Every service business now runs on a stack of software, whether it was chosen deliberately or accumulated like sediment: something for the website, something for bookings, something holding customer records, something sending the emails, something for invoices, plus the spreadsheet that secretly runs everything. Each piece bills monthly, and each holds a different fraction of the business hostage.
The own-versus-rent question gets argued as ideology, and it should not be. It is a per-piece financial decision with a clear rule: rent what is commodity, own what is your edge. Here is how that rule lands on each layer of a service business stack, from a team that builds the owned kind for a living and still tells plenty of people to keep renting.
Builder subscriptions made sense when custom sites cost a fortune. Today the rent buys you a permanent monthly bill, a design ceiling, pages a search engine has to squint at, and an export button that produces rubble. We moved our own site off a builder and wrote about the escape. A fast, owned site is no longer the premium option; it is the sane default, and it is the asset every other layer sends traffic to.
Generic booking tools are fine at generic jobs. The moment your scheduling has real shape, crews, travel buffers, deposits, multi-stage jobs, approval steps, the generic tool starts dictating how you operate, and staff start maintaining workarounds in spreadsheets. That is the tell. When the workaround spreadsheet appears, the tool is no longer serving the operation; the operation is serving the tool. Booking is usually the first piece worth owning because it touches revenue directly: it is where the two-minute checkout lives or dies.
Big CRM subscriptions are built for everyone, which is why they fit no one and go half-used. Whether you rent or build here, one thing is non-negotiable: a complete export of your customer history, on demand, in a usable format. The relationships in that database are the business. When client records live inside the same platform as bookings and payments, follow-up stops being a separate discipline, which is exactly why we build them together.
Nobody should build their own email delivery infrastructure; that is solved, regulated plumbing, and renting it is correct forever. What you should own is the logic that decides who gets which message when, because that logic must read your operational truth: job finished, invoice paid, client gone quiet. Rent the sending; own the brain that pulls the trigger.
The platforms are theirs, Google will always own the reviews. The loop is yours to own: the automatic ask after delivery, the response workflow, the numbers that tell you whether it is working. Likewise analytics: the measurement plumbing is commodity, but the five numbers that steer your Monday deserve to live where you can see them, not across four logins.
Add the subscriptions the piece replaces, over five years. Add the hours of workaround labor at what you actually pay. Add, honestly, the revenue cost of the ceiling, the feature requests that will never ship because you are not the vendor's median customer. When that total clears the build cost, and at real volumes it does earlier than people expect, the build stops being an expense and becomes an acquisition: you are buying an asset that was previously landlording you.
The flagship example we operate: Book With NAV replaced an entire rented stack for real estate media, and 7,500+ shoots later, the platform is not a cost line. It is the business's chassis, and it belongs to the business.
List every tool, its yearly cost, what fraction of it you use, the workarounds it forces, and whether you could leave with your data tomorrow. Most owners have never seen that list in one place. It tends to reorganize priorities on sight.
A build done as one giant reveal is risky. A build done in phases, each shipping something the business uses, retires risk continuously, which is the only way we structure them.
Payment processing rails, email delivery infrastructure, video hosting, accounting cores. Regulated or hyper-commodity layers are rented forever by everyone sane.
The workaround spreadsheet, and the sentence we cannot do that because the software will not let us, said about something customers keep asking for.
Everything we write about, we build. Thirty minutes maps it to your business.