Complex Easy vs Simple Hard

Between January 2021 and December 2023, monthly U.S. fills of semaglutide rose 442%, from 471,876 to 2,555,308. Over the same window, the share of U.S. adults who met both the aerobic and muscle-strengthening exercise guidelines held near 22.5%. One number more than quintupled. The other did not move. The drug and the gym both target the same goal, and in growing numbers, people chose the drug.

This is a pattern, not an isolated case. Look at money, sleep, and appearance and the same shape appears: adoption of the harder-to-explain, easier-to-keep-up option rises while the plain behavior it could replace stays flat.

The rising option is what we call complex-easy, the flat one simple-hard. Both terms are easy to misread, so each deserves a precise definition. Easy does not mean effortless. It means the option removes the daily self-regulation the plain behavior demands, the meal-by-meal restraint or the nightly routine; it can still cost money, carry side effects, or require a procedure. Complex does not mean the act is complicated, since a weekly injection or a tap to pay later is simple to do. It means you now lean on a medical, technical, or financial system, with its own costs and risks, instead of on your own repeated willpower. Eat less and move more is simple-hard: anyone can do it with nothing, but it is hard to keep up. A weekly appetite-suppressing injection is complex-easy: simple to administer, complex in what it commits you to, and easy on the part people find hardest. The trade is not effort for ease but one cost for another: instead of paying in willpower, you pay in money, complexity, and risk, much of it carried by an outside system.

This piece is a synthesis of public sources, not Happily first-party data. It is a field note: we collected published series on weight, money, sleep, and appearance and looked for what they share. The numbers below all come from named public datasets, listed in the methodology box and the references. Where a behavior has only a single recent endpoint or a qualitative "unchanged" reading, we draw it as a flat reference line, which is consistent with what those sources report.

The fix keeps rising, the habit stays flat

The four domains share one shape. The paid, more complex option climbs; the plainer behavior it could substitute for is flat.

Show each series as percent change since it began and the whole pattern resolves into a single view: the paid fixes climb, the habits sit on the zero line.

The same divergence appears domain by domain, across each series' own history.

Weight: ask your doctor vs go to the gym

Semaglutide is the clearest case. Monthly mean U.S. fills rose from 659,492 in 2021 to 1,148,123 in 2022 to 2,270,564 in 2023. Month to month the rise was steeper: fills went from 471,876 in January 2021 to 2,555,308 in December 2023, a 442% increase. By 2025, 12% of U.S. adults reported currently using a GLP-1 drug and 18% reported ever having used one; among adults diagnosed as overweight or obese, those figures rose to 23% and 34%. The behavior that the drug stands in for, meeting the activity guidelines, sat at 22.5% of adults 25 and older in 2022 and is treated here as a flat reference.

Money: smoothing payments faster than building buffers

The money series tells the same story in a different currency. Buy now, pay later use among U.S. adults rose from 10% in 2021 to 15% in 2024, and the share of BNPL users who paid late climbed from 15% to 24% over the same years. The simple-hard alternative, holding a cash cushion, did not grow: the share of adults who could cover a $400 emergency with cash or its equivalent was 68% in 2021 and 63% in each of 2022, 2023, and 2024. People got better at spreading a payment across weeks, not at having the money on hand.

Sleep: a product category that has not moved the behavior

Sleep shows the divergence over a longer arc. Adult melatonin use rose from 0.4% of U.S. adults in 1999-2000 to 2.1% in 2017-2018, and high-dose use above 5 mg a day rose from 0.08% in 2005-2006 to 0.28% by 2017-2018. By 2024, 12.9% of U.S. adults used some sleep aid most days or every day, split across over-the-counter or supplement products (5.7%), prescription medication (5.2%), and marijuana or CBD (3.7%). Yet the share of adults reporting insufficient sleep was unchanged from 2013 to 2022, ranging by state from 30% in Vermont to 46% in Hawaii. The category grew; the problem it sells against did not shrink.

Appearance: rising worldwide

Appearance is the one global series, and it lacks a simple-hard counterpart to chart against, so it stands alone as adoption only. Total aesthetic procedures reported by ISAPS rose from 30,439,576 in 2021 to 37,951,364 in 2024, a 24.7% increase. Both halves grew: surgical procedures from 12,840,688 to 17,415,678, and non-surgical from 17,598,888 to 20,535,686. The same direction holds in a fourth domain measured very differently from the others.

Even this is a floor. ISAPS projects from board-certified plastic surgeons, so it captures the surgeon-reported slice and largely misses the much larger non-surgical, clinic-based market: the injectables, lasers, and skin tightening done by dermatologists, medspas, and hospital skin centers. The scale of what it leaves out shows up in a single market. The U.S. International Trade Administration counts about 7,000 aesthetic clinics in Thailand and a medical-aesthetics market forecast to reach $7.5 billion by 2027, roughly three times its 2020 level. Read that way, the rising line here understates the shift toward the easier paid option rather than overstating it.

Why the complex option wins even when an easy one exists

What follows is interpretation grounded in published behavioral science, not a measured finding from this data. The series above show adoption of complex-easy options rising, but they contain no parallel adoption series for the simple-easy alternative, so the reasons people skip the cheap simple fix cannot be read off these charts. They are an inference the data invites, weighed against the most relevant published research.

