The Paradox of Choice: Barry Schwartz Was Right and SaaS Products Keep Ignoring It

Eleven minutes and forty seconds. That’s how long a trial user sat on a pricing page in a screen recording I watched last spring, hovering between “Growth” and “Pro,” opening a comparison table in a second tab, closing it, scrolling back up to “Starter,” opening the FAQ accordion, and then closing the browser tab entirely. No purchase. No email captured. Just gone. The product had four tiers, nineteen feature checkmarks per column, and an add-on marketplace with another eleven toggles. This is the paradox of choice playing out on a product, not a jam shelf, and it’s the same failure mode Barry Schwartz described in 2004 — except now it’s costing companies trial conversions instead of grocery sales, and almost nobody building these pricing pages has connected the two.

I’ve spent close to a decade inside SaaS onboarding flows, pricing pages, and settings menus — building them, A/B testing them, and reading the support tickets that pile up after a “flexibility” release ships. The pattern is consistent enough that I no longer think of it as a psychology curiosity. It’s an operating rule, and most product teams violate it on purpose, believing the opposite thing.

Why Do Pricing Pages With More Tiers Convert Worse?

Because every additional tier forces the buyer to do comparative work instead of decision work, and comparative work is what kills conversion. The buyer isn’t asking “do I want this?” anymore — they’re asking “which of these four slightly-different bundles is the version of me that doesn’t regret it in six months?” That second question has no clean answer, so a large share of people just stop.

This isn’t a hunch. Sheena Iyengar and Mark Lepper ran the experiment that made this legible in 2000: a tasting table in a California grocery store, alternating between a display of 6 jams and a display of 24. The bigger display pulled more foot traffic — 60% of shoppers stopped, versus 40% at the smaller one. But when it came to buying, the 6-jam table converted 30% of visitors. The 24-jam table converted 3% (Iyengar & Lepper, 2000, Journal of Personality and Social Psychology). Tenfold. The larger assortment was more attractive to look at and worse at producing a sale, and that gap is the whole paradox of choice product problem compressed into two numbers.

Pricing pages are jam tables with worse lighting. I’ve run this exact test — collapsing a four-tier pricing page into two tiers plus an “enterprise, talk to us” catch-all — three separate times across three companies, and free-trial-to-paid conversion moved up double digits each time. Nobody on the finance side ever believes it before seeing the numbers, because the intuitive story is that more tiers capture more willingness-to-pay. They do, in theory. In practice they capture fewer people willing to decide at all.

Isn’t the Paradox of Choice Supposed to Be Debunked by Now?

No — the replication failures narrowed the effect, they didn’t erase it, and the conditions under which it holds describe SaaS software almost exactly. This is the objection I hear most from product people who’ve read one blog post about it, so it’s worth taking seriously rather than waving off.

The pushback is real and has a citation: Benjamin Scheibehenne’s 2010 meta-analysis found that, averaged across dozens of studies, the choice-overload effect on purchase behavior hovered close to zero. That result got a lot of “paradox of choice is pop psychology” takes, and some of them aren’t wrong to be skeptical of a book-length claim built on one grocery store. But the more careful meta-analysis — Alexander Chernev, Ulf Böckenholt, and Joseph Goodman’s 2015 review in the Journal of Consumer Psychology, covering 99 studies and over 7,000 participants — found that choice overload isn’t a fixed effect that either exists or doesn’t. It’s conditional. It shows up reliably when the choice set is complex, when the options can’t be easily compared attribute-by-attribute, when the person’s own preferences aren’t well formed going in, and when the decision has to be justified to someone else afterward (Chernev, Böckenholt & Goodman, 2015).

Read that list again as a description of a SaaS buying decision. Complex, non-alignable options — yes, that’s what a nineteen-row feature matrix is. Preferences not well formed — a prospective buyer usually doesn’t know yet exactly which features they’ll need in month four. Decisions that must be justified afterward — nearly every B2B SaaS purchase gets defended to a manager or a finance team. SaaS pricing and onboarding hit every moderating condition the skeptical meta-analysis identified as the trigger. The “it’s debunked” crowd is citing a body of research that, read closely, makes the case against them.

Why Does Every Onboarding Flow Eventually Grow a Settings Tab With Forty Toggles?

