Off the shelf or custom software: the five-year sum that actually decides it.

If a product exists that does eighty per cent of what you need, and the missing twenty per cent is preference rather than principle, buy the product and change your process. It will be cheaper, it will be live next week, and somebody else will maintain it forever. That advice costs us work and it is still the right advice. A supplier who has never talked a client out of building something has not been paying attention.

What the original version of this guide lacked was numbers. Buyers are told to "compare total cost of ownership" without anyone showing them what the ownership actually costs. So this revision does the arithmetic: the published 2026 prices of the four categories of software a small UK business is most likely to be paying for, the five-year sum, and the same sum for a built application at fixed UK prices, including ours. Then a worked example with the one variable that changes the answer.

One rule for reading it. Subscription prices are quoted per seat or per user because that is how the vendors quote them, and per-seat pricing is the reason the answer changes with headcount. All figures exclude VAT unless the vendor states otherwise, and the dollar-priced tools are shown in dollars because that is how they bill.

These are the prices on each vendor’s own page on 14 September 2026. Three of the four categories have a clear UK-priced leader; CRM and booking are mostly billed in dollars. Promotional discounts are noted but the standard price is the one to plan on, because the promotion ends in six months and the software does not. Where a vendor’s page would not show us a full price list, the table records only what we could read and says so in the method section.

Read on the vendors’ own pricing pages on 14 September 2026. Dollar prices are as published; the vendors do not show a sterling equivalent on the pages we read. Prices change; the date at the top of this page is the date of the last check.

Five years is the right horizon because it is roughly how long a piece of business software lives before somebody wants to replace it, and because per-seat subscriptions compound in a way a one-off build does not. The build side of the table uses the fixed UK prices we publish, which sit inside the market bands in our cost guide: a single-process tool £4,000 to £10,000, a business application £10,000 to £20,000, a platform with mobile apps £20,000 to £30,000 and up.

Two assumptions are ours rather than published figures, and you should replace them with your own. Hosting for a small application is taken as £100 a month, which is generous for most; and a change budget of £2,000 a year is allowed for the features you will want once you have used it. A built application that gets no changes for five years is either perfect or abandoned.

We publish our prices because the comparison only works if the build side has a number: a single-process tool £4,000 to £10,000, a business application £10,000 to £20,000, a startup MVP £10,000 to £20,000, a platform with mobile apps £20,000 to £30,000 and up, all fixed against a written scope with an exclusions list, no day rates. Hosting and running costs are stated in the quote rather than discovered later. If a product already does what you need, the AI adoption audit at £2,500 to £6,000 will say so and name it, and the fee is credited against any build that follows.

Subscription rows assume the seat counts shown and no price rises, which flatters them. Build rows assume a fixed price, £100 a month hosting and £2,000 a year of changes. VAT excluded throughout.

Take a twenty-person services firm with five people who sell, everyone on a rota, one bookkeeper and a front desk that takes bookings. Buying the mid-tier of each product above, the five-year bill is £7,440 in sterling (Deputy Core for twenty at £5,100 and Xero Grow at £2,340) plus $13,320 in dollar-billed tools (Pipedrive Growth for five at $11,700 and Acuity Standard at $1,620). Even if you priced every dollar at a pound, which no exchange rate has done, the total is £20,760. Choose HubSpot Professional instead of Pipedrive and the dollar side rises to $30,120, a ceiling of £37,560 at parity.

Against that, one built business application that replaced the CRM and booking tools would cost £24,000 to £34,000 over the same five years on the assumptions in the table. On subscriptions alone, buying wins, and it is not close. This is the sum most articles stop at, and it is why the default is buy.

Now add the variable the subscriptions do not show. Suppose two people each spend five hours a week moving bookings into the CRM, chasing the rota against the bookings, and re-entering both into the invoicing run. Ten hours a week at £20 an hour, for forty-eight weeks, is £9,600 a year and £48,000 over five years. The buy route is now about £55,000 to £69,000 all in; the build route is £24,000 to £34,000 and removes the rekeying because the application does the whole flow. The answer flipped, and nothing changed except that somebody counted the hours.

That is the whole framework. If nobody rekeys anything between your tools, buy. If they do, cost the hours honestly before you renew, and be suspicious of any build quote that does not ask you for that number. The hours are also the figure most likely to be wrong in either direction: owners routinely guess low because the work is spread across several people in small pieces, and suppliers routinely guess high because it justifies the build. Have the people who do the work keep a tally for a fortnight before anyone quotes.

The rule of thumb is that a product covering eighty per cent of the need is a buy. The trouble is that buyers rarely agree on what the twenty per cent contains, so here is a test that works in practice. Write down the missing features. For each one, ask what happens on a Tuesday if you do not have it. If the answer is "someone does it a slightly different way", that is preference, and the product wins. If the answer is "we cannot deliver the service", "we break a regulatory duty", or "a person has to rekey it somewhere else", that is principle, and it goes on the build side of the ledger with a cost next to it.

