What is the fastest way for an agency to get new clients?

Past clients and referrals. People who have already paid you and people who trust someone who has paid you convert in weeks. Directories, content and cold outbound all work, but each takes months before it produces signed work. Start with the list of everyone you have ever invoiced.

How many clients should a small agency have?

Enough that losing the largest one does not threaten payroll. A common working rule is that no single client should exceed 25 to 30 percent of revenue. An 8-person shop billing $1.2M a year is usually healthier with six to ten active clients than with two large ones.

When should you fire a client?

When the account loses money after you count the unbilled hours, when invoices are repeatedly late after a direct conversation, or when the client is abusive to your team. Raise the problem once, in writing, with a specific fix. If nothing changes, end it under the termination clause and hand over properly.

What should a client onboarding process include?

A signed contract and a paid deposit before any work, a single named decision-maker on the client side, access to every system you need, an agreed communication rhythm, and a kickoff meeting that restates scope. Most of it fits on a two-page checklist.

Clients: how to win them, onboard them, get paid by them, and when to let them go

An agency has two kinds of client problems. The first kind is having too few. The second kind is having the wrong ones, on the wrong terms, and discovering it in month three when the invoice is 40 days late and the scope has doubled. Owners treat these as separate problems. They are the same problem seen from two ends: a business that takes whatever work arrives will also accept whatever terms arrive with it.

I have run a services business long enough to have made every mistake on this page. I have started work without a deposit because the client seemed nice. I have kept a client for two years who cost me a senior developer. I have chased $38,000 for five months with no clause in the contract that let me stop work. Each of those was expensive, and each was preventable with a document or a sentence that takes ten minutes to write.

The five articles in this hub follow a client from first conversation to last invoice.

Winning work is a pipeline problem with long lead times

When work dries up, the instinct is to do something loud: redesign the website, buy a list, post every day for a week. None of that produces a signed contract in the next 60 days. The channels that produce work quickly are the quiet ones: people who already paid you, and people who trust someone who paid you.

The pillar article, how to get clients, lays out seven channels with an honest lead time for each. Referrals and past-client reactivation produce conversations in days and contracts in weeks. Partnerships with adjacent agencies take a quarter. Directories and content take six months or more before they pay for themselves. Cold outbound sits in the middle and works only when each message is about one specific company. The article ends with a week-by-week plan for a 90-day drought, because that is when most owners go looking.

The uncomfortable arithmetic is this. If you need $900,000 a year and your average engagement is $60,000, you need 15 signed projects. If you close one qualified conversation in four, you need 60 qualified conversations a year, which is five a month, every month, including the months when you are too busy to think about sales. Most agencies run zero in the busy months and then wonder why the quarter after a big project is empty.

The first two weeks set the terms for the next two years

A signed contract is the start of the risky period. In the first two weeks the client learns how you work: whether you start without being paid, whether anyone can send you requests, whether “quick question” gets a same-day answer at 9pm. Whatever you tolerate in those two weeks becomes the standing arrangement.

The client onboarding process article is a working checklist with four phases: before signature, the day the contract is signed, week one, and the first 30 days. It covers the deposit, access to accounts, the single decision-maker, and the 30-day check-in that catches a bad fit while it is still cheap to fix. The checklist itself is on the page and available as an editable document. It pairs with the project kickoff agenda in the Scope hub, which covers the meeting itself.

Onboarding is also where you find out whether the paperwork holds. If the client balks at a deposit, pushes back on naming one approver, or wants to start before the statement of work is signed, you have learned something useful for free.

Difficult clients come in five recognizable types

After enough years you stop being surprised. The same five behaviors show up in every portfolio: the scope creeper who adds work one small favor at a time, the ghost who disappears when you need an approval, the micromanager who wants to direct your team hour by hour, the slow payer, and the committee where nobody can say yes.

The article on difficult clients gives each type a cost and a response. The cost matters because owners underestimate it. A client who goes silent for three weeks does more than delay a project. They leave two people half-allocated, and at a loaded cost of $600 a day each, that silence burns $18,000 of capacity. Once you put a number on it, the response stops feeling confrontational and starts feeling like basic management. Most of the responses are contract mechanics: a change process for the creeper (see how to handle scope creep), deemed acceptance for the ghost, a named approver for the committee.

Getting paid is decided in the contract

A late invoice is the most common client problem and the one owners handle worst. They wait too long, write apologetic emails, and keep working the whole time. By day 60 the client owes two invoices, has all the work, and has no reason to hurry.

What to do when a client is not paying sets out an escalation ladder with a day count for each step, from a plain reminder on day 1 past due through suspension of work on day 21 and a formal demand letter around day 45. It includes the wording for each message. It also covers the three clauses that prevent most of the problem: a deposit before work starts, a right to suspend work when an invoice is overdue, and ownership of the deliverables passing to the client only on full payment. That last clause connects to the question of who owns the code, which matters more than most agencies realize.

Ending a client is a normal business decision

Every agency has one client it should have let go a year ago. The account looks fine on revenue and terrible on margin. The team dreads the calls. The owner keeps it because the monthly number is comforting and replacing it feels hard.

How to fire a client covers the decision and the mechanics: how to work out whether the account really loses money, the one conversation you owe them first, the termination clause and notice period, the email wording, and the handover. A clean exit protects you. A client you ended well will sometimes refer you. A client you abandoned mid-sprint will tell people about it for years.

Letting a bad client go is far easier when the pipeline has something in it. That is why this hub starts with getting clients and why the two halves belong together. An owner with five qualified conversations a month can afford standards. An owner with none will sign anyone, skip the deposit, and be back here in six months reading about unpaid invoices.

If you only have time for one article, read the one that matches your current problem. If you have an afternoon, read them in order and compare each stage against how you run it today. The documents referenced throughout are collected on the templates page.