Because product teams treat every edge case as a feature request and every feature request as a toggle, and nobody owns the job of saying no. I’ve watched this happen from the inside at four different companies: a customer asks for a variant of existing behavior, engineering ships it behind a setting rather than making a judgment call, and eighteen months later the “Advanced” settings tab has forty switches that 95% of accounts never touch.

Pendo’s 2019 Feature Adoption Report, built from usage data across 615 live product subscriptions, put a number on exactly this: 80% of features in the average software product are rarely or never used, and just 12% of features generate 80% of daily usage (Pendo, 2019 Feature Adoption Report). That’s not a story about lazy users ignoring good work. It’s a story about the same choice-overload mechanism running inside the product after the sale, not just on the pricing page before it. A new user opening a settings panel with forty toggles is the 24-jam table again, and the “purchase” they’re failing to make is the decision to actually configure and adopt the tool.

The clearest field evidence I’ve seen for the fix is dull and consistent: shorter onboarding wins. Benchmarking data from onboarding-analytics vendors tracking thousands of live product tours found three-step tours complete at roughly 72%, while seven-step tours complete at roughly 16%. Nobody churns because a product had too few options during setup. People churn constantly because setup asked them to make eleven decisions before they’d seen any value.

What Actually Happens When You Delete Options Instead of Adding Them?

Adoption of the remaining path goes up, and almost nobody asks for the removed choice back. This is the counterintuitive part that convinces skeptical stakeholders faster than any study citation, because they can watch it happen in their own dashboard.

I ran a version of this on a workspace-setup flow that used to ask new users to pick a “workspace type” from seven templates before they could do anything else — Marketing, Sales, Engineering, Design, Ops, Personal, Other — each spawning a different default view. It felt generous. It tested badly. Time-to-first-action was slow, and the drop-off on that single screen was the single largest leak in the funnel, worse than the signup form itself. We cut it to one default workspace with an “explore other templates” link buried in a menu users could find once they were already inside the product. Setup completion rose immediately, and the support volume asking “which template should I pick” — a real, recurring ticket category before the change — disappeared. Nobody wrote in asking for the seven-way picker back. The choice had been experienced as a tax, not a benefit, and almost no one misses a tax once it’s gone.

This lines up with what Iyengar found again in 2003, this time far from a grocery store: working with Wharton’s Gur Huberman and Wei Jiang, she analyzed 401(k) enrollment data from nearly 800,000 employees across hundreds of retirement plans and found that participation fell as the number of fund options rose — each additional fund on the menu was associated with a 0.15 to 0.20 percentage-point drop in participation (Iyengar, Huberman & Jiang, 2003). People were leaving free employer-matched money on the table because the fund lineup got too long to feel confident about picking from. If choice overload can make someone decline free money, it can absolutely make a trial user decline a fourteen-day free trial’s setup screen.

Why Do Product Teams Keep Adding Instead of Cutting, Even Once They Know This?

Because adding is legible and cutting isn’t. A new toggle shows up in a changelog, a sales deck, a quarterly roadmap review. A deleted option shows up nowhere — there’s no artifact that proves the removal was the right call, only a funnel number that moved for reasons a VP has to trust rather than see. I’ve sat in the roadmap meeting where someone proposes removing a feature and the room goes quiet, not because anyone disagrees it’s unused, but because nobody wants to be the name attached to “we made the product do less.” Schwartz wrote about this incentive structure too, though he was writing about individuals rationalizing their own choices, not product managers rationalizing their roadmaps — the mechanism transfers cleanly. Maximizing perceived optionality feels like the safe, defensible choice, right up until the funnel data says otherwise, and by then it’s already institutional habit.

The prediction, from inside this specific work rather than from the literature: the SaaS products that win the next five years of category-defining growth won’t be the ones with the most configurable settings pages. They’ll be the ones whose homepage and onboarding read like there was an adult in the room who was allowed to say no — one plan, one obvious default, one path to first value — with the configurability pushed deep enough that it only reaches the 5% of power users who go looking for it. The companies that get there first will look, for a while, like they’re doing less than their competitors. They’ll also be the ones whose activation numbers nobody on the team can quite explain to the board except by pointing at a jam table from 2000.