A B2B SaaS client put a discount on the table last week. Sixty percent off the self-serve plan, aimed at the very large pool of free users who sign up, look around, and never pay anything.
It is the most natural idea in the world. You have people who didn’t buy. Price is the thing you control. Lower the price.
We killed it inside twenty minutes, and the reason it died is worth more than the discount would have been.
The arithmetic kills most discounts before the strategy does
Start with the money, because it is the part nobody checks.
This client’s self-serve users spend something like five to ten dollars a month. A sixty percent cut takes a five-dollar bill down to two. The entire lever — the whole instrument being pulled — is three dollars a month.
Nobody’s purchase decision turns on three dollars a month. Not a developer expensing it, not a small startup, not a hobbyist. At that price the transaction is already below the threshold where a person deliberates. These users are not sitting on the fence about the price. They are not on the fence at all.
A discount only functions when the saving is large enough to be felt against the thing it is buying. Ten percent off a forty-thousand-dollar contract is four thousand dollars, and a procurement officer will notice. Sixty percent off a five-dollar plan is a rounding error with a coupon code attached.
Before you design a discount, compute what it is worth to the person receiving it, in their currency, not in percent. If the answer is smaller than lunch, it is not an incentive. It is a gesture. The same arithmetic discipline that decides where a value gate belongs applies here: work the absolute number before the strategy.
The second problem: a public discount leaks upward
The other half of this client’s business — nearly all of its revenue, in fact — is six-figure annual enterprise contracts, several of them in live negotiation that week.
Put sixty percent off on the self-serve plan and it does not stay on the self-serve plan. It becomes a number in a procurement conversation. It becomes a screenshot in an email from a buyer who has just decided that your list price is theater. You cannot run a deep public discount on one tier while defending list price on another, because the buyer on the expensive tier is reading both pages.
That is what makes a self-serve discount genuinely dangerous rather than merely ineffective. The downside is not that it fails to work. The downside is that it costs margin on the deals that pay for the company.
It is a specific case of a general rule: the discount you offer at the bottom sets the ceiling on what you can defend at the top. If two tiers are visible to the same buyer — and on a public pricing page they always are — they are one pricing system, not two.
Like this? I send one idea like it most weeks. One PLG idea a week → subscribe
Then the real question, which has nothing to do with price
Once both arguments were on the table, the conversation moved to where it should have started. Which users is this even aimed at?
Split the free base on one question. Has this user connected their data and actually run the thing the product exists to do — reached the First Value Moment — or not?
For the ones who have, price may well be a live lever. They have seen the product work on their own data. They know what they would be buying. A well-sized, non-contaminating offer can move that person, and often usage growth moves them anyway without an offer at all.
For the ones who haven’t, price is answering a question they never asked. They do not hold an opinion about whether the product is worth five dollars, because they have never watched it do anything. Discounting to this group is marking down a meal for someone who never opened the menu. The number was never the obstacle.
The working rule we landed on: if a user has not reached first value within fourteen days, no price will change their behavior. Not a discount, not a longer free window, not a better-designed pricing page. They are stuck on the Golden Path, not on the price.
What actually moves the non-activated
The lever for that group is time and a person, not money.
The client’s own experience bears it out. When someone reaches out to a lapsed trial user and simply offers to extend the window, the reply is almost never “it was too expensive.” It is some version of “I ran out of time.” They signed up in a week that got away from them, the setup step needed thirty minutes they did not have, and the trial quietly expired while they were doing their actual job.
That is not a pricing failure. It is a scheduling failure. The fix is to give the time back and name the step: here is the one thing to do, here are another twenty days, here is my calendar if the setup is the part that is blocking you.
Unglamorous, doesn’t scale beautifully, works. That is the usual profile of the thing you should do first — and it is the same reason a product tour is wasted on a user who never got in. Whatever you send to the non-activated, the activation is the prerequisite, not the message.
One more honesty check: the segments don’t ladder
The last argument for the discount was strategic. Convert the self-serve users cheaply now, and some of them grow into enterprise accounts later.
The client’s own data said no. The person paying five dollars a month is, with rare exceptions, not a smaller version of the person signing a six-figure contract. Different buyer, different job, different problem. Discounting the small plan to manufacture large customers optimizes a ladder that does not exist.
That does not make the self-serve tier worthless. It makes it a different business with a different job — lead generation, distribution, developer goodwill, whatever it is in your case — and it should be measured against that job rather than against a graduation rate that never happens.
The order of operations
Before reaching for a discount on a trial or a free tier, three questions, in this order.
- Is the discount big enough to be felt? Compute the absolute saving, not the percentage. Under a threshold it is not an incentive.
- Where does it leak? If a more expensive tier is visible to the same buyer, the discount sets your defensible ceiling everywhere, not just where you posted it.
- Who is it aimed at, and did they ever reach first value? If not, you are pricing a product they have never used. Fix the path, give back the time, send a person. Then talk about money.
Most of the time you reach question three and find you were never looking at a pricing problem at all. It is the same read error that makes people compare a conversion rate to a benchmark without checking the gate that produced it — the number is real, and it is answering a different question than the one being asked.
The users who didn’t pay did not weigh the price and decline. They never got far enough in to have a price to weigh.
Three dollars was never the reason.
Every engagement starts by separating the users who reached first value from the ones who never did, before a single lever gets pulled. That is the first phase of the Trial Conversion Engine.
Running a trial you’re not happy with? Book a call and I’ll walk your funnel live.
