)
)
Key takeaways
The Core Five requirements separate real PSA software from a glorified task tracker: Resourcing and forecasting, time tracking, budget and profitability tracking, client reporting, and AI agents that plan and do costed work.
"Nice to have" and "must-have" aren't the same list for every team: The framework below sorts core capabilities from genuine extras, so you're not paying for the wrong thing.
Integrations are a requirement, not a bonus feature: Software that doesn't talk to your CRM or accounting platform just relocates the manual reconciliation work.
Billing accuracy is the feature that pays for the rest: Unifying time, budgets, and invoicing in one platform closes the gap between hours worked and dollars collected.
AI is now a requirement to evaluate, not a demo perk: An AI feature that isn't working from your own cost and delivery data can't actually answer what a deliverable cost to produce.
"Are we ready to invoice this account yet?" It's the 25th of the month, and the honest answer is nobody's fully sure. The timesheet export needs cleaning up. The budget spreadsheet hasn't been touched since the kickoff call. Someone has to cross-reference both against the scope document before anyone can send a number to the client with a straight face.
I ran that ritual more times than I'd like to admit. Not because anyone on the team was careless — because the tools we had were never built to answer that question cleanly. A timer plus three spreadsheets can track effort. It can't tell you, on demand, whether the effort matches the money.
That gap is exactly what a real PSA platform is supposed to close. But "PSA software" has become a loose enough label that plenty of tools claim the category without covering the ground that actually matters. This chapter is the checklist I wish I'd had: what to expect, what to demand, and what quietly breaks when a feature is missing.
Requirements read differently depending on whose signature is on the purchase order:
What's in this guide
This is the fourth chapter of Teamwork.com's PSA guide. The rest of it:
Core PSA features & requirements — you're here
Professional services project management software: what agencies actually need
What "core" actually means here
Professional services automation requirements aren't a wish list. They map to a small number of decisions every services business has to make, over and over: who works on what, whether the account is still profitable, and what the client gets told about it. A professional services automation platform earns the name by answering all three from the same data. It doesn't earn it by bolting a timer onto a kanban board and calling it done.
Internally, the way we think about it at Teamwork.com is that a feature only counts as "core" if removing it would force someone back into a spreadsheet. That's a deliberately low bar and a genuinely useful one: it's the same test I'd have applied on the agency side, if I'd known to ask it.
Six categories pass that test — call it the Core Six requirements. Integrations are less a sixth category than a precondition for the other five working at all.
The Core Five requirements
1. Resourcing & forecasting
Good resourcing shows you capacity and workload for every person, on every account, far enough in advance to actually redistribute work — not just a snapshot of who's busy today. It should flag when a new project is going to collide with an existing commitment before you staff it, not after someone's already double-booked.
Without it, the same two or three people absorb every urgent request, because they're the ones a manager can picture clearly enough to trust. The rest of the team's real capacity stays invisible. That's not a motivation problem. It's a visibility problem wearing a motivation costume.
Teamwork.com's Resource Scheduler shows capacity across every project someone's staffed on, not just the one you happen to be looking at, so the double-booking shows up before you commit to it, not during the kickoff call. Teamwork.com's AI Smart Scheduler goes a step further and suggests who to staff based on availability and role, instead of leaving that judgment call to whoever remembers who's free.
)
2. Time tracking
Time tracking sounds like the simplest requirement on this list, and it's the one most tools get partially right and functionally wrong. The bar isn't "can people log hours." It's whether that time attaches cleanly to a project, a task, and a billable rate, in a way that flows straight into a budget and an invoice without anyone re-entering it somewhere else.
Without that connection, time tracking becomes a compliance exercise — numbers logged because someone's asked to log them, disconnected from whether the account is actually on budget. I've sat in enough end-of-month reconciliations to know that gap is exactly where unbilled hours quietly disappear.
Data point: Industry estimates suggest poor time tracking costs professional services workers somewhere around $50,000 a year in unrecorded, unbilled revenue. That's not one dramatic missed invoice — it's a few untracked minutes here, a call that never got logged there, compounding over 12 months.
3. Budgeting and profitability tracking
This is the category that turns "the team is busy" into "the account is profitable," and it's the one most spreadsheet setups never really solve, because it requires timesheets, budgets, and billing to already agree with each other. Good budget tracking sets a billable rate and a cost rate per person, tracks spend against the budget as work happens, and surfaces overspend while there's still time to do something about it.
Without it, you find out an account went over budget the same way I always did, at month-end, from a spreadsheet, after the work that caused it was already delivered for free.
Worked example. A six-person team on a $20,000-a-month retainer overservices by just 10% — a handful of extra rounds nobody logged as change requests. That's roughly six weeks of unpaid work a year, or about $2,000 a month handed back to a client who never asked for a discount. Multiply that across even four similar accounts and it's $96,000 of margin gone before anyone notices it's missing.
Teamwork.com's Budget Tracking checks spend against billable and cost rates in real time, so that gap shows up as an alert, not a surprise. Teamwork.com's Profitability Reports then roll it up account by account, so the conversation with leadership is a number, not a guess.
)
This is also where invoicing accuracy lives or dies; it's worth a real example, because the gap between "roughly accurate" and "actually accurate" is bigger money than most finance teams realize — worth coming back to once all four requirements are on the table.
4. Client reporting and collaboration
A client shouldn't have to ask for a status update, and your team shouldn't have to build one from scratch every time they do. Good reporting pulls budget usage, task progress, and timelines into a view you can share directly, on a schedule, without a manual assembly step the night before the call.
