The first 90 days after you leave: what has to be running
An operational plan for days 30, 60 and 90 of full-time freelancing, built around the reason cash gaps appear in your busiest month rather than your slowest.
Almost everyone leaving employment prepares for the wrong risk. The fear is an empty pipeline, so the preparation is a client list and a savings buffer. The failure that actually shows up in the first quarter is different: the work arrived, you did it, and the money hadn't. IPSE's research on the UK self-employed found 35% were not paid on time by a client in the last 12 months, with an average amount owed of £5,230. The page does not state its sample size or field dates, so treat it as a description of prevalence rather than a precise estimate. Even read loosely, it points at the right thing: late is more common than absent.
There is a mechanism underneath that, and it is the one worth building your first 90 days around: the month you are busiest is the month you invoice latest. Delivery crowds out billing, billing is the only step in the chain you control, and the delay lands exactly when your costs are highest. Everything below is arranged so that the two steps you control run on dates rather than on how the month felt.
The busiest month is the latest-invoiced month
Time from work done to cash received has two halves. The second half belongs to the client and is bounded by law: under the EU late payment directive the default B2B period is 30 calendar days from receipt of the invoice, with 60 days as the outer limit unless expressly agreed and not grossly unfair. The first half belongs entirely to you, has no legal floor, and is invisible.
Work the illustration. Two deliveries in June, on the 12th and the 26th, worth €14,000 together. June was your best month, so the invoices go out on 8 July. Thirty days from receipt puts payment due around 7 August: 56 days after the first piece of work and 42 after the second. Invoice on delivery instead and the same €14,000 on the same terms is due on 12 July and 25 July. Identical revenue on identical terms, roughly four weeks earlier, and those four weeks are the ones in which you were paying for a busy month.
The size of the effect is easy to quantify for yourself. At €80,000 of annual revenue, every day of average lag between finishing work and sending the invoice is €80,000 ÷ 365 = €219 sitting in transit permanently. Cutting a median lag of 18 days to 3 releases 15 × €219 = €3,285 of cash, once.
Notice what that is and isn't. Raising your rate 10% on €80,000 is €8,000 a year and it recurs. Fixing invoice lag is a one-time level shift in working capital, not a flow. The rate rise is worth more over any horizon longer than a few months. The lag fix arrives immediately, costs nothing, needs no client's agreement, and is the difference between a late payer being a nuisance and being a crisis. Do the cheap one first and then do the valuable one. The mechanics of where the days go are laid out in more detail in time-to-cash: where freelancers actually lose money.
What has to be running, and by when
By | What must be running | How you know it is | What its absence costs |
|---|---|---|---|
Day 30 | Hours recorded the same day, in one place that invoices read from | You can produce billable hours by project without reconstructing a week from memory | Reconstructed weeks lose hours, and they lose most in the busiest month |
Day 30 | Two fixed invoicing dates a month, independent of workload | Both dates are in the calendar with nothing else on them | Revenue ÷ 365 per day of lag, compounding across every project |
Day 30 | Tax reserve moved the day money lands, not at month end | Reserve balance is at least your set-aside rate × cash received to date | A reserve spent by accident is a liability you meet from next quarter's revenue |
Day 60 | Every open invoice carrying a due date and the applicable statutory interest | You could state the due date of your oldest invoice from memory | UK statutory interest is 11.75% for the second half of 2026; EU minimum is 10.40% — unclaimed by default |
Day 60 | A chase sequence triggered by dates rather than by nerve | Nothing is more than seven days overdue without a message already sent | A 2022 survey of New York freelancers found 76% spent 1–2 hours a week chasing |
Day 60 | One weekly slot for pipeline that survives a full week of delivery | The slot happened in the busiest week of the last month | Sales lag: the selling you skip in month two is the empty month four |
Day 90 | A rolling forward view of cash at least 13 weeks out | You can name the lowest cash point in the next quarter | A bank balance says nothing about a quarter containing two late payers |
Day 90 | Effective hourly rate computed at each project close | Your last three projects each have a number | Repricing on impression instead of evidence |
Day 90 | Reserve expressed in months of fixed costs, not in currency | One number you can say out loud | The same late payment is a nuisance at six months and an emergency at one |
Day 30: the two systems that must not depend on your mood
Recording hours and sending invoices are the only two links in the chain that are entirely yours, and both degrade under load. That is why they get built first, before the systems that feel more urgent.
Hours first, because they are the input to everything else. The requirement is not a timer; it is that hours reach invoices without a copying step. A timer that produces a report you then retype into an invoice has two failure points and both of them fail in a busy week. This is the actual difference between time-tracking tools and workspaces that bill: what a Harvest-style tracker does after the timer stops is the question to ask before you pick one, and the answer determines whether your billable share is measured or estimated.
