in the room
Current outcome rate, current cost, current coverage. The team that runs the floor should own those numbers, because they will be asked about them afterwards.
Wire the handoff into the existing floor
Transfers land with your agents, in their tooling, with the context attached. If this step is done badly the whole thing fails, which is why we treat it as the product.
Move the ratio deliberately
Expand AI coverage as the outcome rate holds, and let the seats that remain move up to the calls where judgement pays. Headcount planning follows the evidence rather than a target set in advance.
Are you telling us to fire our agents?
No. We are telling you the ratio changes. The volume work moves to AI and the judgement work stays with people, which usually means fewer seats doing higher-value calls rather than an empty floor. The 200-to-50 example on this page kept the SLAs; it did not remove the human layer.
We outsource to a BPO. Does this end that relationship?
Not necessarily, and the better BPOs are already moving this way themselves. If your provider brings the AI, you get the change without changing vendors. If they will not, you have a decision to make about who owns the automation layer in your account. We work either way, including through the BPO.
What about the calls that genuinely need a person?
They get a person. That is the design, not a concession. What changes is that the person arrives already knowing who they are speaking to, what has been said, and what was promised, so the call starts at the decision instead of the introduction.
Our BPO contract has committed volumes.
Then the honest first move is a process outside that commitment, or the volume you currently do not work at all. Most books have a long tail nobody can afford to call. That is the cheapest place to prove this.
How do you price against a per-seat contract?
Per seat is a cost input; we price against the outcome with a baseline agreed before we start. The calculator puts both side by side using your own seat cost and volume.
Bring the AI yourself before someone brings it to your client.
The clients you run today are all being sold voice AI this year. The only open question is who sells it to them. We would rather it was you, with us as the backbone and the client relationship staying yours.