I Did Free Work for a Leaving Client. They Stayed. The Math Behind Free.

You have done it too. A client says they want to pause the retainer, money is tight, the timing is off. You feel the urge to slash your price to keep them. Offer 20% off, maybe 30%. Hold the account.

I did something else. The client was Zulal, on a monthly retainer, and they were on their way out. Their site loaded slow. PageSpeed sat at 39 on mobile. So I fixed it. Took it to 77. I charged them nothing.

They stayed.

Most people read that and assume I got lucky or that I was being soft. Neither. There was math behind it, and the math is the whole point of this post.

A discount says "I was overcharging you"

See, when a client wants to leave and you drop the price, watch what you actually told them. You told them the old number was negotiable. You told them you can deliver the same thing for less. Which means the first number was air.

That is the trap with retention discounts. You think you are giving the client a reason to stay. What you are giving them is a reason to distrust every invoice you have ever sent. The relationship survives, the respect does not.

A discount is a downgrade of your own work in front of the person paying for it. That is lazy. It is the easiest move and it is the one that costs you the most later.

Free work, done right, sends the opposite signal. It does not touch the price. It says the price was always fair, and here is more on top, because I want this to work.

What actually happened with Zulal

Their PageSpeed was 39 on mobile. Bad. Images were uncompressed, fonts were loading render-blocking, and there was a heavy third-party widget firing on load that nobody was using anymore.

I did not ask. I did not send a quote. I spent an evening on it. Compressed and converted the images to modern formats, deferred the font loading, killed the dead widget, and moved the layout shift offenders into reserved space so the page stopped jumping. PageSpeed went from 39 to 77.

Then I sent them the before and after. Two screenshots, one line: "noticed your site was slow, fixed it, no charge."

Here is the part people miss. I did not do it only as a strategy. I did it because I loved doing it. The work itself is the reward for me. The flow state of fixing a broken thing, watching the number climb, that is dopamine I would chase for free anyway. The strategy and the love sat in the same act. When those two line up, you have found work worth doing.

They replied within the hour. They did not leave. The retainer ran on for several more months after that.

The two principles doing the work

There is real psychology under this, and it is worth naming so you can use it on purpose instead of by accident.

The first is reciprocity. Robert Cialdini documented it across cultures: when someone gives you something unrequested, you feel a pull to give back. A waiter who drops a single mint with the bill lifts tips. The free fix created a debt in the client's head that a discount never could. A discount is them taking something from you. A gift is you giving something to them. The direction of the favor flips, and so does the feeling.

The second is proof of competence. A leaving client is rarely leaving over money alone. Money is the reason they say out loud. Underneath it is a quieter question: is this person still worth it? When I fixed the speed without being asked, I answered that question with a number they could see. 39 to 77. No meeting, no pitch deck, no "let me tell you about our value." The work argued for itself.

Reciprocity plus proof of competence beats a discount every time, because a discount answers neither question. A discount just makes you cheaper. Cheaper is not a reason to stay. Cheaper is a reason to leave slower.

The asymmetric return

Here is the math that made the evening worth it.

The fix cost me maybe four hours. Call it an evening I would have half-spent scrolling anyway. The downside was capped at that. Four hours, gone, worst case.

The upside was not capped. A client who stays on retainer is months of revenue. A client who feels they owe you sends referrals, the kind that show up already trusting you because their friend told them you go above what you charge for. Every client I have ever had came organically, never from cold outbound, and the engine behind that is exactly this: people talking about work that surprised them.

Small known cost, large unknown payoff. That shape is the only kind of bet worth making with free work. You are not giving away your time. You are buying a lottery ticket where the ticket is cheap and the floor is your craft getting one more rep.

For example, that same Zulal site is now something I can point to. The number is real, the work is mine, and it earns its keep in conversations long after the evening ended.

When free work is strategy and when you are just being unpaid

This is where most people get burned, so read this part twice.

Free work is strategy when three things are true. The work is small and bounded, so your downside is capped. You chose to do it, nobody asked or expected it. And the relationship already has trust in it, so the gift lands as generosity, not desperation.

Free work is just being unpaid when any of those break. A client who demands free work is not in a reciprocity loop, they are in an extraction loop, and the more you give the less they value it. Open-ended free work with no edge to it bleeds you out, because there is no finish line and no number to point at when you are done. And free work for someone who has not earned trust yet reads as weakness, the same way a discount does.

The tell is simple. Ask yourself: am I giving this, or is it being taken from me? If you decided, picked the scope, and set the boundary, it is a gift and it will work like one. If they asked, kept asking, and you caved, that is not strategy. That is a leak.

I have a hard floor on this. I will fix a client's PageSpeed for free in an evening for goodwill. I will not rebuild their checkout flow for free because they asked nicely. One is a gift I control. The other is a job I am not getting paid for. The size of the ask is the line.

How to actually run this

If you want to try it, here is the version you can use this week.

Pick a client who matters and who you have history with. Find one thing on their product that is broken and that they have not noticed. Slow page, a typo on the pricing section, a form that fails on mobile, a meta description that is empty. Small. Something you can finish in one sitting.

Fix it without telling them first. Then send proof, not a story. A before number and an after number. One line, no pitch, no "by the way we also offer." The second you attach a sell to a gift, it stops being a gift and becomes a sample, and people smell that instantly.

Then stop. Do not do it again next week as a habit. The power is in it being unexpected. A gift you give every Monday is just a service you forgot to invoice.

So here is the question to sit with. The next time a good client starts drifting, are you going to cut your price and quietly tell them your work was never worth what you charged, or are you going to spend one evening proving it was worth more? One of those keeps the client and burns your floor. The other keeps the client and raises it. Pick the one you can defend a year from now.

If you have done free work that paid off, or one that bled you dry, reply and tell me which side of the line it sat on. I read everything.