Killing Your Best Service Line: When an Agency Should Sunset a Profitable Offering

The hardest cut we ever made was a service line that still had a 38% margin and a waiting list. It funded payroll for two years. Killing it was the right call, and we waited about nine months too long to make it.

Most agency advice tells you how to launch new offerings. Almost nobody talks about the inverse problem: when to retire one that's still working. That gap is where a lot of mid-sized shops quietly stall — not because they picked the wrong service, but because they couldn't let go of one that had outlived its strategic role.

Why profitable services are the hardest to kill

Unprofitable work is easy. You look at the P&L, you have an uncomfortable meeting, you move on. Profitable work is sticky in ways that don't show up in a spreadsheet.

It anchors your sales pitch. It's what referrals come in for. It's the thing two of your best engineers are emotionally attached to because they built the playbook. And critically — it's almost always the highest-utilisation line, which means it's also the one with the most operational gravity.

That gravity is the problem. Every hour those engineers spend on the legacy line is an hour they're not spending on whatever your next bet is. The opportunity cost is real, but invisible, and it compounds.

A profitable service line you should have killed two years ago looks identical on the P&L to one you should keep forever. The difference is in the second derivative, not the headline number.

The five signals it's time to sunset

We use five signals. None of them alone is enough. Three or more, and we start the conversation seriously.

1. Margin is high but trending down

Not the absolute margin — the trajectory. A service that went from 55% to 38% over three years is telling you something even if 38% still looks healthy. Usually it means commoditisation: more competitors, more open-source equivalents, or clients getting smarter about scoping.

We pull margin per quarter, not per year. Annual smoothing hides the slide.

2. The work no longer teaches you anything

Early in a service line's life, every project produces a new internal tool, a new pattern, a new bit of tribal knowledge. Mature service lines produce... invoices. If your team hasn't shipped a meaningful internal improvement to how you deliver this work in 12+ months, you've extracted the learning value. You're now just running the machine.

3. Senior engineers avoid it

This one is brutal but reliable. When you staff the service line, who raises their hand? If it's consistently your juniors and your contractors, and your senior people quietly route around it, that's the market inside your own company telling you the work isn't interesting anymore. Senior engineers vote with their calendars.

4. The client profile has drifted downmarket

Look at the last ten clients who bought this service versus the first ten. If the average deal size is shrinking, the sophistication is dropping, and the procurement process is getting heavier — you're sliding into a segment you probably don't want to serve. Profitable today, race-to-the-bottom tomorrow.

5. It blocks the thing you actually want to build

This is the one that matters most. If you've been trying to move into product work, or a higher-margin specialism, or an AI-adjacent practice — and the legacy line keeps eating the calendar of the exact people you'd need to do it — the service isn't just neutral. It's actively defending the status quo.

The math nobody runs

Most agencies evaluate a service line on contribution margin: revenue minus direct delivery cost. That's necessary but not sufficient. The number you actually want is something like strategic contribution per senior-engineer-hour.

Here's a rough sketch of how we model it:

def strategic_value(service):
    # Direct economics
    gross_margin = service.revenue - service.delivery_cost
    
    # Opportunity cost: what those hours could earn elsewhere
    senior_hours = service.senior_engineer_hours
    next_best_rate = 220  # blended rate on target service line
    opportunity_cost = senior_hours * next_best_rate
    
    # Strategic multiplier: does this work compound?
    # 1.0 = neutral, >1 = builds capability, <1 = drains it
    compounding = service.learning_velocity * service.referral_quality
    
    return (gross_margin - opportunity_cost) * compounding

The numbers in that function are illustrative — your blended rates and multipliers will differ. The point is the shape: a service line can have positive gross margin and negative strategic value at the same time. That's the gap most agencies never measure.

When we ran this on our legacy line, gross margin was solidly positive. Strategic value was negative for four straight quarters before we acted. We had the data. We just didn't want to look at it.

How to sunset without blowing up the business

Killing a service line isn't a single decision. It's a roughly six-month operational project.

Phase 1: Stop selling it (privately)

The first move is internal. Stop putting it on new proposals. Don't announce anything — just quietly take it out of the default deck. You'll be surprised how often a service stays on the menu purely because nobody removed it.

If a client specifically asks, you can still take the work. But you stop volunteering it. This alone usually drops inbound by 40–60% within a quarter.

Phase 2: Raise the price

For the work that still comes in, push the price up 25–40%. Two things happen. Either clients say yes — in which case the strategic-value math gets a lot better and you've bought yourself runway — or they say no, and you've accelerated the wind-down without making the decision feel personal.

This is also how you tell the difference between clients who genuinely value the service and clients who were just buying the cheapest option in the room.

Phase 3: Document and offload

For existing engagements, document everything. Runbooks, decision logs, the weird edge cases that only live in one engineer's head. You want this stuff written down whether you're handing work to a partner agency, training a junior team to maintain it, or just protecting yourself from support requests two years from now.

We've had good results referring tail-end clients to smaller agencies that specialise in exactly the work we're leaving. It's a clean handoff, the client is served, and you sometimes get a referral fee or a reciprocal relationship out of it.

Phase 4: Reallocate the people, not just the hours

This is where most sunsets go wrong. You free up 600 senior-engineer-hours a quarter and then... nothing changes, because nobody owns redirecting them. The capacity gets absorbed by whatever's loudest.

Before you start the sunset, you need a named owner for the freed capacity and a specific destination — a new service line, an internal product, a sales-engineering push, whatever your next bet is. If you can't name it, you're not ready to sunset. You're just creating slack that will get eaten by drift.

The conversation with your team

Engineers who've spent years on a service line will have feelings about it being retired. Be honest. Don't dress it up as a pivot or a strategic refresh. Tell them the service worked, it's been good to the business, and the business now needs their skills pointed somewhere with more upside.

The people who built the playbook are usually the best candidates to design the next one. Frame it that way and most of them will lean in. The ones who don't are telling you something useful about what they actually want to work on, and that's a conversation worth having anyway.

What we'd do

If you suspect you've got a service line that's quietly past its sell-by date, don't start with a strategy offsite. Start with three spreadsheets: margin trajectory by quarter for the last three years, senior-engineer-hours by service line for the last six months, and a list of every internal improvement shipped to each service line in the last year.

If the first one's sliding, the second one's heavy, and the third one's empty — you already know. The rest is just choosing when, and being honest with your team about why. If you want a second pair of eyes on the math or the transition plan, that's the kind of work we do under our consulting practice.