What I Learned Refunding a Webflow Client for the First Time in 2026

What Does It Actually Feel Like to Refund a Webflow Client?

On April 26, 2026, I sent a Razorpay payout to a client for the first time in five years of running my Webflow practice from Bengaluru. The refund was for one hundred and seventy thousand rupees, which represents a month and a half of work on a B2B SaaS marketing site that, by week four, we both knew was not going to work out. Sending that payout took me forty minutes longer than it should have, because I sat staring at the confirmation screen, trying to decide if I was making a mistake.

Five years is a long stretch to go without ever refunding anyone. According to Freelancer Map's 2025 European freelance survey, 38 percent of freelancers refund at least one client in their first three years, and the median refund happens around month fourteen. I held out longer than most, mostly because I have always considered refunds a sign of failure. After this experience I think I had that wrong.

I want to walk through what led to the refund, what it cost me beyond the money, what it taught me about how I scope and run client engagements, and the policy change I made in May to never end up in the same spot again.

How Did the Engagement Go Wrong in the First Place?

The client was a fintech founder in Singapore who reached out in early March looking for a homepage rebuild and a small CMS-driven blog launch. We had a clean discovery call, I sent a fixed-fee proposal of three hundred and forty thousand rupees, and she paid the first half on signature within forty-eight hours. The brief said it would be a four week build. By week three I knew it would be eight.

The drift was on both sides. She added two new sections after week one, then changed the brand colors after week two, then asked for an integration with a custom analytics tool I had never used. I said yes to all three because the relationship felt good and the work itself was interesting. By the end of week three, my scope had expanded by 60 percent and my margin had collapsed.

By week four, every Webflow Designer session felt heavy. I dreaded opening the project. I missed two small deadlines for the first time in years. According to Asana's 2025 work patterns report, freelancers who report dread on a specific project complete that project late 73 percent of the time. The data lined up with my own experience exactly.

Why Did You Choose to Refund Instead of Push Through?

Three reasons. First, the relationship was clearly going to end badly even if I pushed through. The new scope I had agreed to was outside my comfort zone, and the final deliverable was going to disappoint both of us. Second, I had four other client commitments that were being neglected because of this one. Pushing through would have damaged three other relationships to save one. Third, I had stopped doing my best work on the project, and shipping subpar work is its own kind of damage to my practice's reputation.

The math felt heavier than it actually was. The refund covered the first payment minus the value of the work I had actually delivered that the client could use independently. That came out to one seventy thousand rupees. Painful in the short term, but recoverable across two weeks of healthy client work, which is how it played out.

The framing that helped me was reading the late Tony Hsieh's section in Delivering Happiness on refund-as-marketing. He argued that a clean refund preserves the option to do business again later, where a dragged-out failure forecloses it. That framing turned out to be exactly right in my case.

What Was the Hardest Part of Sending the Refund?

The conversation that preceded it. I called the client on a Tuesday morning. I told her plainly that the project had drifted past what either of us scoped, that I was the wrong person to finish it given the new requirements, and that I would refund the difference between her first payment and the deliverables she could use independently. She was surprised but not upset. The whole call took fourteen minutes.

What was harder than the call was the two days before it, where I rehearsed every possible reaction. I worried she would write a public review of my practice. I worried about referral damage. I worried about whether the refund itself would look like an admission of malpractice. None of those concerns materialized. She asked thoughtful questions, accepted the refund without negotiation, and wished me well. The catastrophe I had imagined did not exist.

The other hard part was telling my partner that night that I had effectively wiped out a month of profit. She was supportive, but I felt embarrassed. According to a Harvard Business Review piece from 2024 on freelancer financial decisions, that specific kind of shame is the most common reason refunds get delayed past the point where they would have been clean.

What Did the Refund Cost Me Beyond the Money?

About a week of mental energy, mostly in the two days before the call and the three days afterward. I had to rebuild my forecast for May, which had assumed the second payment would land. I had to update my CRM, my accountant's tracking sheet, and my GST records to reflect the reversed transaction. None of that was hard, just tedious.

The bigger cost was creative. I struggled to start new work for about ten days afterward. Every fresh client brief felt like it might turn into the same kind of drift. I overcorrected on scoping with the next two clients I onboarded, which produced a different kind of friction. It took until mid-May for my scoping rhythm to feel normal again.