Frame the everyday choice on two axes. One axis is how hard the option is to sustain day to day; the other is how complex it is under the hood. That gives four cells, and the cell that keeps winning is complex-easy.

Several published findings help explain why the simple-easy fix loses. The first is that price and effort act as signals of efficacy. Shiv, Carmon, and Ariely (2005) found that people who paid a discounted price for an energy drink solved fewer puzzles than people who paid full price for the identical product, and Waber, Shiv, Carmon, and Ariely (2008) found that a higher-priced placebo produced more pain relief than the same pill described as discounted. A cheap, simple fix carries the opposite signal. Closely related is the effort heuristic of Kruger, Wirtz, Van Boven, and Altermatt (2004): people rate work as higher quality when they believe more effort went into it. A 2023 replication is mixed, with the monetary-value effect unsupported, so the defensible version is narrow: the effect holds for perceived quality and liking of ambiguous things, which is the form that matters here, where a serious-looking remedy reads as the one that works.

A second strand is about agency. The simple-hard route asks for willpower every single day; the complex-easy route asks for one decision that an outside system then carries out. Johnson and Goldstein (2003) showed how powerfully defaults shape consequential choices: countries with opt-out organ-donation systems show far higher consent than opt-in countries, because the path of least resistance does the work. A weekly injection, an autopay financing plan, or a procedure converts a recurring willpower problem into a single act of outsourced agency. Commensurability also plays a part. A serious goal seems to deserve a serious remedy, and the medical or technical framing of the complex option borrows authority that a glass of water before a meal does not have.

The honest counter-tension cuts the other way. Alter and Oppenheimer (2009) showed that information which is easy to process is judged as more true, more likeable, and more credible, which should favor the simple, easy-to-grasp option, not the complex one. Processing fluency and the price-as-quality and effort heuristics pull in opposite directions. The resolution is about the kind of decision. When a choice is deliberate and low-stakes, fluency tends to win and simple looks credible. When the decision is fast and low-deliberation but the goal is high-stakes and personal (weight, money, sleep, the way you look), the efficacy and agency signals dominate, and complex-easy beats simple-easy. That is the condition these four domains share, and it is consistent with the rising adoption the data shows.

There is one more reason the simple option struggles, and it runs against the fluency effect rather than with it. New things are intrinsically rewarding. Novel stimuli activate the brain's dopamine reward system (Bunzeck and Düzel, 2006), and people reach for variety and the unfamiliar even when the familiar option works fine (McAlister and Pessemier, 1982). The complex-easy options are new, branded, and carry a story of a breakthrough; the simple-easy ones are old, unbranded, and feel stale. Water before a meal promises nothing, while a new injection, app, or procedure carries the perennial promise that this time the goal will finally be reached. Where fluency makes the familiar feel safe, novelty makes the new feel exciting, and for a goal the familiar approach has visibly failed to deliver, excitement tends to win. But novelty drives adoption, not adherence. That may be part of why uptake keeps climbing, as each new method draws its own burst of attention while the last one's shine fades, and it sits uneasily next to the high rates at which these solutions are later abandoned.

What this means

If you design advice, benefits, or programs that compete with a paid complex option, the takeaway is not that the simple option is worse. It is that the simple option loses on perceived efficacy and on signaling, not on cost or effort, so it has to win those back.

Engineering out the friction

Step back, and this looks less like a run of individual choices than like an environment being redesigned around us. Technology has always cut effort, and mostly that is progress; the washing machine and the calculator freed people for better things. But there is a difference between outsourcing a chore and outsourcing the practice of self-regulation itself. The simple-hard behaviors were never only about the result. Doing them rehearsed patience, restraint, and the everyday discipline of choosing the harder thing, and that rehearsal may matter as much as the outcome. A market that strips the friction from every goal at once strips out the rehearsal too. The worry is not that individuals are weak. It is that we are engineering the friction out of daily life faster than we are replacing what it built, and a capacity left unpracticed can fade, the way a sense of direction fades once we hand it to the GPS (Sparrow, Liu, and Wegner, 2011). Whether that adds up to a slow erosion of self-regulation across a population, a character crisis we are building rather than one anyone chose, is a question these four rising lines raise and cannot settle. It is the kind of claim a longitudinal study would have to test.

Limitations

• Geographies are mixed. Weight, money, and sleep are U.S. series; appearance is global. The domains are not directly comparable and are lined up for pattern, not for a pooled estimate.
• ISAPS counts procedures, not unique people, and includes medical tourists, so the appearance series can rise without the number of individuals rising as fast. It also projects only from board-certified plastic surgeons, which makes the figure a conservative floor: it excludes the much larger non-surgical, clinic-based market of injectables, lasers, and skin tightening delivered by dermatologists, medspas, and non-surgeon providers. Sizing that full market would take a consumer survey paired with clinic-throughput data, not a surgeon-based projection.
• No simple-easy adoption series exists in this data. The interpretation of why the cheap simple fix is skipped is grounded in published behavioral science, not measured here.
• Where search interest is referenced as a method, Google Trends is normalized interest, not volume.
• Everything here is correlational. No intervention was run, and no causal claim is made; the series show association and direction, not cause.