Most questions asked of a provider are answered by the brochure. The ones below are not, and the value is frequently in whether an answer exists at all rather than in what it is.
What is your attrition rate, by role, over two years?
The single most informative question. A provider who has the figure is measuring the thing that will affect you most; one who has to estimate is not.
Expect a number that is high by the standards of employment generally, because this industry is. What matters is whether it is being tracked, whether it differs sharply between roles, and what the provider believes drives it.
Who owns the recruitment decision?
If you cannot interview, and cannot decline, you are being allocated a person rather than making a hire. Both arrangements exist and only one of them is usually described accurately.
The follow-up: how many candidates will I see, and what happens if I decline all of them?
What happens when somebody leaves?
Replacement timeline. Whether you pay during the gap. Who pays for the second onboarding. And whether the leaver is required to write a handover, which almost never happens unless it was agreed at the start.
Is the person dedicated, and what does that mean here?
Dedicated is used loosely. Ask whether the person works on any other account, whether they could be moved without your agreement, and what happens to your work when they are on leave.
The leave question is the one that surprises people. Philippine statutory leave plus public holidays is a real number of days and somebody should have told you what happens on them.
Who employs the staff?
The provider, or a subcontractor. Subcontracting is legitimate and it changes who holds the employment relationship and what happens if the subcontract ends. It is answerable in one sentence and should be asked in writing.
What is the office arrangement, precisely?
Whether staff are on site or working from home, which changed in this industry after 2020 and has not changed back everywhere. Whether the connection is redundant. What the power backup is and how long it lasts. Whether there is a documented plan for a typhoon.
The last two are not formalities in this country and the entry on infrastructure explains why.
How is my data handled?
Where it sits, who can access it, whether staff use their own devices, what happens on the day somebody leaves. The answer should include a process for revoking access rather than an assurance about trustworthiness.
What do you do badly?
Ask it directly. Every provider has an answer and the useful signal is whether they give one.
A provider who says nothing goes wrong has either not been in business long or is not telling you things, and both matter more than anything on the price list.
Can I speak to a client with a team like mine?
Not a case study. A conversation, with somebody running a comparable team in a comparable role. The entry on checking references sets out what to ask them.
The question they should ask you
How many hours a week of describable, checkable work the role contains, and how you measured it.
A provider who does not ask, or who accepts an estimate without probing it, is not going to tell you when a role is too thin. That is the failure the first part of this journal is entirely about, and it is avoidable in the first conversation by either party.
How to run the conversation
Send the questions in advance. A provider given time to look up their attrition figure gives you a real one; a provider ambushed gives you an estimate, and you learn less from the ambush than you think.
The exception is the question about what they do badly, which is worth asking live.
What your answers should be
Have the written role, the measured volume and the named person who will brief it. A provider assessing whether you are a good client is looking for those three, and turning up without them makes the good providers cautious and the poor ones enthusiastic.
The follow-up that gets the real answer
After any figure, ask how it is measured. Attrition can be counted as leavers over headcount, as leavers over average headcount, annualised from a quarter, or with the first ninety days excluded, and those produce very different numbers from the same underlying reality.
A provider who explains their method is giving you something usable. One who has not thought about the method has a number rather than a measurement.