Nepheli Consulting turned one in September. Four of us, a handful of clients, still alive, still profitable. By the standards of first-year consultancies that's a pass. But a retrospective that only lists wins is a marketing page, not a retrospective, so here's the honest version: the things I got wrong, in roughly the order they hurt.
I priced like an employee, not a business
My first day rate was, essentially, my old salary divided by working days, plus a bit for nerves. It felt honest. It was actually a mistake that took six months to unwind.
An employee's day comes with paid leave, sick days, equipment, training, and someone else doing the sales, admin, and accounting. A consultancy's day rate has to fund all of that, plus the unbillable weeks between engagements, plus the company's own future. When I finally did the maths properly, my "honest" rate meant every fully-billed month was quietly subsidising the machinery around it, and there was nothing left for the slow months.
The uncomfortable discovery: raising rates lost us almost nobody. The clients who valued the work didn't blink. The one prospect who pushed back hardest on price turned out, on a later engagement, to also push back hardest on invoices, scope, and everything else. Price sensitivity at the start of a relationship is data. I now treat it as such.
I said yes to wrong-fit clients
In month two, someone offers you money and your pipeline is empty. You say yes. Everyone does. Some of those yeses were fine. Two were not.
One engagement was fundamentally a staffing arrangement dressed up as consulting: they wanted a body in their standups executing tickets, not an outcome. Nothing wrong with staff augmentation as a business, but it's not the business I set out to build, and it consumed our best engineer for months while teaching us nothing and building no reusable expertise.
The other was a client whose real problem was organisational, two teams at war over who owned the platform, and who hired us hoping that a third party saying "use Kubernetes properly" would end the war. It didn't. We delivered good technical work into an environment that couldn't absorb it, and everyone left mildly dissatisfied.
The lesson isn't "be picky", which is easy to say with a full pipeline. It's that wrong-fit engagements have a cost beyond the opportunity cost: they erode the team's morale, they produce no case study you'd want to show anyone, and they train you to tolerate the wrong kind of work. I now ask one question before anything technical: if this project succeeds completely, who inside your organisation is happy about that? If the answer is fuzzy, so is the engagement.
Nobody warned me about the pipeline
Actually, several people warned me. I didn't listen, because in year zero the pipeline fills itself: your network knows you just left a good job, everyone has a project, the first three clients arrive without you doing anything that resembles sales.
Then, around month seven, I looked up from delivery and realised the network-effect well was dry, the current engagements all ended within the same quarter, and I had done precisely zero business development since we started, because we were busy. Which is exactly when you can least afford to start, because consulting sales cycles run one to three months minimum.
We got lucky with timing on a renewal. But the structural fix was accepting that sales is not a phase, it's a permanent background process: a fixed number of hours per week, protected on the calendar like client work, spent on conversations, writing, and follow-ups, especially when we're fully booked. Fully booked is when you sound most attractive and negotiate best. The worst time to look for work is when you need it, and the pipeline you eat in March is the one you planted in December.
Invoicing is not revenue
I knew, intellectually, that invoices get paid late. I did not appreciate what net-30 terms plus a client's own approval workflow plus an August holiday season does to a small company's bank account. We had a stretch where the business was profitable on paper and uncomfortably tight in reality, purely because three invoices sat in three different approval queues.
What changed: deposits for new clients, shorter payment terms as a default rather than an apology, invoicing the moment a milestone lands instead of batching at month-end, and a follow-up cadence that starts politely at day one overdue, not day thirty. None of this is adversarial. Good clients don't mind; their accounts payable departments are simply queues, and squeaky invoices get processed. Also: a cash buffer of a few months' costs is not a nice-to-have. It's what lets you decline the wrong-fit client in the paragraph above.
Staying technical while running a company
This is the fight I expected, and it's been harder than expected anyway. The CEO job, sales, contracts, hiring, invoices, the accountant, the lawyer, expands to fill any space you give it, and it's all urgent in a way that reading Cilium release notes never is. I could feel the technical edge dulling around month eight: reviewing the team's Terraform and realising I was rubber-stamping patterns I hadn't personally used.
What works, partially: I keep myself billable on one engagement at a time, in a real delivery role, not "architect who attends the kickoff". I take the on-call shifts nobody wants. And I kept my certifications current not because clients check, they rarely do, but because studying for a renewal is a forcing function to touch the parts of the stack that daily work doesn't reach. It's an imperfect equilibrium; some weeks the company wins and the code loses. But a DevOps consultancy whose CEO can't debug the cluster is on a clock.
What I'd tell myself on day one
- Your rate is wrong. Add 40%, then read about what utilisation rates real consultancies achieve, then add more.
- The first "no" you give a prospect will feel terrifying and will be the most profitable decision of the quarter.
- Block four hours a week for sales starting now, not when you need it. You will want to skip it every single week. Don't.
- Cash in the bank, not signed contracts, is what lets you sleep.
- Hire slower than feels right, but when someone great appears, move faster than feels comfortable. We got one of each decision right and one wrong, and the difference is stark.
- Write things down publicly. Half our inbound this year traced back to something I published for free.
Year one's report card: survived, learned, kept the team paid and mostly happy. Year two's job is to make fewer of these mistakes and, presumably, discover an entirely new set. I'll report back.