The Upgrade Problem That Has Nothing to Do With Price

Every founder I know has sat in a room debating whether the price should go up ten percent or fifteen. Almost none of them have sat in a room debating whether the free trial should require a credit card. That asymmetry is the subject of this essay, because a 2026 survey of 200 B2B software products found that the first conversation barely moves the number anyone actually cares about, and the second one might move it more than anything else you will do this year.

The survey, run by Kyle Poyar’s Growth Unhinged newsletter in partnership with ChartMogul and ProductLed, asked 200 operators what actually improved their free-to-paid conversion over the prior twelve months. The answers cluster around a finding that contradicts almost every founder’s intuition about pricing. I want to walk through what the data actually says, because I have made the mistake this essay is about, and I suspect most people reading it have too.

Why raising the price doesn’t move the number you think it does

The direct finding is this: simply raising or lowering the sticker price barely changes free-to-paid conversion rates. That is exactly why a careful price increase behaves like close to free money when it doesn’t spook anyone, and why a price cut rarely buys the growth founders expect from it.

The intuition almost everyone starts with is simple. Higher price, more friction, fewer conversions. Lower price, less friction, more conversions. It feels like physics. It is closer to superstition. Buyers deciding whether to convert from a free trial to a paid plan are mostly not comparing your number against a mental ceiling. They are deciding whether the product already proved its value during the trial. If it did, the price is a formality. If it didn’t, a lower price does not fix that. It just makes the miss cheaper for you.

I raised Sendspark’s price by a meaningful amount two years ago, nervous the whole week before it went live. Churn barely moved. What moved instead, later, was the packaging around a completely different plan tier, and it moved conversion more than the price change ever had. That is the pattern the survey confirms at scale: the decision to upgrade gets made upstream of the price tag, during the trial itself, and the number on the page is mostly a formality by the time someone reaches it.

This tracks with broader research on the topic. McKinsey’s work on B2B pricing power has repeatedly found that most B2B companies underestimate how much room they actually have on price, often by close to an order of magnitude, because the internal fear of churn is calibrated to a handful of vivid, memorable objections rather than the real base rate of who actually leaves. The founders who are most afraid to touch their price are usually the ones who have looked at the data least, not most.

What a 200-product survey actually found

The headline numbers are worth sitting with directly. The median free-to-paid conversion rate across the 200 surveyed products was 8 percent. A single percentage point of improvement on that baseline is worth roughly 15 percent more revenue per trial, which is a startling amount of upside sitting inside a number most founders treat as a fixed fact of their industry rather than something they can move.

Pricing and packaging changes were the single most cited category when survey participants were asked for their highest-impact growth experiment of the past year. That sounds like it contradicts the first finding, until you notice that “pricing” in the survey responses almost never meant the number. It meant the structure around the number: which plans exist, what each plan gates, how a trial actually behaves before someone hits the paywall. Founders describe changing their pricing. What most of them actually changed was the packaging.

That distinction is the whole essay. Pricing is one lever with one value. Packaging is a system with dozens of small decisions inside it, and the survey found that the system, not the lever, is where conversion actually gets won or lost.

Harvard Business Review’s long-running coverage of pricing research makes a related point about durability. A one-time price adjustment affects the next cohort of buyers who see the new number. A packaging change affects every trial user who passes through the funnel afterward, indefinitely, because it alters the shape of the decision itself rather than one input to it. That compounding quality is part of why packaging outperforms the number over any meaningful time horizon, even when the initial effort to ship it is larger.

The three packaging levers that actually moved conversion

Three categories of change showed up repeatedly among the operators who reported real gains: adjusting usage limits and feature gates, changing which plans exist, and adjusting trial mechanics. None of these touch the headline price. All of them change the shape of the decision a trial user is facing when the trial ends.

On limits and gates, giving free users a capped allotment of AI credits drove more usage, not less, because hitting a limit is the moment a user finally understands what the feature is worth to them. Paywalling specific features behind a rate limit did something similar: it converts a vague “you should upgrade” message into a concrete, felt constraint at the exact moment someone wants to do more.