Without it, reporting becomes invisible unbilled labor — real hours spent producing something that generates zero revenue and satisfies nobody, because it's inconsistent from client to client and from month to month.
Data point: Manual reporting reportedly runs about an hour per client, per week. Across 15 active clients, that's roughly 60 hours a month — more than a third of a full-time role, doing work that never appears on an invoice. Databox's research on agency reporting is worth a look if the number sounds too high to be real.
5. AI agents
This is the newest requirement on the list, and the one most feature checklists haven't caught up to yet. The bar isn't "does it have an AI chatbot." It's whether an AI agent can actually plan and do a defined piece of work — drafting a status report, synthesizing findings, pricing a proposal — using your own project and cost data, and show up afterward as a costed, supervised line item with an owner, not an invisible feature nobody's tracking.
Without that connection, "AI features" become a demo trick: impressive on a sales call, disconnected from your actual budgets and rates the moment you're back in your own account data. A generic assistant can draft a paragraph. It can't tell you what a deliverable actually cost to produce, because nobody ever gave it your rates or your timesheets to work from.
What actually passes the test: an agent that can draft status reports, synthesize findings, and price proposals from your own actuals — costed and supervised like a member of the team, not a black box bolted on top of the platform. That's the difference between an agentic PSA and project management software with a chatbot added on: one prices what it does, the other doesn't know how.
Integrations: the precondition, not the extra
None of the above works in isolation. Resourcing needs to know what's on the calendar. Budgets need to know what's been invoiced. Time tracking needs to reach the accounting system without a CSV export in between. A professional services automation platform that doesn't connect to the CRM, accounting software, and calendar tools you already run just moves the manual reconciliation work from one spreadsheet to two systems that don't talk to each other; in practice, that's worse.
We asked over 1,000 senior leaders at client service firms which capabilities their current tech actually fails at for our Sprint to AI research. Resource management came up for 42%, and connecting third-party tools for 40% — the two requirement categories on this list that are easiest to underrate until you're the one stuck exporting a CSV at 6pm.
Must-have vs. nice-to-have
Not every capability belongs on the non-negotiable list, and treating them as equally important is how procurement decisions get bogged down comparing features that don't matter to the ones that do.
The first six rows are the requirements. The last two are genuinely optional, and it's worth being honest with a vendor's sales team about which list you're evaluating against — a feature demo can make a nice-to-have look load-bearing if you let it.
Self-audit checklist
Before you sign anything, ask whether your shortlisted tool can answer these five questions without exporting to a spreadsheet first:
Can I see who's overbooked next month, not just this week?
Can I trace a logged hour to a specific budget line and rate?
Would an overspent account show up as an alert, or would I find out at invoicing?
Could I hand a client a live report link right now, without building anything first?
Could an AI agent draft or price something using our own numbers, or is it guessing?
If the honest answer to any of those is "sort of, with a workaround," that's not a professional services automation platform. That's a task tracker with a time field bolted on.
That checklist isn't really about procurement comfort. It's the same five questions that show up later as a line in a board deck — whether margin held, whether an account renewed, whether headcount had to grow to cover a process a system should have carried. A vendor who can't answer them isn't failing a demo. They're asking you to sign off on carrying that risk manually, indefinitely.
What this looks like inside Teamwork.com
Run the five self-audit questions above against Teamwork.com specifically, and each one maps to a real, connected part of the platform rather than five separate tools stitched together after the sale:
"Can I see who's overbooked next month, not just this week?" Teamwork.com's Resource Scheduler and AI Smart Scheduler show capacity across every project someone's staffed on, weeks out, not a same-day snapshot.
"Can I trace a logged hour to a specific budget line and rate?" Time tracking flows straight into Budget Tracking, which checks spend against billable and cost rates as the work happens.
"Would an overspent account show up as an alert, or would I find out at invoicing?" Profitability Reports roll budget status up account by account, so the answer is a live number rather than a month-end reconstruction.
"Could I hand a client a live report link right now, without building anything first?" Client-facing reporting pulls from the same task, time, and budget data — nothing gets rebuilt from a spreadsheet the night before.
"Could an AI agent draft or price something using our own numbers, or is it guessing?" Teamwork.com's AI Teammates work from that same connected data, so a drafted report or a priced proposal reflects what work actually costs, not a generic estimate.
)
Invoicing accuracy is where all four requirements show up in one number. Teamwork.com's budgeting and profitability tools unify time tracking, budgets, and billing in one place — the same unification behind SugarCRM's result below.
SugarCRM's invoicing accuracy. SugarCRM, an enterprise SaaS company, unified projects, time tracking, and billing on a single platform and got invoicing accuracy close to perfect — less than $20,000 credited back against more than $10 million in annual invoicing. That's a credit rate under 0.2%, on a scale where even a small percentage of billing errors translates into real client-trust damage and real revenue leakage. That result isn't a time-tracking win or a billing win in isolation — it's what happens when resourcing, time, budgets, and reporting are already the same system instead of separate requirements evaluated separately and bought as four different tools. Read the full SugarCRM customer story for how that unification worked in practice.
Where this goes next
None of these five categories work well in isolation, which is really the whole argument for treating PSA software as its own category rather than a set of features to bolt onto a project tool. Chapter 2 goes deeper on that distinction if you haven't read it yet. The next question — once you know what to require — is how to actually compare vendors against this list without getting talked out of it by a good demo. That's the Fit Framework, covered next in Chapter 5: How to Choose the Best PSA Software.
)
)
)
)
)
)
)