Sooner or later, almost every contact or back-office operation ends up with a variable component in the agent's pay. The underlying idea is reasonable: pay more to whoever does the work better. The problem is that incentive pay doesn't motivate in the abstract. It motivates exactly what it measures, with a literalness that tends to surprise whoever designed it.
The pattern repeats. A bonus on handle time is introduced, handle time drops the following month, and a quarter later reopens go up. A bonus on cases closed is introduced and premature closures appear. Nobody cheated: the scheme asked for that and the team delivered it.
An incentive scheme is an instruction, not a prize
It helps to stop reading the bonus as recognition and start reading it as what it is: the most precise description the company ever gives of what it considers good work. The manual says one thing, training says another, and the bonus calculation tells the truth. When the three disagree, the bonus wins.
Hence the first practical rule: before picking metrics, write in one sentence which behaviour you want to see more often. If that sentence can't be written, the scheme isn't ready to be paid out.
What can carry money on top of it
Not every metric survives having money attached. One works as an incentive only if it meets three conditions at once.
- The agent controls it. If the result depends on incoming volume, on the quality of the list, or on a staffing decision, it isn't theirs.
- It's measured the same way every day. If the definition shifts depending on who pulls the report, month-end will be spent arguing about the calculation instead of the work.
- It has a counterweight. No volume metric should ever pay on its own.
Most broken schemes fail the first condition. Paying sales conversion to an agent who doesn't choose who to call transfers a commercial risk they never took. Paying individual SLA attainment when the SLA breaks for lack of headcount is worse: it punishes precisely the person who stayed to cover.
Volume and quality, always paired
The rule that prevents the most damage is simple: every productivity metric comes with a quality metric, and the quality one works as a gate, not as an addend.
The difference isn't cosmetic. If quality and productivity add up separately, an agent can offset mediocre quality with volume. If quality is a gate, there is no bonus below a given standard regardless of output. The message is different and so is the behaviour.
For that gate to be legitimate, the score has to come from an audited sample with written criteria and calibration between auditors, not from the impression of whoever supervised that shift. How that sample is built is in quality control by audited sampling. A bonus resting on badly measured quality isn't an incentive: it's a lottery with merit written on it.
Be careful with customer satisfaction
Putting satisfaction into individual incentive pay is tempting and usually goes badly. Only a fraction of customers respond, skewed towards the very happy and the very angry, and the per-agent sample in a month ends up too small to separate performance from chance. On top of that, much dissatisfaction doesn't originate in the interaction at all but in the product, the price, or a policy the agent merely relays.
If it is used, keep it at team level, with a threshold rather than a slope, and read it alongside other signals. What each instrument actually measures is in CSAT, NPS and CES.
Adherence: useful, but not as the only lever
Schedule adherence is genuinely within the agent's control and it does affect service, so it's a reasonable candidate. The risk is making it the centre of the scheme: you end up rewarding presence rather than outcome, and the team learns to be logged in without being available. Why adherence weighs so heavily on real coverage is in shrinkage and adherence in BPO.
Individual, team, or both
Individual pay moves behaviour fast and erodes cooperation: nobody helps the person next to them if it costs them their own number. Team pay protects collaboration and dilutes the signal, especially in large groups. A mix usually works better in practice: an individual component on what the person truly controls, and a team component on what can only be achieved together, such as shift coverage or clearing a whole queue.
When the team component is applied to very large groups it stops being felt. The natural unit is normally the supervisor's cell, which is also the unit where something can be corrected. How that cell is organised is in how a BPO team is structured.
The size of the bonus changes the agent's economics
A bonus that is too small changes nothing: chasing it isn't worth the effort. One that is too large turns monthly income into something uncertain, and income uncertainty is one of the most consistent reasons people quit in this industry, alongside the schedule. What actually makes people stay is worked through in attrition in BPO.
There is no universally correct percentage; it depends on the process and the local labour market. What is constant is the criterion: the fixed part has to be enough to live on without the bonus. If it isn't, the scheme stopped being an incentive and became a transfer of risk onto the agent.
The agent should be able to do the maths
A good scheme fits on one page and the agent can reproduce the calculation from data they can see. If they have to wait for month-end to find out what they earned, the incentive wasn't operating during the month, which is exactly when it should have been. Three conditions help: few metrics — two or three, not seven — daily visibility of progress, and a channel to dispute a calculation with a response deadline.
If the agent can't work out their own bonus, the scheme isn't incentivising anything: it's handing out money at month-end.
The legal side, in general terms
In Colombia, how variable pay is agreed has effects well beyond that month's payroll. As a general rule, what directly remunerates the service rendered counts as salary and enters the base for benefits, contributions and severance; salary-exclusion agreements and occasional bonuses have their own conditions and limits, and commissions are normally treated as variable salary. It also helps for the rules of the scheme — what is measured, how it is calculated, when it is paid and how it can be changed — to be written and communicated rather than held up by custom.
This is general framing and not legal advice. Any scheme should be reviewed with legal or employment counsel before being implemented, particularly if it changes what was already being paid.
Signs the scheme is doing damage
- The incentivised metric improves while another, non-incentivised one gets worse at the same time.
- Agents argue about the calculation more than they discuss cases.
- Almost the whole team hits the maximum, or almost nobody reaches it.
- Shortcuts appear that nobody taught: quick closures, convenient dispositions, avoidable transfers.
- The supervisor spends month-end making manual adjustments so the result “comes out fair”.
Any one of those five shows up before the damage reaches the service. What to watch on the dashboard to catch them early is in what to measure in a BPO operation from month one.
How smartBPO works it
We design incentive pay with quality as a gate rather than an addend, and with few metrics the agent genuinely controls. We write the rules before starting and publish them, with progress visible during the month and a channel to dispute calculations. We separate what belongs to the agent from what belongs to the team and from what is an operational decision, so nobody is charged for a result that depended on a staffing call. And we revisit the scheme when the process changes, because an incentive that was right a year ago may be rewarding the wrong behaviour today. We don't claim an incentive scheme improves quality on its own: it supports what supervision and feedback are already doing, and it shows badly when it's set up wrong.