compare
An agency with nine people is running work for twelve clients. Four are retainers, six are fixed projects, two are signed and dormant. Someone suggests ClickUp, because it does everything, and that is true. A week later the same person has watched three tutorials that each recommend a different structure, found a consultancy that charges for setting it up, and is no longer sure whether the tool is the answer or the next project.
The existence of a paid industry around structuring this particular tool is the most useful thing to know before adopting it. It is not a criticism. It is a description of what flexibility costs. A tool that will accept any hierarchy requires the agency to choose one, write it down, and defend it against twelve clients who all want something slightly different. What follows is which of those decisions matter, and which capabilities an agency needs that turn out to sit on specific paid tiers.
Most project management advice assumes one team working on one product with a shared backlog. Agency work breaks that assumption in five ways, and every structural argument comes back to them.
People are shared and projects are not. A designer touches five clients in a week. Any structure organised purely by client hides the only number that matters for planning, which is how much of that designer's week is already committed.
Confidentiality runs between projects, not around the company. Client A must not see client B, and often must not know client B exists. This is the inverse of a product team, where everything internal is visible internally.
A large share of elapsed time is waiting. Feedback rounds, approvals, assets that have not arrived. Work blocked by a client is not work in progress, and a board that treats them identically misrepresents both the agency's speed and the client's.
Work ends, and then arrives again. Projects finish, clients leave, retainers repeat the same six tasks every month. One structure has to hold both without becoming a graveyard.
Hours are money. Not as a reporting nicety. As the thing that determines whether a project made a profit, and whether a retainer is being quietly overspent.
Any tool can be made to handle the first four with enough configuration. The fifth is where plan boundaries start to matter.
ClickUp's structure runs Space, then Folder, then List, then Task, with subtasks below that. Four container levels is generous, and it is exactly why guides disagree: there are several defensible ways to map an agency onto them.
The two common patterns are a Space per client, and a Space per function with a Folder per client. The published agency guides tend toward a small fixed number of Spaces, typically one for each stage of the agency's own operation, with clients living a level down. Either can work. What matters is picking one and never mixing them, because a half migrated hierarchy is worse than either pure form.
Three practical constraints should inform the choice.
Spaces are capped on the free plan at five. An agency with twelve clients cannot use a Space per client on Free Forever, so the choice is partly made by the plan. Lists per Space are limited to 40 on Free, 200 on Unlimited, 400 on Business and 1,000 on Enterprise.
Permissions attach to containers, so the container choice is the access model. Deciding where a client sits in the hierarchy is deciding what a client can be shown. Working that out after the fact means moving hundreds of tasks.
Onboarding cost scales with depth. Every new hire, and every freelancer brought in for three weeks, has to learn where things go. A structure that takes a paragraph to explain will be followed. One that takes a page will be approximated, and approximation is how a shared structure dies.
The recommendation that survives contact with reality is the shallowest structure that supports the access model. Depth is available and it is not free.
The hierarchy also has to answer what happens to work that repeats and work that has stopped, and this is where most agency setups accumulate rubbish.
Retainers should generate fresh tasks each cycle from a template rather than having the same task reopened every month. Reopening produces one record with a long history and no countable output. Creating fresh tasks produces a monthly list of what was delivered, which is the evidence that makes a renewal conversation short.
Dormant and finished clients need a stated destination. Archiving whole containers keeps searchable history without occupying the space that active clients need, and it should happen on a schedule rather than when someone notices the sidebar has become unreadable. An agency that never archives ends up scrolling past six former clients to reach today's work, which is a small tax paid several hundred times a week.
This is the detail that changes the arithmetic for agencies, and it is easy to miss when comparing feature lists.
On Free Forever, guests must be given full access, with no permission control. Guest visibility settings are not available on that plan either. For an agency, that means the free plan cannot safely be used for anything a client is invited into, because the only options are full access or no access. Guests with permission control begin on Unlimited.
Prices as published on ClickUp's pricing page, per user per month:
| Plan | Billed yearly | Billed monthly | Guests |
|---|---|---|---|
| Free Forever | $0 | $0 | Full access only, no permission control |
| Unlimited | $7 | $10 | Guests with permission control |
| Business | $12 | $19 | Guests with permission control |
| Enterprise | Contact sales | Contact sales | Plus custom permissions and control over who may add guests |
Two further items belong in this section because they decide how client facing an agency can be. Removing ClickUp's branding from outgoing email starts on Business. White labelling is Enterprise only. An agency that intends clients to experience the tool as part of its own service should price that expectation before designing around it.
