The End of Buying the Whole Suite: Why Small Businesses Are Paying for 400 Features to Use 6

If you run a small business or a nonprofit in Asheville, you've probably been buying software the same way for a decade or more. You pick the best-looking platform in whatever category you need. You pay per seat, per month. You bend the work around whatever the vendor decided your work should look like. Need a CRM? Rent Salesforce's version of one. Need project management? Pay for somebody else's opinion of what that means. The subscriptions pile up. Nobody reads the renewal emails. And by year three, most organizations around Western North Carolina are running a software stack nobody on the team ever affirmatively chose.

That deal worked because building your own tool was impossible. Not hard. Impossible, for a five-person nonprofit or a regional outfitter. Custom software cost six figures, took months, and demanded a developer on permanent retainer. Against that backdrop, paying $30 a seat per month to rent somebody else's sprawling platform wasn't a mistake. It was the right call.

The premise that made it the right call has quietly broken. Most businesses haven't noticed yet.

Why the whole-suite model ever made sense

Give the last decade its due. Storage was expensive. Laptops struggled to run anything heavy. Building custom software was slow and costly. Vendors had every reason to ship big, sprawling platforms with hundreds of features: that's how they justified the infrastructure investment, and economies of scale only worked at the largest scope. A small business couldn't realistically build its own software. Renting somebody else's well-funded platform was a legitimate trade.

And the big platforms did get good. Monday, Airtable, HubSpot, QuickBooks Online. Capable tools, most of them. That isn't the argument. The argument is that the conditions that justified renting them have moved, and the economics haven't caught up yet.

What actually changed

Three things shifted in the last couple of years. None of them were loud.

First, storage got trivially cheap. A terabyte of consumer hard drive capacity ran $30 to $40 back in 2015. Today it's under $20 for bulk drives, and about $20 to $25 per terabyte on NAS-class storage (per Backblaze's Drive Stats and John McCallum's long-running memory-price survey). A four-bay enclosure stocked with enough drives to cover a small business's entire operation — files, documents, backups, photo archives — lands in the low thousands as a one-time purchase. A few years ago that was maybe a year of cloud storage for a mid-sized team.

Second, AI got good at working against data it doesn't own. This is the one most people miss. The shift isn't that AI got cheap. It's that AI got portable. A capable model like Claude or ChatGPT can now reach into the files on your laptop, your office computer, or a small storage box sitting on your network, and do real work against them. Your data doesn't have to migrate into a vendor's cloud for the AI to help. The AI travels to the data. Does the job. Leaves. Your data never has to become somebody else's asset on somebody else's server.

Third, building a small, focused tool got radically faster. The expensive part of custom software was always the developer labor, and AI has compressed that enormously. The kind of small internal tool that would have cost $30,000 and three months in 2019 can often be built in an afternoon now, against your real data, in the shape of your real workflow. That's the change that matters more than anything else on this list.

Stack those three on top of each other and the arithmetic of renting a 400-feature platform to use six of them quietly stops adding up.

What you're actually paying

Most owners haven't audited their own stack in years, so the waste is worse than it feels. Small teams pay list price. They don't get volume discounts. Vendor pricing is deliberately structured to push small teams into higher tiers to unlock basic things like multi-factor authentication or permissions that don't embarrass you in front of your staff. And that's before anyone starts counting seats nobody actually logs into.

The feature number is the one that stings. Pendo analyzed anonymized usage across hundreds of subscriptions and found that 12% of features drive 80% of daily usage. The other 88% mostly exist to differentiate the marketing page. You are paying, every month, for 400 features so that six of them can do their job.

The seat number is worse. Zylo's 2025 SaaS Management Index, built on eight years of data across more than 40 million SaaS licenses, found that 53% of licenses go unused or underused each month. Half the seats you're paying for are sitting idle. It isn't malice. Nobody is trying to waste the money. The audits just never happen.

The greenfield-only problem

There's an honest flaw in how most modern platforms are designed. They assume you're starting fresh. Airtable is a beautiful tool if you're building your workflow from scratch. Same with Monday. Same with almost every new-age SaaS suite from the last five years. The pitch is always some version of "come do your work inside our model."

If you have no existing workflow, that's a fair trade. But most small businesses and nonprofits don't have a clean slate. They have fifteen years of accumulated habits, spreadsheets, customer files, intake processes, institutional knowledge — most of it working well enough, some of it load-bearing. Forcing all of that into a SaaS platform's opinionated structure is a brutal migration tax. Once you've paid it, switching costs hold you there, because re-migrating is even worse.

So established businesses (which is to say, most businesses) end up paying twice. Once in subscriptions. Once in the friction of running their actual work through a system that was designed for somebody else.

What micro-software actually looks like

The alternative isn't building your own Salesforce. It's building exactly the thing you do, and nothing else.

