Where your pipeline is actually leaking

You have four proposals out and none of them have come back. The reflex is to send more proposals.

That reflex is usually wrong, and it is the most expensive way to be wrong. Every additional proposal costs the same research, the same call, the same writing and the same three follow-ups as the four already sitting unanswered. If something further down the pipeline is broken, volume doesn't fix it. It multiplies the cost of it.

A freelance pipeline has five joints. Work leaks at each one, and the fix at each one is different. This post is a diagnosis, not a tactics manual: for every stage there's the symptom, the most likely cause, and the single change worth testing. Where the fix needs a whole method, it's linked rather than repeated.

Write down five numbers before you change anything

You can't diagnose a leak you haven't measured. Five numbers, last 90 days, reconstructed from your sent folder in about twenty minutes.

Here is an illustrative set, not observed data, so you can see the shape of a diagnosis:

The instinct on seeing that column is to raise the 32. Look at the last row instead. Two closes and zero repeats means the whole three months produced no compounding, so next quarter starts from the same 32.

Most client-flow tools track stages one to four and stop at the signature, which is where the money question actually starts. That gap is the comparison on the Dubsado alternative page, and it's the reason your fifth number rarely lives anywhere.

Doubling the top of the funnel doubles the price of the leak

Take the illustration above. Six proposals produced two closes. Say each of those six cost you five hours across research, a call, writing and follow-up. That's 30 unbilled hours to win two projects.

Double the conversations and you spend 60 unbilled hours to win four. The conversion problem is unchanged, you've just paid twice as much to have it. Fix the proposal stage first, from 2-in-6 to 3-in-6, and the same 30 hours buys three projects.

This only becomes visible when acquisition hours are attached to the project that resulted. Run a finished project through the effective hourly rate calculator with the hours that won it counted alongside the hours that delivered it, and the number moves more than most people expect.

The five stages, diagnosed

The rest of this post is why each of those is the likely cause, and what to do if it isn't.

Not enough conversations is usually a positioning problem, not an effort problem

Symptom: you send messages, join communities, keep a profile updated, and the volume of actual conversations stays flat.

The default explanation is that you aren't doing enough. Sometimes true. More often the problem is that nobody who reads your description can immediately think of a situation it applies to. "Freelance designer" describes a job. "I fix checkout flows for subscription businesses" describes a problem someone can recognize as theirs.

There is also a real market change underneath this stage, and it is worth separating from your own performance. Fiverr's Q2 2026 results report 2.7 million annual active buyers, down 21.9% year over year, with annual spend per remaining buyer up 15.6% to $368. CEO Micha Kaufman described AI absorbing "high-volume, low-value, transactional tasks." Upwork's Q1 2026 results show the same shape more gently: active clients down 3%, spend per client up 5%.

Fewer buyers, each spending more. If your description sits at the transactional end of a category, conversation volume is genuinely falling and no amount of extra sending fixes that. Upwork's In-Demand Skills 2026 reports every AI-referencing skill up 109% year over year on its marketplace, and its Future Workforce Index 2026 puts freelancers doing AI work at 34% more per hour. Both come from an interested party and the growth list publishes no declines, so read them as direction rather than as a forecast.

The change to test: one sentence, naming a buyer type and an outcome, used identically everywhere for 30 days. If conversation volume doesn't move, the problem is channel choice rather than wording, and the channels are ranked by cost per client won in where freelance work actually comes from in 2026.

A conversation becomes a brief when someone names the problem out loud

Symptom: pleasant calls that end with warmth and no date.

The cause is almost always that the call was about you. You explained your process, your tools, your previous clients. The buyer left with an impression instead of a decision, and impressions decay in about four days.

A brief exists when three things are on the table: a defined problem, a budget signal, and the name of whoever signs. Missing any one of them, you don't have a brief, you have a conversation you'll be reconstructing from memory in a fortnight.