Invoicing second, and the key word is fixed. Not "invoice promptly," which is a resolution. Two dates a month, say the 1st and the 15th, on which every deliverable completed since the last date is billed, whether the project is finished or not, whether it feels premature or not. A fixed date removes the judgment call, and the judgment call is what the busy month eats.
The tax reserve is the third day-30 item and it belongs here because the first deadline arrives sooner than people expect. In the US, estimated tax is due 15 September 2026 and 15 January 2027 for someone who left employment this autumn, and the safe harbour is 90% of this year's tax or 100% of last year's, rising to 110% if last year's AGI was above $150,000. In the UK, Making Tax Digital for Income Tax went live on 6 April 2026 for anyone with qualifying income above £50,000, with cumulative quarterly updates due 7 August, 7 November, 7 February and 7 May. Two details catch people: qualifying income is gross, before expenses, and HMRC tests it from the prior year's Self Assessment return — so in your first year out you are probably not yet in scope, and you will be, on a threshold that drops to £30,000 for 2027 and £20,000 for 2028. Build the record now on the assumption you will be filing quarterly, because retrofitting a year of receipts is worse than the reserve itself. A monthly set-aside calculator with US, UK and German presets will give you the percentage to move; the discipline is moving it on receipt rather than on the last day of the month.
Day 60: terms that carry a number, and an hour you protect
By day 60 you have invoices going out reliably and you will have discovered that some of them are not being paid. Two things change.
The first is that your invoices start carrying the entitlement you already have. For UK B2B debts, statutory interest is Bank of England base rate plus 8%, and with Bank Rate at 3.75% the rate fixed for 1 July to 31 December 2026 is 11.75%, plus fixed compensation of £40, £70 or £100 depending on the size of the debt. In the EU, interest runs automatically at at least eight percentage points above the ECB reference rate, which with the main refinancing rate at 2.40% means a minimum of 10.40% for the second half of 2026, plus €40 of fixed recovery compensation per late invoice — and several member states set a higher margin, so check yours. You will rarely charge it. Stating it changes the conversation anyway, and it costs one line.
Choosing what your default terms should be in the first place is a separate question, and one that is expensive to revisit once clients are used to an answer. It sits with the other month-one commitments in the first 30 days: six decisions that are expensive to reverse.
The second change is that chasing becomes scheduled work rather than an emotional event. A 2022 survey of New York freelancers, run with several creative guilds, found that 76% spent one to two hours a week chasing payment. That survey covers one state, one year, and does not disclose its sample size, so read it as a signal about how much time this consumes rather than a national statistic. An hour a week spent deciding whether today is the day to send a reminder is an hour spent on a decision that should have been made once, in advance, as a schedule.
The third day-60 item is the weekly pipeline slot, and it is the one that gets sacrificed. The reason to protect it is arithmetic, not discipline. If your average time from first conversation to signed work is six weeks, then a month in which you sell nothing is not a quiet month — it is a quiet month six to ten weeks later, when you have committed to costs on the assumption of continuity. The slot has to survive the busiest week of the month or it isn't a system.
Day 90: a forward view instead of a balance
By the end of the first quarter you have enough history to stop navigating by bank balance. A balance is a fact about the past. What you need is the lowest point in the next thirteen weeks, which depends on invoices already sent, work already committed, and fixed costs already known.
Two numbers make that view honest. The first is your reserve expressed in months of fixed costs rather than in currency, because €14,000 means nothing until it is divided by €2,800 a month to give five months. A runway calculator that takes cash on hand, average monthly costs and expected income will do the division and, more usefully, show how much a single 45-day-late invoice moves it. The second is the effective hourly rate on each project at close, computed as fee divided by every hour the project consumed including scoping, revisions and chasing. Three projects with three numbers tell you which kind of work to sell more of. Zero projects with numbers tell you to raise your rate across the board, which is the wrong instrument.
Everything in this post is operational, and none of it fixes a pricing problem: if your effective rate at project close is below target on every project, no invoicing cadence will repair that. The full sequence of financial milestones, from first paying client to the month you can pay yourself consistently, is set out in the complete guide to going freelance organized by money milestones.
The rule for when to accept the next project
Measure one thing for the whole of your first quarter: the median number of days between finishing a piece of work and sending the invoice for it.
If that median is above seven days, the next project does not improve your position. It adds revenue you will bill late, at the moment your capacity to bill is already the binding constraint, and it widens the gap between the month you earn and the month you get paid. Fix the median first. It takes a calendar entry and about an hour.
If the median is at or under seven days and it held through your busiest month, you have the system that lets more work turn into more cash rather than more exposure. Then take the project, and reserve the rate conversation for the one after that.
Worklyn generates invoices directly from logged hours and completed deliverables, so the invoice runs off the work rather than off a separate act of remembering, which is the day-30 list in one line.