KPIs for Insurance Brokers: What to Measure Beyond Revenue

Revenue per broker is the number most brokerages report on, and it deserves its place. It is the commercial result, it is comparable year on year, and it is the measure a principal is accountable for. What it does not do well is explain itself. Three things limit it as a management measure. The first is timing. A renewal conversation held in March shows up in a June number, and a new business enquiry handled badly in one quarter shows up as an absence in the next. By the time the figure moves, the conversations that moved it are months old and nobody can recall them precisely enough to act.

The second is attribution. Revenue moves for reasons that sit outside any individual's control. The 2026 Vero SME Insurance Index, which surveyed 1,250 SMEs and 250 large businesses in September 2025, found that 47 per cent of businesses reported a decline in revenue over the preceding 12 months, with micro and small businesses under the greatest pressure. The same research reports that six in 10 smaller businesses saw their premiums increase. A book can grow on rate while the broker holding it is losing clients, and it can shrink while the broker holding it is doing the best work of their career. A single figure cannot separate those two situations.

The third is granularity. Revenue per broker is one number per person per period. There is no version of that number that tells a broker what to do differently on Tuesday, which is what a coaching conversation needs. Development happens at the level of the behaviour, and revenue is three or four steps downstream of any behaviour.

None of this is an argument against measuring revenue. It is an argument for measuring alongside it. The activity that produces a commercial result runs through client conversations, and a brokerage that records those conversations is holding more evidence about performance than its reporting usually draws on. Some of that evidence is already counted: the call metrics a phone system reports natively describe how reachable a firm is and what a conversation costs. The rest sits inside the recordings, in what was actually said, and it is the part that explains the numbers rather than restating them.

A useful KPI set does three jobs that a revenue figure alone cannot. It gives a broker something they can change this month. It gives a manager something specific to talk about in a review. And it lets a firm notice a problem in the conversation layer before it arrives in the commercial layer, which is generally somewhere between one and three renewal cycles later.

Broker performance metrics fall into four categories, and a set that covers all four behaves very differently from a set that covers one or two. Each category answers a different question, and each one has a known weakness that another category checks.

Activity measures what a broker did. Calls taken and made, quotes issued, renewal conversations completed before expiry, client review meetings held, referrals requested. These are the easiest to count, because most of them are already recorded somewhere in the phone system or the CRM without anyone deciding to collect them. Activity answers whether the work happened at the volume the role assumes.

Quality measures how the conversation was handled. This is the category that most often has no number attached to it, because it requires a written definition before it can be scored at all. A brokerage that has decided what excellent sounds like on its own calls can score against that definition: whether expectations were set before the client asked, whether cover was translated into the client's own operation, whether an unraised exposure was raised. Those four observable behaviours are specific enough to be scored consistently by two different reviewers, which is the test a quality measure has to pass.

Client outcome measures what changed for the client. Retention at renewal is the obvious one. The more useful ones sit earlier: the proportion of clients who have had a documented risk conversation in the past 12 months, the proportion of recommendations that were acted on, referral rate, claims experience where a claim occurred. The Vero research gives some sense of the room available here. Around 60 per cent of businesses reported receiving risk advice from their broker, and two thirds of those acted on it. Among businesses that had not developed a business continuity plan, 22 per cent said a recommendation from their insurance broker would motivate them to develop one.

Compliance discipline measures whether the elements the firm has decided a conversation should contain actually happened. A brokerage sets that list itself, and what belongs on it depends on the service, the advice and the client. Common examples are a disclosure made and confirmed, the scope of the advice stated, the fee arrangement explained, and a file note completed inside the firm's own window. This category is often run as an exception report, which makes it the one most likely to be misread: a clean report can mean nothing went wrong, or it can mean nothing was looked at.

Each category carries a weakness the others cover. Activity without quality rewards volume. Quality without client outcome rewards performing to a scorecard. Client outcome without activity rewards holding a good book rather than working it. Compliance discipline without the other three describes a broker who completes the form and never has the conversation. Read together, the four produce a picture; read alone, any one of them points somewhere unhelpful. A workable set is six to eight measures in total, spanning all four categories, with one or two per category. Larger sets tend to be assembled because each individual measure looked reasonable, and they stop being read within a quarter.

The difficulties below are properties of measurement rather than of the people doing it, and they show up in well-run firms as readily as anywhere else.

Firms that get value from broker KPIs tend to share a small number of design choices, most of which are decisions rather than tools.

An illustrative target, offered as an example of the shape rather than as a recommended value. Assume a firm decides every commercial client should have one documented risk conversation a year, and that it holds 240 commercial clients spread evenly across six brokers. That is 40 clients each, 40 conversations each over a year, and across roughly 46 working weeks it lands at about one documented risk conversation per broker per week. Every figure here is an assumption, including the even spread and the working year. The point of the example is the shape rather than the number: the target is weekly, countable, and entirely within the broker's control, which is what a revenue target is not.

Callyx.ai works on the call recordings a brokerage already holds. Nothing changes about how brokers take calls, and no review task is added to anyone's day. The Core capability is included on all plans; Comply, Coach, Insights and Learn sit on the Professional plan.

Results are delivered after every call to an inbox, a CRM or a practice management system, depending on how the firm is set up. Callyx.ai is Australian-owned and runs on Australian data infrastructure. None of this replaces the judgement of the person running the review. It changes what that person is working from.

None of these five steps requires new software. They require a decision about what the firm wants to be able to see, and a written definition for each thing on the list. Step five is where a KPI set turns into coaching rather than reporting, and it is the one most often deferred, because it is the only step that needs a recurring slot in somebody's calendar.

Revenue per broker describes the result and not the work. It arrives late, it moves with rate and market conditions as well as with performance, and it offers a broker nothing to act on this month. Measuring alongside it, across activity, quality, client outcome and compliance discipline, gives a brokerage a picture that explains the commercial number instead of restating it.

The design choices matter more than the tooling. Six to eight measures rather than twenty. Targets on what a broker can influence inside a month. Activity expressed as a floor. Quality scored against a definition the team helped write. Every measure paired with a counterweight, everything segmented before anyone is compared, and the numbers going to the individual rather than to a wall.

Two of the four categories have historically been difficult to measure at any scale, because quality and compliance discipline both live inside the conversation rather than beside it. For a brokerage that already records and retains its client calls, that is a reading problem rather than a data problem. Callyx.ai turns those recordings into scored, searchable evidence against a standard the brokerage sets itself, which is what makes a KPI set describe the whole team rather than the part of it someone had time to listen to.

Every measure on the list below evidences something real and every one of them bends under a target. The third column is not a reason to drop the measure. It is the reason each measure is read next to its counterweight.