The custom software vs off-the-shelf decision is no longer a default. For a decade the answer was “buy whatever has the biggest market share,” because building a CRM or an ERP from scratch took two years and fifteen engineers. That logic broke in two places at once. Zylo reports SaaS vendors raised prices 10 to 20 percent in 2025, while IT budgets grew under 3 percent, and AI-assisted development has cut the cost of writing and maintaining code. The five-year math that pointed to “buy” in 2022 now points the other way for a specific class of workflows.
For a mid-market enterprise running a digital transformation UAE programme, this is one of the highest-leverage calls in the plan. Get it right and you build a moat where it counts and pay rent only where it does not. Get it wrong and you either own a maintenance trap or hand your differentiator to a vendor whose roadmap you do not control. This guide gives you the decision framework: total cost of ownership over five years, differentiation, time-to-value, integration, and lock-in.
Key Takeaways
- Decide per workflow, not per company. Buy commodity (payroll, general ledger, expense). Build where the software is the differentiator.
- Compare on five-year total cost of ownership, not the year-one sticker. Year one favours SaaS; year five often favours custom once renewal increases and integration cost compound.
- McKinsey found large IT projects run 45 percent over budget and deliver 56 percent less value than predicted. Custom done badly is worse than off-the-shelf done adequately.
- Three signals point to build: the workflow is your competitive moat, integration cost would dominate, or a vendor roadmap is blocking a change you need now.
- Score every option on strategic fit, optionality, and risk profile, not cost alone. A cheap SaaS contract with deep lock-in carries risk a TCO spreadsheet hides.
Total cost of ownership (TCO): the full five-year cost of a system, not just the licence or build price. It includes implementation, integration, training, hosting, support, renewal increases, and the internal staff time to run it.
Off-the-shelf (COTS): commercial off-the-shelf software bought as a licence or subscription, configured rather than built. The vendor owns the roadmap, maintenance, and compliance updates.
Why the Custom Software vs Off-the-Shelf Question Has Reopened
For most of the last fifteen years the answer was settled. SaaS won on speed to deploy, cost per seat, and maintenance burden. Custom software was reserved for the rare case where no commercial product fit at all. That was the correct call when a serious build meant 18 to 36 months and a large standing engineering team.
Two structural shifts reopened the question. The first is pricing. Major enterprise vendors pushed through broad price increases through 2025, often packaged as AI feature add-ons and renewal uplifts. Gartner forecasts worldwide IT spending to reach USD 6.08 trillion in 2026, up 9.8 percent, and attributes much of the rise in software cost to GenAI features now embedded in products enterprises already own, meaning a meaningful share of that spend absorbs price rises rather than buying new capability. When a renewal climbs 15 percent a year and the budget grows 3 percent, the gap compounds into a real number over five years.
The second shift is on the build side. AI-assisted development has made small teams faster, with experienced engineers shipping in days what used to take weeks, and it has lowered the ongoing maintenance burden through automated test generation, documentation, and refactoring. The work that once required a dedicated platform team is partly shouldered by tooling. The result is a moved equilibrium. The question in 2026 is not whether to build or buy. It is which specific workflows justify a build and which are still better served by a mature commercial product.
When Off-the-Shelf Is Still the Right Answer
Most workflows are not differentiators, and for those, off-the-shelf wins on TCO and time-to-value. General ledger accounting, payroll, expense management, and basic CRM contact management are categories where the commercial market has converged on best practice. Customisation buys you almost no competitive advantage there, while the maintenance burden of a custom build compounds for years. Building these is rarely justified.
Regulation-heavy categories also favour buying. Tax engines, regulatory reporting platforms, and AML transaction monitoring are domains where vendors invest continuously to keep rules current as the law changes. Replicating that compliance maintenance internally usually costs more than the licence fee, year after year. The exception is a genuinely unusual regulatory profile, such as a multi-entity structure in uncommon jurisdictions or a novel product type that commercial rules engines do not model.
