Most people measure a new remote seat by whether it feels like it’s going well. That works until month four, when someone asks what it’s actually delivering and nobody can answer with anything except an impression.
Ninety days is the right window — long enough for a real answer, short enough to change course. Here is what to look at in each month, and what to ignore.
Month one: is the work correct?
Only one question this month, and it isn’t about volume. It is whether the output meets your standard.
The check that works takes half an hour: pull ten pieces of their actual work and look at them properly. Ten loads they tracked — were the check calls made, are the notes usable by the next shift? Ten invoices they built — complete, correct references, right the first time? Ten calls, if the seat is phone-based — listen to three.
That sample tells you more than any report. It also tells you something a report never will: whether the errors are random, which is normal and coachable, or systematic, which means they misunderstood the job and nobody caught it.
Expect a rough patch around week three. Supervision comes off and volume goes up at the same time, and something gets missed. That is the normal shape of learning a job, not a sign the hire was wrong. What matters is that the feedback is specific and fast.
Month two: has it given time back?
This is the one people forget to measure, and it is the actual point of the hire.
Ask the person whose work was supposed to be offloaded what changed. If your carrier rep is still chasing PODs, the seat has not done its job — however well the new person is performing. That is a scoping failure, not a performance one, and the fix is redistributing the work rather than replacing the person.
Two useful measures, both rough and both fine:
- Hours back. Ask your team for an honest estimate of hours a week no longer spent on the handed-over work. Precision doesn’t matter; direction does.
- What filled the gap. If the freed hours went into selling or coverage, you have a return. If they quietly evaporated, you have a cost — and that is worth knowing at month two rather than month twelve.
Month three: is it compounding?
By now the difference between an adequate seat and a good one is visible, and it shows in three places.
- Questions are down. Not zero — down. Someone who has stopped asking anything has usually stopped caring or started guessing.
- They flag things before you do. “This shipper keeps moving Thursday appointments” is worth more than any metric on this page, because it means they are thinking about the work rather than processing it.
- The exception rate is falling. Fewer things needing your involvement, on the same volume.
What to measure per seat
Whatever you already use is usually right. If you have nothing, these are enough:
- Track and trace — check calls completed on schedule, exceptions escalated inside your rule, hours between an issue occurring and someone knowing.
- Back office — days from delivery to invoice, share of invoices going out complete first time, ageing over 45 days.
- Carrier sales — loads covered, margin, and how much of the coverage came from repeat carriers rather than the board.
- Sales development — qualified conversations booked, not dials made.
- Customer service — first response time and how many threads close without escalation.
What not to measure
Hours logged and screenshots. Monitoring software measures presence, not work, and it tells a capable person you don’t trust them. The output sample in month one answers the same question better and doesn’t cost you their goodwill.
Raw activity. Calls made, emails sent, tickets touched. Easy to count, easy to inflate, and it rewards volume over judgement. An SDR who books three good conversations beats one who makes two hundred dials and books none.
The ninety-day conversation
Put it in the calendar at the start, with your vendor and with the person. Three questions: what is working, what is still landing back on your team, and what should this seat be doing in the next quarter that it isn’t doing now.
Most seats that quietly underperform do so because nobody ever had that conversation, and by month six the disappointment has hardened into a decision.
The short version
Month one, check the work is correct by sampling ten real pieces of it. Month two, check your team actually got time back — that is the point of the whole exercise. Month three, check it is compounding. Ignore hours logged and raw activity, and book the ninety-day conversation before you need it.