Then look at the eighty per cent you are getting. Off-the-shelf products bundle features for their median customer, and you will pay for a great deal you never open. That is not a reason to build; a subscription you half-use is still cheaper than a build you fully use, as the table shows. It is a reason to buy the lowest plan that clears your principle list, rather than the plan the sales call recommends.

The one case where the 80/20 rule misleads is where the twenty per cent is the integration between products. A CRM that does everything and a rota that does everything, with a person in between, is two products at a hundred per cent and a process at sixty. The gap is never on either vendor’s feature list, because it belongs to neither of them.

Every business above about ten people runs on several products, and the money is lost in the joins. There are three ways to close a join. A person does it, which is the £9,600 a year in the worked example. A low-code connector does it, which is cheap at low volume and, as our automation cost guide sets out from the platforms’ own prices, climbs steeply once a workflow is busy. Or the join is built, either as a small integration between the two products you keep, or by replacing both with one application.

The small integration is the option most buyers overlook, and it is often the right one. Keep Xero, keep the rota tool, and build the £4,000 to £10,000 piece that moves the data between them and does the one thing neither product does. You keep the vendors’ maintenance on the ninety per cent and own only the join. Ask any supplier quoting you a full replacement whether this smaller version would do; if the answer is a straight no with no reasoning, get another quote.

Two integration costs belong in any comparison and rarely appear. The first is the product’s API: some vendors charge more for the plan that has one, and the plan you are on may not. The second is the data you will lose if you ever leave, which is why the questions in our supplier guide include who owns the export.

The original version of this guide listed three situations where building pays back. They hold, and the arithmetic above makes each one testable rather than rhetorical. In each case the question is the same: does this situation add a line to the buy column that the subscriptions do not show, and is that line bigger than the difference between the two five-year totals? If the honest answer is no, it is preference, and preference is a reason to configure a product rather than to commission one.

The expensive mistake is not buying, and it is not building. It is buying a product and then spending two years customising it until it is a bespoke system with none of the ownership benefits and all of the constraints. Configuration is fine; every product in the table expects it. Heavy customisation of somebody else’s platform, particularly through a specialist consultant who bends the product into a shape it resists, is the worst of both worlds, and it is where a great deal of money quietly disappears.

The tell is the invoice. If you are paying a consultancy a recurring fee to keep a product doing something it was not designed to do, add that fee to the subscription in the five-year sum and look at the total again. You have usually already made the build decision without admitting it, at a higher price and with the vendor able to break your customisation on any release.

The clean version of "buy now, build later" is different and sensible. Buy the product to learn what you actually need, use it plainly for a year, and then build only the part that turned out to matter. What you must not do in the meantime is customise the product so heavily that leaving it becomes a project of its own.

There is no official statistic that splits UK businesses into buyers and builders, and any article giving you one has made it up. What the government does measure is the direction of travel. The Department for Science, Innovation and Technology’s UK Business Data Survey 2026, fieldwork October 2025 to January 2026, found that 64 per cent of large businesses, 61 per cent of medium and 55 per cent of small businesses used a public cloud provider or a third party via software or a web solution, against 42 per cent of sole traders. Most business software is now rented, which is the buy default in the data.

The same survey found 41 per cent of businesses handling digitised data using AI-based technologies, rising to 82 per cent of large businesses; the Office for National Statistics, asking a narrower question of businesses with ten or more employees, put AI use at around 35 per cent in June 2026, up from around 12 per cent in late 2023, with only 10 per cent of adopters using it extensively. Cost was cited as a barrier by only 7 to 14 per cent of businesses across size bands. The barrier is not money; it is knowing what to do.

That matches what we see in audits. The businesses that benefit from building are rarely the ones that walk in wanting to; they are the ones that have bought three tools, hired a person to sit between them, and never added it up. The ones that walk in wanting to build usually want a product that already exists, and the most useful thing an audit does for them is name it, with the plan they need and the price of it, so the decision is made on the same sum as everyone else’s.

We read the pricing pages of HubSpot, Pipedrive, Deputy, RotaCloud, Xero, Sage, Calendly and Acuity Scheduling on 14 September 2026 and recorded the published price of each plan, the billing basis and any promotion, in the currency shown. Pipedrive and Sage returned only partial pages to automated reading, so for those we record only the figures we could see: Pipedrive’s annual per-seat prices and Sage’s starting price after promotion. Five-year totals are our arithmetic on those prices at the seat counts stated, with no price rises assumed.

Build prices are Augustova’s published ladder; hosting and change budgets are stated assumptions, not published figures. Adoption statistics are from DSIT’s UK Business Data Survey 2026 and the ONS Business Insights and Conditions Survey as reported in July 2026. Nothing in this guide is a client account. All prices exclude VAT unless the vendor states otherwise, and we will revise the tables as the sources change.