Are We Building Better Tools or Just Adding More Features?

The Feature Treadmill

Most software today is not getting better. It is getting bigger. Usage data shows roughly 80% of features in a typical product are rarely or never used, while the average desk worker now juggles nearly twice as many apps as in 2019.

The Giant knife is a perfect picture of modern software. Every tool works. Together, they make something nobody wants to use. This pattern has a name, feature creep, and it has quietly become the default way products grow.

This article asks a simple question with hard numbers: are we building better tools, or just adding more features? The short answer is that most teams are adding, and their users are paying for it in time, focus and money.

Most Features Go Unused

Just 12% of features generate 80% of daily usage in the average software product. That finding comes from Pendo's Feature Adoption Report, which analyzed usage across 615 product subscriptions with more than a year of data.

The flip side is harsher. The same study found that 80% of features are rarely or never used. Pendo then put a price on it: a software company with $50 million in revenue might spend about $8.4 million building features customers barely touch. Across public cloud companies, the estimate reached $29.5 billion.

The waste is not only in what vendors build. It is also in what buyers pay for. Torii's 2026 research found that 65% of SaaS licenses sit unused or underused. Companies are buying seats for tools that nobody opens, filled with features that nobody clicks.

Tool Sprawl

The Hidden Tax on Attention

Workers lose about 9% of their working year just reorienting after switching apps. That is the headline from a Harvard Business Review study that tracked 137 users across 20 teams at three Fortune 500 companies for up to five weeks.

The researchers found the average user toggled between apps and websites nearly 1,200 times a day. Each switch cost a little over two seconds to refocus. Small on its own, those seconds added up to just under four hours a week, or roughly five working weeks a year.

The app count keeps rising. Gartner found the average desk worker used 11 applications for work, up from 6 in 2019. Okta's Businesses at Work 2025 report showed the average company now deploys 101 apps, crossing 100 for the first time. Torii's discovery data, which also catches unapproved tools, counts 831 apps in the average organization.

The most telling number may be the last one. Ivanti found that 74% of IT professionals see clear overlap and redundancy in their tools, yet 63% say consolidation is not a high priority. Everyone sees the mess. Few are cleaning it up.

Why Companies Keep Adding Features

Feature creep is rarely a design failure. It is an incentive problem. Teams are rewarded for shipping, not for removing, so the product grows in one direction only.

Five forces keep the treadmill running:

•  Roadmaps measure output. Release notes and launch counts are easy to report. Adoption, time saved and fewer support tickets are harder to show, so they get less attention.

•  Sales checklists drive the backlog. One large prospect asks for a niche capability, and it ships for everyone. Comparison grids reward the longest list, not the best workflow.

•  Competitors copy each other. When a rival launches something, matching it feels safer than ignoring it, even when few customers asked.

•  Removal feels risky. Deleting a feature can upset a small, loud group of users. Keeping it costs nothing visible, so it stays forever.

•  Pricing rewards bundles. Higher tiers need more features to justify the price, so vendors pack them with extras rather than better core tools.

The result is a product that looks stronger in a demo and feels weaker in daily use. Every new menu item makes the useful ones a little harder to find.

The AI Layer

More Features, Unclear Value

AI is now the fastest way to add features, and the data shows a wide gap between adding AI and getting value from it. Torii's discovery data counted 694 new AI-native apps in 2025, up from just 11 in 2020.

The value has not kept pace with the volume. In McKinsey's State of AI 2025 survey, 64% of respondents said AI is enabling innovation, yet only 39% reported any EBIT impact at the enterprise level. Most of those said AI drives less than 5% of their EBIT.

The picture from MIT is starker. As Fortune reported, MIT's NANDA initiative found that about 95% of generative AI pilots at companies were delivering little to no measurable impact on profit and loss. The research pointed to weak integration into real workflows, not weak models, as the main cause.

This is feature creep at a new speed. An AI button inside every app is easy to ship. An AI that removes three steps from a real workflow is hard to build, and that is where the value lives.

What Better Tools Actually Look Like

A better tool is measured by the work it removes, not the features it adds. The products people love tend to share a few traits, and none of them show up on a feature checklist.

Three principles separate the two:

1.  Depth over breadth. If 12% of features drive 80% of usage, the best investment is usually making that 12% faster, clearer and more reliable.

2.  Subtraction as a skill. Strong product teams treat removing a feature as a win. A smaller surface means less to learn, less to maintain and fewer bugs.

3.  Fewer handoffs. With only 27% of enterprise apps connected to each other, integration often creates more value than any new capability. Every switch you remove gives back seconds that add up to weeks.

Lessons From Products That Chose Focus

The best proof that subtraction works comes from companies that tried it at scale. Three cases stand out, and each one changed a product by making it simpler, not bigger.

Microsoft Office 2007: the features were already there. By the mid 2000s, Office had grown to roughly 2,500 features. Microsoft's research found that about 90% of the feature requests it received were for things Office could already do. Users simply could not find them. Instead of adding more, the team rebuilt the interface around the Ribbon. Microsoft later reported that Word and Excel users were using about four times as many features as before, and PowerPoint users about five times as many.

Google's spring cleaning: focus over sprawl. In 2013, Google shut down Google Reader and seven other products, bringing the total closed since its 2011 cleanup effort to 70. The company said it needed to stop spreading itself too thin. Many of the retired services duplicated other Google products. The move was unpopular with Reader fans, which shows why removal takes courage.

Apple in 1998: four products instead of dozens. When Steve Jobs returned to Apple, the company sold a confusing range of overlapping models. He cut the lineup down to a simple grid of four computers: consumer and professional, desktop and portable. That focus helped set up the iMac and the turnaround that followed.

The common thread is simple. Each team stopped asking what else it could add and started asking what was getting in the way.

Metrics That Separate Better Tools From Bigger Ones

What a team measures decides what it builds. If the only number on the dashboard is features shipped, the product will keep growing whether or not it gets better.

These metrics shift the focus from output to outcomes:

The goal is not to track all of these at once. Pick two that match your biggest problem and review them every quarter. A product that ships fewer features but cuts task time in half is the better tool, even if its release notes look thin.

A Practical Checklist for Builders and Buyers

Both sides of the market can break the cycle with a few simple habits.

For product teams

• Tag every feature and track its adoption before planning the next one

• Set an adoption target for each new feature at launch, and review it after 90 days

• Run a quarterly "sunset review" and retire anything below your usage threshold

• Ask "what step does this remove?" before approving any AI feature

• Report time saved and tasks completed alongside features shipped

For software buyers and IT leaders

• Audit actual license usage, since up to 65% of seats can sit idle

• Map which tools overlap and pick one owner per job to be done

• Favor tools that integrate with your core stack over standalone point tools

• Pilot AI tools against one measurable workflow, not a general promise

• Count app switches for one high-volume team, then target the worst handoffs first

Conclusion

Build Less, Help More

The numbers point one way. We are mostly adding features, not building better tools. Eight in ten features go unused, workers lose about 9% of their year to app switching, and only a small share of AI projects show real financial returns.

The fix is not to stop building. It is to change what counts as progress. A release that removes a step, connects two tools or makes the core 12% faster is worth more than ten new menu items.

The next great product will not be the one with the most features. It will be the one people barely notice, because it simply gets the work done.