If you’re scaling a multi-vertical agency and still staring at total revenue like it tells you everything, you’re kidding yourself. Revenue hides the mess. It hides bad-fit clients, overworked staff, sloppy scoping, and verticals that look exciting in the sales deck but chew through margin.
There are agencies that hit $3 million, $5 million, even $10 million in annual revenue, and still do not know which vertical actually pays the bills. They know they’re “growing”. Great. But getting bigger isn't the same as getting smarter, and blind momentum is a dangerous way to run a business.
Track gross margin by vertical
Start here. Not total agency margin. Vertical margin.
If you serve healthcare, construction, SaaS, professional services, and ecommerce, each one behaves differently. Different buying cycles. Different delivery expectations. Different levels of account management handholding. If you lump them together, the strong verticals carry the weak ones, and you miss the problem until cash gets tight.
Track gross margin by vertical every month, ex GST, with labor costs included properly. That means salaries, super, contractor spend, and the ugly bits people love to ignore. If one vertical brings in $120,000 a month and spits out 18% gross margin while another brings in $90,000 at 42%, the second one wins. Every day of the week.
The last time somebody ignored this, one fast-growing vertical looked brilliant on the topline and absolutely rotten underneath. The pricing was fixed, two bad-fit accounts were cut loose, and that lifted the agency's gross margin from 31% to 38% inside two quarters. Same agency. Better decisions.
If you can’t answer “Which vertical actually makes us money?” in under thirty seconds, you’ve already got a reporting problem.
Watch sales velocity, not just lead volume
The leads are nice. Closed business is nicer.
Too many agency owners brag about how many opportunities to sit in the pipeline. Who cares? A bloated pipeline usually means one of two things: weak qualification or slow sales. Sometimes both, which is a fun little disaster.
Track:
- Lead-to-opportunity conversion by vertical.
- Opportunity-to-close rate by vertical.
- Average sales cycle length by vertical.
- Average first-year revenue per new client.
- CAC payback period.
You need to know how quickly each vertical turns effort into cash. A vertical with a 2%-win rate and a 120-day sales cycle might still work, but only if the contract values justify the drag. A vertical with a 28-day cycle and smaller retainers can be far more useful when you need predictable monthly intake.
For Australian agencies, this matters even more when you’re hiring ahead of demand. Wages don’t wait politely while procurement “circles back after EOFY”.
Measure utilization, but don’t worship it
Utilization matters. Blind worship of utilization wrecks teams.
Tracking billable utilisation by role, by team, and by vertical is essential but treating one flat target as gospel is a mistake. A strategist at 90% billable for three straight months isn't efficient. They're trapped. They stop thinking, start reacting, and quality falls off a cliff.
For most service teams, healthy utilisation sits in the 70% to 80% range, with room for internal improvement work, proposal support, and fixing delivery issues before they turn into expensive apologies. If project managers or senior specialists keep sitting above 85%, capacity is gone. Hire, re-scope, or raise prices. Pick one.
Track these together:
- Billable utilisation.
- Effective hourly rate.
- Scope creep percentage.
- Write-offs and rework time.
- Average turnaround time.
Why together? Because a team can look “fully utilised” while burning hours on work, nobody scoped, nobody billed, and nobody wants to repeat. That’s not productivity. That’s unpaid overtime with a dashboard.
Retention tells you if your model actually scales
New business gets attention because it’s loud. Retention is quieter, but it tells the truth.
New business gets attention because it’s loud. Retention is quieter, but it tells the truth. Getting retention right requires more than tracking numbers; the underlying customer retention strategies you have in place determines whether the data ever improves.
Watch:
- Revenue retention by vertical.
- Client tenure by vertical.
- Net revenue retention if you upsell existing accounts.
- Concentration risk, meaning how much revenue sits with your top 5 or 10 clients.
- Delivery issue rate in the first 90 days.
I’m harsh on early churn. If clients leave within six months, that’s rarely “just how the market is”. It usually means bad sales fit, bad onboarding, or a delivery model built on wishful thinking.
One agency with strong acquisition numbers in healthcare had lousy retention. The issue wasn't demand. The issue was onboarding. The process looked fine on paper and completely collapsed once busy clinic managers got involved. The first 30 days were rebuilt; reporting was simplified, and churn dropped by 22% over the next two quarters.
Track operational fit before expanding a vertical
Not every vertical deserves scale. Some deserve respectful goodbye.
This is where most agencies get sentimental. “But we’ve won great logos there.” Good for you. Are those logos profitable? Are they repeatable? Can your team deliver without reinventing the wheel every time?
Look at operational fit through a few ugly but useful questions:
- How much custom process does this vertical demand?
- How much senior oversight does it need?
- How often do scopes blow out?
- Can we templatise delivery?
- Does the client tech stack make execution smoother or harder?
Take healthcare. If your team works with clinics or allied health providers, your workflows often get tighter when clients already run solid systems. Delivery has been seen to improve fast when accounts use specialist medical practice software because the reporting is cleaner, admin friction drops, and the client side can actually act on what you send them. That doesn’t magically save a bad engagement, but it removes a lot of dumb operational drag.
That’s the point. Scale the verticals where your delivery model gets sharper with repetition, not shakier.
Talent depth is a scaling metric too
Most agency dashboards underplay talent because it feels less neat than sales and finance. That’s a mistake.
If you want to scale across several verticals, you need to know whether your hiring pipeline matches the complexity of the work. Otherwise sales get ahead, delivery scrambles, and your best people spend Friday night fixing work that should never have gone out.
Track:
- Time to hire by role.
- Time to productivity for new hires.
- Bench strength in each vertical, meaning how many people can lead work without supervision.
- Staff turnover by team.
- Training completion tied to real delivery capability.
This matters even more in technical or industrial accounts. If you’re servicing manufacturers, utilities, or mining-adjacent businesses, domain familiarity counts. A candidate with hands-on exposure or a qualification like Certificate III in Instrumentation and Control often ramps faster because they already understand the environment, the terminology, and why clients care about precision. That doesn’t replace agency skill, but it cuts the nonsense phase where everyone pretends to understand each other.
Time to competence is a real metric. If a hire takes six months to become useful in one vertical and eight weeks in another, that affects the scaling plan whether it's convenient or not.
If a metric doesn’t change a decision, bin it
That’s the filter.
Every metric on your dashboard should change one of these decisions:
- Where do you invest.
- What do you sell.
- Who you hire.
- Which clients do you keep.
- How do you price.
If it doesn’t help you make a sharper call, toss it. Nobody needs another dashboard that looks impressive in a board meeting and useless on Monday morning.
A lean scorecard with ten hard metrics beats a bloated one with fifty vanity charts every time. Track margin by vertical. Track sales speed. Track utilisation with context. Track retention. Track operational fit. Track talent depth. Then act on what the data shows, even when the answer is uncomfortable.
%201.png)
%201.png)
%201.png)