Picture a six-person Asheville nonprofit with maybe 40 recurring donors. They don't need an enterprise CRM. What they actually need is a list of donors, renewal dates, a notes field, and a nudge when someone is overdue. Four features. A custom tool built against their real donor list costs basically nothing per month to run once it exists. The CRM that does the same thing badly, plus 395 other features they'll never touch, costs $50 to $200 a month forever. And that CRM holds their donor list as leverage at renewal time.

Or picture a small WNC guide service taking trip-planning intake. What they need is a form that captures the details that matter for their specific trips (party size, fitness level, gear rentals, dietary restrictions, pickup location), routes the booking to the guide leading that day, and sends the client an automated follow-up with the trail-specific briefing. One tool. One job. Not Monday plus Mailchimp plus Typeform plus Zapier plus a CRM stitched together with webhooks. An afternoon of AI-assisted work, and a file on a machine they control.

This isn't theoretical for me. I run a personal AI assistant, and I use a frontier cloud model (Claude) to power it. But the assistant's memory, file history, project notes, and institutional context all live on my own machine. The AI is a visitor. It shows up, does work against data I control, and leaves. I'm moving back to Asheville from Oregon this spring, and one of the things coming back with me is that local context system. Nothing fancy. A laptop, an organized folder structure, a storage box. The point isn't that every small business in town should build the same setup. The point is that the pieces you'd need to keep your own data while still getting the benefit of frontier AI are finally boring enough to put together.

What "local" actually means

A plain-language definition, because "local" has become a technical word that gets in the way. Local just means the data lives on a machine you own. Your laptop. Your office desktop. A small storage box plugged into your office network. It doesn't mean the AI has to run on that hardware too. In most cases the smart play is still to let a capable cloud AI do the thinking. What matters is that your data isn't sitting on somebody else's server, waiting to become a bargaining chip the next time that vendor raises prices, gets acquired, changes terms, or decides your subscription tier no longer qualifies for data export.

An analogy that might help: it's the difference between owning a house and renting a storage unit. If you own the house, you can still hire a contractor to come do work on it. They bring the expertise. They leave when the job is done. Your stuff stays where it is, under your control. If you rent a storage unit, the company running the unit decides the rent, the access hours, the terms, and what happens to your stuff if you miss a payment. Most small businesses have quietly rented a dozen storage units from a dozen different companies for their most valuable data — customer lists, project files, financials, years of email — without noticing they did it.

Working with AI in 2026 should feel like the first arrangement. Your data is in the house. When you need something done with it, a capable AI shows up, does the work, and leaves. You keep what matters. You pay for the work that got done, not for the privilege of hosting the house.

What to keep, what to replace

A useful triage. Some categories of SaaS have real, durable leverage and you should keep paying without much argument. Payments processing (Stripe and its peers have moats that aren't going anywhere soon). Email deliverability (getting mail into inboxes is a specialized discipline worth renting). Payroll and HR compliance (the legal leverage alone pays for itself). E-signature platforms. Basically anything where the vendor is doing expensive regulatory, deliverability, or fraud-prevention work on your behalf.

Other categories are worth reconsidering: project management platforms, internal CRMs, knowledge bases, intake forms, dashboards, small-team collaboration tools, documentation systems. This is where you're most likely to be paying for 400 features to use six, and where a purpose-built tool against your own data can do a better job for a fraction of the cost.

And then there's the thing you should own no matter what tools sit on top of it: your data. The customer list. The project history. The financial records. The institutional knowledge. Whatever software is in fashion this year, the underlying asset is the information your business has accumulated over time. That information belongs on a machine you control, with backups you own, and AI you can point at it when you need to.

The shift is already happening

At the enterprise level, the repatriation conversation is old news. Andreessen Horowitz's "Cost of Cloud" analysis framed it back in 2021. 37signals publicly documented roughly $2 million in first-year savings after leaving AWS, with total projected savings north of $10 million over five years. IDC's 2024 workloads survey found that roughly 80% of enterprises expected to repatriate at least some compute or storage in the following twelve months. The dollar stakes at that scale are big enough that CFOs have gotten involved and the conversation happens in public.

Small businesses and nonprofits haven't followed yet. The dollar stakes are smaller. The audits don't happen. But the underlying capability shift applies at every scale: infrastructure ownership is cheaper than renting, AI can work against your own data, and custom tools are no longer enterprise-only. The organizations that notice this first will save real money, move faster, and exit the decade owning the asset instead of renting it.

Where to start

A two-hour exercise, worth doing before your next big renewal. Pull the credit card statement for the last twelve months and list every recurring software charge on it. Beside each one, write down the two or three features your team actually uses. Cross off the features that exist mostly to justify the invoice. What's left is where the whole-suite economics might still make sense. Everything else is a candidate for replacement, with a smaller subscription, a custom micro-tool, or nothing at all.

That's the work we do at SiteSprint. Building the small, focused tools that replace the bloated suites, and helping Asheville-area businesses and nonprofits figure out what actually belongs in the cloud versus what should live on a machine they own. If your software stack has quietly eaten a bigger share of your operating budget than you realized, we should talk.