The change to test is small and mechanical. In the last two minutes of every call, say what you think the problem is in one sentence, ask whether that's right, and propose a specific next step with a date attached. "I'll send scope and pricing on Thursday, can you look at it before your Monday meeting" is a next step. "I'll send some thoughts" is not.

If the buyer won't name a decision-maker or an approximate budget, that's information, not rudeness. The questions that predict whether a project will be profitable are set out in five questions that predict whether a project will be profitable, and two of them belong in this call.

Briefs that sit for a week are a qualification problem wearing a scheduling costume

Symptom: the brief arrives, you're pleased, and eight days later you still haven't sent anything.

People read this as procrastination. It's more often reluctance with a good reason behind it. Some part of you knows this brief is vague, or the budget is unstated, or there are four stakeholders, and writing a bespoke document for it feels like a bad trade. So it waits.

Two changes, in this order. First, qualify in ten minutes: if you can't name the decision-maker and the approximate budget, send three questions rather than a proposal. Second, once it is qualified, turn it around inside 48 hours. Speed at this stage is a genuine advantage because most of your competition is also sitting on it for a week.

Turnaround gets faster when the document isn't built from scratch each time. The free proposal generator runs in the browser with no signup and no watermark, which is enough for a two-page scope-and-price document sent the same day.

Proposals that go silent are structured badly more often than priced badly

Symptom: the proposal goes out, nothing comes back, and you assume it was too expensive.

Price is the explanation people reach for because it's the only variable they think about. It's rarely the one that moved. A proposal with a single number gives the buyer a yes/no decision, and yes/no decisions default to "let me think about it" and then to nothing. Two or three priced options change the question from whether to buy to which to buy.

Three other things break at this stage more often than price does:

The proposal has no expiry, so there's no reason to decide this week rather than next quarter. A validity date is not a pressure tactic, it's the reason a document gets opened. The free quote generator has a valid-until field built into it for exactly this.

The follow-up isn't scheduled. Most freelancers follow up once, apologetically, and stop. Book two follow-ups in your calendar at the moment you hit send.

And you can't tell whether it was read. A proposal that was never opened is a different problem from one that was opened four times and not answered, and treating them the same wastes the follow-up. Worklyn's proposals show open activity and convert an accepted document straight into a contract or invoice, which removes the days between "yes" and the first payable document.

What actually goes in the document, and what to leave out of it, is covered in the proposal document: what to include and what to leave out. If the silence is coming from the buyer's own risk rather than your price, what the person hiring you is actually worried about is the better diagnosis.

Closes that don't repeat are the most expensive leak on the list

Symptom: the work goes well, the invoice gets paid, and six months later you've heard nothing.

This is last because it's least visible and it costs the most. Look at the illustrative numbers again. Winning that second project took roughly five unbilled hours through the normal pipeline. Winning it from a client who already paid you takes one message and a scoping call, because pricing is settled, vendor onboarding is done, and the invoice already goes to the right person in accounts payable.

The cause is not that clients were unhappy. It's that nothing triggers the contact. Prospecting has a trigger, which is an empty calendar. Repeat business has none, so it happens by accident or not at all.

The change to test is one calendar event: on handover day, book a 30-day check that asks whether the thing you built did what it was supposed to do. That question is about their business, and it surfaces the next piece of work more reliably than any pitch. The full cadence, and the messages that go with each touch, are in more work from the clients you already have.

Thirty minutes now, and the one experiment to run this month

In the next thirty minutes: reconstruct your own five numbers from the last 90 days. Conversations, briefs, proposals, closes, repeats. Then find the biggest percentage drop between two adjacent rows. That's your leak, and it's usually not the first row.

This month: change one thing at that stage and leave everything else alone. One variable, 30 days, then compare the same five numbers. Changing three things at once is how you end up with a better quarter and no idea which change caused it.

Worklyn's CFO Mode answers questions from your own records, including what each client has actually paid you and which of them ever came back, so the fifth number stops being the one nobody tracks.