The Lean Stack: Fewer Tools, One System, and Data You Actually Own — Peaceful Media
← Blog

The Lean Stack: Fewer Tools, One System, and Data You Actually Own

Most small businesses run a dozen subscriptions that barely speak to each other. Here is the eight-layer stack we build instead: what each layer does, what it consolidates, what it costs, and when you should not do it.

The eight layers of the lean stack drawn as a labelled stack on a warm studio backdrop: domain, edge and security, hosting, the site itself, data, payments, email, and analytics.

The short answerA lean stack is a small set of owned layers that together do the job a dozen SaaS subscriptions usually do: a domain, an edge/security layer, hosting, a custom-built site, a database, payments, transactional email, and analytics. The point is not to avoid tools but to need a handful instead of a dozen, and to own the customer data at the centre. It typically costs less per month than the subscriptions it replaces, but it is the wrong choice for a team that needs to change everything themselves without a developer.

Most of the small businesses we meet are running about a dozen subscriptions. A site on one platform, a funnel on another, email somewhere else, a form tool, a scheduler, a course host, a pipeline tracker. Each one arrived to solve a real problem. None of them share a customer record.

The monthly total is usually the smaller cost. The larger one is that nobody can answer a simple question (who bought twice, who opened the last three emails and never replied) without exporting spreadsheets and joining them by hand.

There is another way to build now, and it has only been practical for a few years. We call it the lean stack.

The eight layers

That is the whole thing. Most of it you set up once and never think about again.

LayerWhat it does
DomainYours, not rented inside a platform
Edge & securitySits in front of everything, absorbs attacks, serves globally
HostingRuns the site, scales without you managing servers
The siteCustom-built, so it does what your business does
DataOne database holding customers, orders, submissions
PaymentsTakes money, handles subscriptions and invoices
EmailSends the transactional mail your system triggers
AnalyticsTells you what actually happened

We build ours on Cloudflare at the edge, Vercel for hosting, Supabase for data, Stripe for payments, and Resend for email. Each is reviewed honestly on the toolkit shelf, including what it’s bad at.

The layers are not the interesting part, though. The interesting part is what one database in the middle lets you stop paying for.

What it consolidates

One build we did replaced a sales pipeline tool, an email automation platform, a form and assessment add-on, a course and video host, and a separate marketing site: five subscriptions, five logins, five places a customer’s details were half-stored.

To be clear: it is not that those tools are bad. We recommend tools in every one of those categories, and we are affiliates for several of them. A specialized tool doing work nothing else can do earns its subscription every month.

The claim is narrower and duller: maybe not this many. Four tools that each solve a fifth of the same problem, none of which share a customer record, is a tax you pay in fees and again in the hours somebody spends every week keeping them in sync.

What it costs

The build is real money up front, and the running costs are usually small enough to be surprising.

Every layer above has either a generous free tier or usage-based pricing that stays near zero until you have real volume. For most small businesses we build this for, the domain renewal ends up the largest recurring line on the bill. That is not a promise (your volume decides it), but it is the common case, and it inverts the usual arithmetic where the software costs more each month than the accountant.

The trade is straightforward. You spend once on the build instead of monthly on rent, and in exchange you carry the responsibility of owning the thing.

When you should not do this

We would rather talk you out of it than sell you the wrong build, so:

If your team needs to change everything themselves, don’t. Restructure pages, add a section, change how a form behaves, all without calling a developer. That is a real requirement, platforms serve it well, and a custom build makes it harder, not easier. This is the same reason we still build a lot of WordPress sites and think it is often the right call.

If you are still working out what you sell, don’t. Pre-product-market-fit, flexibility beats ownership. Rent everything, change your mind cheaply, and build when the shape of the business stops moving.

If it is working, don’t. A stack nobody is fighting is not a problem that needs solving. The businesses this helps are the ones already paying somebody to keep four tools talking to each other.

What it looks like when it works

Michel Peruch’s build is the clearest example we can show. In seven days it went live with a brand, a site, an AI business diagnostic at the front door, a mini-CRM staging every lead from New through Won, and email sequences carrying prospects forward, all run from one admin, on one database.

The diagnostic is the part worth noticing. A visitor answers fifteen questions and gets a scored report written for their answers. That is not a feature you can buy as a subscription; it exists because the quiz, the scoring, the report, and the pipeline are all the same system talking to itself.

We also did it to ourselves. This site runs on the same stack, and we wrote up the whole rebuild, including the part where our own site scored a C− and one of our fixes made things measurably worse.

Where to start

Not with a rebuild. Start by writing down what you pay for and what each one is for. Most people find two or three they forgot they were buying.

From there, two free things that need no account:

  • The Website Analyzer grades what you have now (speed, machine-readability, accessibility) in about thirty seconds.
  • The Launch Planner maps the sequence if you are heading toward a new or rebuilt website. There is a template for exactly that, alongside ones for other launch types.

And if you want the honest read on any individual tool before you commit to it, that is what the toolkit is for, including who should skip each one.

FAQ

Related questions

What is a lean tech stack?

A small set of layers that together cover what a business needs online: a domain you own, an edge and security layer, hosting, the site itself, a database, payments, transactional email, and analytics. Eight layers, most of which you never think about again once they are set up. The contrast is with a dozen separate subscriptions that each own a slice of your business and charge monthly for it.

How many SaaS tools does a small business actually need?

Fewer than most run, but more than zero. It depends on whether a tool is doing work no one else can do. A great tool for a specialized job earns its subscription. The waste is usually four tools that each solve a fifth of the same problem, none of which share a customer record, plus the hours somebody spends every week keeping them in sync.

Is it cheaper to build custom than to pay for SaaS?

Often, on a monthly basis, once the build is paid for. The layers in a lean stack have generous free tiers and usage-based pricing, so a small business frequently finds the domain renewal is the largest recurring line on the bill. The build itself is real money up front, which is exactly why this is a bad trade for a business that is still working out what it sells.

Should I use an all-in-one platform or separate tools?

All-in-one platforms are good at getting you live fast, and if one fits how you work, use it. They become expensive in a different currency later: your pricing page, your funnel logic, and your customer list live inside somebody else's product, and moving is a project. Choose the platform when speed matters most, and own the stack when the business has settled into a shape worth building around.

Who should not build a lean stack?

A business whose team needs to restructure pages, add features, and change how things work without calling anyone. That is a real requirement and platforms serve it well. Also anyone pre-product-market-fit: if what you sell may change in six months, rent your stack, stay flexible, and build when the shape stops moving.

Why does owning your customer data matter?

Because the customer record is the business. When it lives across four platforms, nobody can answer simple questions (who bought twice, who opened the last three emails and never replied) without exporting spreadsheets. When it lives in one database you own, those are queries, and the automations that act on them are yours to change.