The standard article on this subject lists fifteen tools with a paragraph each. It is not useful, for a structural reason: the answer depends entirely on how many clients you have, and the same list cannot be right for someone with two and someone with twenty.
So start from what the business has to do rather than from what is available.
There are only four jobs
Strip out the craft you are actually paid for and everything left over falls into four categories. Every tool you will ever consider is trying to sell you into one of them.
- Find work. Conversations that have not become projects yet, and remembering to continue them.
- Do the work. Knowing what to do next, and not losing anything a client asked for.
- Get paid. Turning delivered work into an invoice and the invoice into money.
- Keep records. Being able to answer, months later, what happened and what it cost.
The fourth is the one people discover late, usually when an accountant or a client asks a question that requires evidence. It is also the cheapest to fix early and the most expensive to fix retroactively.
What you need, by stage
The stages are defined by client count rather than by revenue, because it is client count that generates administrative load.
| Where you are | What you need | What you do not need yet |
|---|---|---|
| First clients, one or two at a time | Somewhere to write things down, a bank account for the business, invoices that are legally correct | A CRM, a project tool, anything with a per-seat price |
| Three to five clients, work overlapping | Time records, a task list split by client, a document template you reuse | Automations, a client portal, a proposal tool |
| Enough that admin has become a day a week | Invoicing that fills itself from tracked hours, payment reminders, a quote you can send as a document | A dedicated helpdesk, marketing tooling |
| First person joining you | Shared visibility of who has what, and one place holding the client history | Anything you are buying because a bigger agency uses it |
What people buy far too early
Four common purchases that solve a problem you do not have yet.
- A CRM at four clients. A CRM manages a pipeline you cannot hold in your head. With four clients you can, and the tool will mostly be an empty database you feel guilty about.
- Project management for solo sequential work. If you do one thing at a time and hand nothing over, a board is a list with extra steps. Buy it when work starts changing hands.
- Marketing automation before there is anything to automate. A newsletter tool with eleven subscribers is a hobby, and an expensive one relative to its audience.
- A second tool for something the first tool nearly does. This is how a five-app stack starts: each addition is individually reasonable and the total is the problem.
The rule for when to actually pay
One test, and it keeps you honest in both directions: buy a tool when the free version of the process has failed twice, and you can name what each failure cost.
"My spreadsheet is messy" is not a failure. "I invoiced nine hours instead of fourteen because I reconstructed the week from memory, twice" is a failure with a price on it, and it justifies a purchase in about four seconds. The named cost is what stops you buying tools out of aspiration, which is the actual failure mode here — not stinginess.
Two shapes that both work
Once you are past the earliest stage, there are two sensible configurations, and which one fits depends on where your work actually is.
- Specialist tools with clean edges. One good tool per job, chosen deliberately. Works well when one part of your work is deep and specialised, and when you are disciplined about not letting the same fact live in three places.
- One tool covering the admin, specialists for the craft. The admin jobs — tasks, time, invoices, client history — sit together, and your professional tools stay separate. Works well when the paperwork is what is eating your week.
What fails is neither of those: five tools that each hold part of the same client, chosen at five different moments, none of which knows what the others know. That is not a stack, it is a sequence of accidents — and it is worth counting the cost before adding the sixth.
The admin side, for one person
What the four jobs above look like when they sit in one place: tasks, tracked time, quotes and invoices with the client history attached to all of it.