The book that helped most during that stretch was Paul Jarvis's Company of One, which I had read before but reread that week. The chapter on saying no specifically to good clients, not just bad ones, mapped onto what had happened. I have since referenced it in my discovery process. My piece on how turning down Webflow clients made my solo practice more profitable reflects much of the same thinking.

What Did the Refund Teach Me About How I Scope Engagements?

The biggest lesson is that scope change requests need a small but real friction. When I said yes too easily in week one and two, I trained the client to keep asking. After this experience I added a written change order step to every fixed-fee engagement. Any scope shift over two hours gets a one-page change order, signed and dated by both parties, with a revised fee.

The change order is not punitive. Many requests are reasonable. The friction is the point. Even a five minute pause to write down the request forces both sides to think about whether it is worth it. According to the Project Management Institute's 2025 freelance benchmarks, projects with formal change order processes finish on time 84 percent of the time, versus 51 percent without them.

The second lesson is that fixed-fee scoping needs to assume a contingency. I now build a 15 percent contingency into every fixed-fee quote, communicated to the client as "scope reserve for the unexpected." Most clients appreciate the honesty. Reserves that go unused get refunded at project close, which has happened on three of the last five engagements.

How Did You Handle the Relationship After the Refund?

I wrote a short follow-up email a week after the refund. I shared the names of two other Webflow partners I trusted who I thought would be better fits for what she actually needed. One of them, a friend in Mumbai who specializes in fintech UX, ended up taking the project. The fintech founder thanked me directly, and I have stayed loosely in touch with her over Slack since.

The referral move was important. Refunding without referring leaves the client unresolved. Referring without refunding leaves the financial issue unresolved. Both together close the loop cleanly. According to Bain's 2025 small business retention research, clients who receive a refund plus a referral are 2.6 times more likely to refer the original freelancer in the future than clients who experience a project failure without resolution.

Two weeks ago I got a new inbound from a different founder in Singapore who mentioned the original client by name. The referral I worried I had damaged ended up multiplying. The mental story I told myself before the call had it backwards on every count.

What Policy Did You Change in May to Avoid This Again?

Three changes. First, I added a written scope reserve and change order clause to every fixed-fee proposal. Second, I added a midpoint check-in at the project halfway mark where we explicitly review scope drift and decide together if a change order is needed. Third, I now flag any project as a fixed-fee risk if discovery surfaces more than two integrations or more than one design system change, and switch those projects to a time-and-materials contract instead of fixed-fee.

The fixed-fee risk flag has already triggered twice in May. Both projects converted to a transparent hourly arrangement, both clients accepted the change, and both projects are running smoothly. The previous version of me would have quoted a fixed fee, agreed to too much, and dragged the engagement out. The new version catches it at discovery.

The change order clause has been used three times in May. Two clients wrote a change order and approved a small fee increase. One client wrote a change order, looked at the fee, and decided to defer the new scope to a future phase. All three outcomes are healthier than the alternative.

What Would You Tell a Webflow Partner Facing the Same Decision?

Refund earlier than you think you should. The clearest signal is dread. If you have started avoiding the Designer for a project, that is your data. Push through and you waste another four weeks of your life. Refund and you reclaim those four weeks for healthy work.

Make the refund proportional and fair, not punitive to yourself. Subtract from the refund the value of deliverables the client can actually use on their own, like brand assets, copy drafts, or a working Webflow staging file. Be transparent about how you calculated it. The client will respect the math.

And send the referral. The referral closes the relationship in a way that no refund alone can. My write-up on why I turned down my highest-paying client last month covers the related case where saying no early prevents needing to refund later.

How to Build a Better Scoping Process This Week

Open your standard proposal template. Add a paragraph on scope reserve. Add a clause on change orders for any work over two hours. Schedule a halfway-mark check-in for every active project, even the ones that feel fine. Update your discovery questions to surface the integration count and the design system risk early. None of this takes more than an hour, and it pays for itself the first time it prevents a refund.

If you have a current project that is making you dread opening Webflow, treat that as the signal. Have the hard conversation this week. It will be easier than the version where you push through another month.

If you want help thinking through whether to refund or how to scope your next engagement to avoid the situation, I am happy to talk it through. Let's chat.