On plan structure, one team found that removing the option to pause a paid plan, rather than only cancel it, increased renewal clarity. Pausing lets someone defer a decision indefinitely; canceling forces an actual choice, and forced choices convert better than deferred ones. Another team found that offering an existing customer a fresh trial of a new tier, something most companies only offer to brand-new signups, opened a second conversion moment that had been sitting unused.

On trial mechanics, one company added a second call to action next to their existing “get started free” button: a credit-card-required premium trial alongside the plain freemium option. That single change lifted premium trial starts by 26 percent, because it let high-intent visitors self-select into the path built for them instead of defaulting everyone into the lowest-commitment option.

None of these three examples required a pricing committee, a board conversation, or a single change to the number on the pricing page. Each one is a small, reversible product decision that a team could ship in a sprint and measure within a month. That is part of why they get skipped. A price change feels consequential enough to justify a meeting. A feature gate feels too small to justify one, right up until you add up what it is actually costing you every month it goes unmeasured.

Lever category	Example change from the survey	Why it moves conversion
Usage limits and feature gates	Capping free AI credits; paywalling a feature behind a rate limit	Turns an abstract "you should upgrade" into a concrete constraint felt at the moment of real intent
Plan structure	Removing the option to pause instead of cancel; offering existing customers a new trial	Forces an active decision instead of letting one get deferred indefinitely
Trial mechanics	Adding a dual call to action: freemium vs. credit-card premium trial	Lets high-intent visitors self-select into the right path; lifted premium trial starts 26 percent

The credit card asymmetry founders are afraid to test

The single largest effect in the survey was this: trials requiring a credit card upfront converted to paid at roughly 5 times the rate of trials that did not. Most founders will not test requiring one, because they are certain it will crater signups, and they are usually right about the signups and wrong about what that costs them.

I spent close to a year assuming a card-required trial would be a mistake for us, without ever actually running the test. The reasoning felt airtight at the time: fewer clicks to start means more people start, and more people starting means more people eventually paying. What that reasoning quietly assumes is that all signups are worth roughly the same amount, which is the part that turns out to be false. A visitor willing to hand over a card before touching the product is a different visitor than one who is not, and the gap in intent between those two groups is larger than the gap in headline signup counts.

When I finally looked honestly at what a smaller, higher-intent trial pool would be worth against a larger, lower-intent one, the arithmetic wasn’t close. A trial that converts at five times the rate needs a fraction of the volume to produce the same number of paying customers, and the customers on the other end tend to need less convincing in a sales conversation because they already made one real commitment before a rep ever spoke to them. I was optimizing for the number that was easy to watch climb on a dashboard, not the one that pays anyone’s salary. That is an easy mistake to make and a genuinely uncomfortable one to admit once you see the data next to it.

The fix is not to assume the answer either way. It is to test the requirement against your actual funnel, watch net paid conversions rather than raw signups, and let the real number decide instead of the instinct.

The test itself is not complicated. Split incoming trial traffic, send half through the existing no-card flow and half through a card-required flow, and hold everything else constant for a full sales cycle so you are not comparing a slow month against a fast one. What you are watching is not which group has more signups. It is which group produces more paying customers per hundred visitors who hit the page, because that is the only number that funds the next quarter. Most teams already have the analytics needed to run that split. What they are usually missing is the willingness to watch the signup count drop for a month before the conversion number tells them whether the drop was worth it.

Why founders reach for the price number instead of the packaging

The price number gets picked because it is a single, visible decision that one person can approve. Packaging requires product, growth, and sales to agree on which limits exist, which gates are active, and how the trial itself behaves, so the harder work with the bigger payoff keeps getting deferred in favor of the easier one with the smaller payoff.

A price change fits on one slide. It has one owner, one meeting, one date it takes effect. Packaging is the opposite of all three. It touches whatever the product team builds, whatever growth measures, and whatever sales promises in a demo, and none of those three groups owns the whole picture by default. I have watched this exact standoff happen inside my own company: everyone agrees packaging probably matters more than the number, and then nobody schedules the meeting, because the number has a clear owner and packaging does not. This is a close cousin of the free-to-paid handoff nobody on your team owns, a related gap I’ve written about before: activation and conversion get split across teams, and the split is exactly where the packaging work falls through.