The second place where agency needs and plan boundaries meet is time. Here the tiers are specific.
| Capability | Free Forever | Unlimited | Business | Enterprise |
|---|---|---|---|---|
| Native time tracking | Trial | Included | Included | Included |
| Timesheets | Trial | Trial | Included | Included |
| Timesheet approvals | Not included | Not included | Not included | Included |
| Dashboards | Trial | Trial | Included | Included |
| Custom exporting | Not included | Trial | Included | Included |
| Billable reports | Not included | Not included | Included | Included |
The practical effect is that an agency billing hours lands on Business. Timesheets, dashboards, billable reports and custom exporting are all on that tier, and those four are the reporting layer an agency actually runs on. At $12 per user per month billed yearly, a nine person agency is at roughly $108 a month, or $171 at the monthly rate. Formal timesheet approval, the step where a manager signs off hours before they become an invoice, is Enterprise.
Other plan boundaries worth knowing before designing a process around them: file storage is 60MB on Free and unlimited on every paid plan, chat message history is 30 days on Free and unlimited above it, and automation allowances run from 5 active rules with 100 executions on Free, to 500 rules and 1,000 executions on Unlimited, to unlimited rules with 5,000 monthly executions on Business. Teams, the grouping used for sharing, are capped at 2 on Free, 3 on Unlimited and 10 on Business.
None of this makes the tool a poor fit. Unlimited storage and unlimited members on the free plan are both unusually generous. The point is that the agency specific capabilities cluster on one tier, so the honest comparison is Business against alternatives, not Free against alternatives.
Configuration effort is a real cost that does not appear on a pricing page, and it recurs.
Setting up a hierarchy for twelve clients, with statuses, custom fields, templates for recurring retainer work and automations, is a project. Done carefully it takes days, and it is done by whoever in the agency is best at the tool, which is usually someone whose time is billable. Consultancies exist to absorb that work, which is evidence of its size rather than of any defect.
The effort then recurs in three ways. Every new client needs the structure applied, which is fine if it is templated and expensive if it is not. Every new person needs it explained. And every quarter the structure drifts, because a client asked for a status that does not exist elsewhere and someone added it to one List only.
The countermeasure is written conventions rather than more configuration. The stage names, the card naming pattern, where a client sits in the hierarchy, and what a custom field means should fit on one page that a freelancer can read in five minutes. Agencies that keep that page current get years out of any tool. Agencies that do not end up with twelve bespoke setups and a person who is the only one who understands them.
This is also the right frame for comparing tools. The question is not which has more features. It is how much of the agency's structure has to be invented versus assumed. A project management tool where the board types, the timeline, the per person view and chat are all present by default invents less, and gives less room for drift, at the cost of accepting someone else's opinions. Running the comparisons side by side is a faster way to see which tools divide access below the workspace level and which include hours, because those two properties decide most agency cases.
Five questions, answerable in an afternoon, that predict whether a tool will hold.
Can a client be shown one project and nothing else, on the plan the agency will actually pay for? Test it with a real second account, not by reading the feature matrix.
Can every person's committed work across all clients be seen on one screen? This requires one owner per task and a date on anything in progress. Without it, capacity planning is guesswork regardless of tool.
Is work waiting on a client a distinct state? If it is not, the board overstates work in progress and understates client delay, and both errors favour blaming the agency.
Are hours recorded against the same records as the work, on that plan? Hours in a separate system mean maintaining the client list twice, and the two lists always diverge.
Can the whole structure be explained on one page? If not, freelancers will not follow it, and the structure is already theoretical.
Write the hierarchy down on one page before configuring anything, decide whether clients live in Spaces or a level below, and test client access with a real guest account on the plan the agency will pay for rather than on a trial. Then check which tier holds timesheets, dashboards and billable reports, because for an agency billing hours that tier is the real price. If the structure is clear and what remains is cost per person for client access and hours, Pinateca is free for up to five people and ten boards, with timesheets available as an add-on rather than as a tier.
Either works, but not both. A Space per client gives the cleanest permission boundary and is capped at five Spaces on the free plan, so it usually requires a paid plan for a real client list. A Folder per client inside a small number of functional Spaces scales further and keeps cross client views simpler. Pick one, write it down, and migrate fully rather than partly.
No. On Free Forever, guests must be given full access with no permission control, and guest visibility settings are not available. Guests with permission control start on Unlimited at $7 per user per month billed yearly. Any agency intending to invite clients should treat a paid plan as the starting point.
Business, at $12 per user per month billed yearly or $19 monthly. Native time tracking is included from Unlimited, but timesheets, dashboards, billable reports and custom exporting are all on Business, and those four are the reporting layer that turns tracked hours into invoices and margin. Formal timesheet approval is Enterprise.
Because the tool accepts every structure and therefore endorses none. Four container levels produce many defensible mappings, and each guide is written from the shape of one kind of agency. That is why the deciding factor should be the access model and the length of the explanation rather than elegance: the shallowest structure that supports client permissions is the one a freelancer will still follow in month six.
Leaving out a distinct state for work sitting with the client. Without it, time the agency did not spend appears as agency delay, the board shows far more work in progress than is real, and nobody can say which clients are holding up their own projects. It costs one status to fix and it changes every conversation about schedule.