The uncomfortable part is that this is not a strategy failure. It is an org design failure wearing a strategy costume. Nobody decided packaging was less important than pricing. It just never got assigned to anyone with the authority to ship the experiment, so the conversation defaults back to the one lever that already has an owner.

Paddle’s pricing research, published for years under the ProfitWell name before the acquisition, has documented this exact organizational pattern across hundreds of SaaS companies: the price gets revisited constantly, sometimes anxiously, while the packaging around it calcifies for years because it sits at the intersection of three roadmaps and belongs fully to none of them. The fix is not a better meeting about pricing. It is naming one person, usually someone in growth or product, who owns the packaging backlog the same way a PM owns a feature backlog, with the explicit mandate to ship and measure gate and limit experiments on a cadence, independent of whatever is happening with the number.

Running a packaging audit, not a pricing audit

A packaging audit asks a different question than a pricing audit. A pricing audit asks what the number should be. A packaging audit asks which limit, gate, or trial mechanic is quietly working against the outcome you actually want, and it deserves its own recurring meeting, separate from any conversation about the number.

The inputs are different from a pricing conversation too. Bring the conversion funnel broken out by trial type, not blended into one aggregate rate, because a card-required trial and a card-not-required trial are two different funnels wearing the same label. Bring a list of every feature gate currently live in the product and an honest answer to whether anyone has ever measured what it does, because most gates get added once and never revisited. Bring exactly one experiment queued for next quarter, because an audit that produces no experiment was really just a status meeting.

A first experiment does not need to be the credit-card test if that feels like too much to start with. It can be smaller: picking the single feature gate that has gone the longest without anyone checking whether it converts or repels, and instrumenting it properly for thirty days before touching anything else. The point of the first audit is not to fix everything the survey identified. It is to prove to the room that packaging experiments are cheap to run and fast to read, so the second audit gets scheduled without anyone having to argue for it.

Aspect	Pricing audit	Packaging audit
Question asked	What should the number be?	Which gate, limit, or mechanic is working against conversion?
Typical owner	Founder or CFO	Product, growth, and sales together
Cadence	Annually or when triggered by a board ask	Continuous, on its own calendar, independent of pricing
Typical output	A new price	A new gate, limit, or trial structure to test

This is a case for treating packaging as an ongoing discipline rather than a project, the same argument I made about the pricing decision most founders refuse to revisit: the audit itself, held on a calendar that does not move, is worth more than any single change it produces. Founder lessons like this one keep circling back to the same root cause, which is that the highest-impact work in a growing company rarely has a natural owner until someone insists on giving it one.

The closing thought

The price was never really the reason people didn’t upgrade. It is the easiest thing to blame because it is the easiest thing to change, and changing it gives you the feeling of having done something about conversion without the discomfort of touching the actual mechanism. The packaging around the price, the gates, the limits, the shape of the trial, is where the real decision gets made, and it is exactly the work most founders defer because no single person owns it.

I think about this every time someone on our team proposes a pricing tweak as the fix for a conversion problem. The tweak is rarely wrong to consider. It is just rarely the actual lever, and reaching for it first is a way of avoiding the harder, cross-functional conversation about the trial itself.

The founders who will pull ahead over the next few years are not the ones who found the perfect number. They are the ones who stopped treating packaging as a one-time launch decision and started treating it the way they already treat their product roadmap: something with an owner, a backlog, and a cadence of small, measured changes. The number will still come up in board meetings, because it is easy to talk about. The packaging work happens quietly in between those meetings, and it is the part that actually shows up in the revenue line six months later. If you want more essays like this one on the operator side of B2B growth, or on where AI is actually changing B2B sales rather than where it’s marketed to be changing it, more of them live at dearmer.com.au, and I’d rather you read the about page than take my word for what qualifies me to write them.

When was the last time you actually tested requiring a credit card on your trial, or measured what a single feature gate does to your conversion rate, instead of reaching for the price instead?