Industries with mature, opinionated platforms (manufacturing ERP, hotel property management, retail point-of-sale) are typically faster and cheaper to deploy off-the-shelf, because the workflow assumptions inside those products encode decades of industry practice. A company building a custom point-of-sale system in 2026 is almost certainly making a mistake unless its requirements sit genuinely outside the standard model. The honest test for any of these: if your version of the workflow looks like everyone else’s, buy it.
When Custom Software Becomes the Better Bet
Custom software earns its keep in three situations. Recognising them is most of the decision.
The Workflow Is the Differentiator
A logistics company’s routing engine, a bank’s lending decision logic, a retailer’s personalisation model. These are places where the software is the competitive moat, not a support function. Off-the-shelf produces parity outputs at best, and parity is not a strategy. This is also where the build-vs-buy literature is most consistent: high-performing firms tend to build the applications that differentiate them and buy everything else. If a workflow is the reason a customer chooses you, owning the code that runs it is a defensible capital decision.
Integration Cost Would Dominate
Mid-market companies often run 30 to 60 SaaS products, each with its own data model and integration surface. The cost of stitching them together with middleware, ETL pipelines, and custom connectors can quietly exceed the cost of building one focused platform with native integration. Map the integration estate honestly, then compare the SaaS-plus-integration stack against a custom alternative over five years. Integration is where buy-side TCO usually hides its real weight.
Speed of Change Is the Constraint
SaaS roadmaps move on the vendor’s schedule, not yours. A feature you need next quarter sits on the vendor’s plan for next year, with no way to escalate. A custom system, built properly, lets the business ship a needed change in weeks. That matters most in regulated markets like the UAE, where new requirements arrive on government timelines that vendors absorb slowly. UAE e-invoicing is a live example: it runs on the Peppol 5-corner model in PINT AE format, transmitted through an Accredited Service Provider, with mandatory adoption from January 2027 for businesses above AED 50 million in revenue. When a deadline like that lands, the teams that can change their own systems move first.
How to Run a Disciplined Build-vs-Buy Evaluation
Most of these decisions go wrong because the evaluation is rushed and the criteria are vague. McKinsey’s study of more than 5,400 large IT projects found they run on average 45 percent over budget, 7 percent over time, and deliver 56 percent less value than predicted, with 17 percent overrunning badly enough to threaten the company’s existence. Software projects carry the highest risk in that data set. A structured evaluation is what separates a defensible build from one of those statistics.
Start by defining the workflow in writing before talking to any vendor or developer. Document current cost, current pain points, and requirements over a five-year horizon. The act of writing it down almost always surfaces requirements the team had never articulated, which is exactly the gap that overruns feed on. Across ERP implementations, inadequate requirement gathering and system design trace to roughly 60 percent of failed projects, and the same root cause sinks custom builds.
Then run a parallel evaluation. Get a serious quote from at least two relevant SaaS vendors, including implementation, integration, and five-year licensing at projected user volumes, with the renewal increase modelled rather than assumed flat. Get a serious estimate from at least two engineering teams, including build, hosting, and an ongoing maintenance budget. Compare on five-year TCO, never on the year-one sticker, because year one almost always favours SaaS while year five often favours custom once renewal and integration costs accumulate.
Finally, score on three dimensions a pure cost model misses. Strategic fit: does the workflow differentiate the business. Optionality: does the choice preserve future flexibility or lock you in. Risk profile: what happens if the vendor raises prices, gets acquired, or sunsets the product. A SaaS contract with deep lock-in and a vendor under acquisition pressure can look cheap on a spreadsheet while carrying strategic risk the spreadsheet never shows.
Why Custom Software Has Become More Viable in the UAE
Specific factors have made custom software development in the UAE more viable in 2026 than even two years ago. The talent market has matured, with engineers in Dubai and Abu Dhabi who have shipped at scale for regional banks and global product companies. AI-assisted development has narrowed the productivity gap between elite global teams and competent regional ones for well-scoped work. And cloud infrastructure costs have fallen as Microsoft, Google, AWS, and Oracle all opened production cloud regions in the country, removing the latency and data-residency friction that used to push workloads offshore.
Regulation and language requirements also tilt toward custom in the UAE context. Arabic-first user experiences, integration with national identity infrastructure such as UAE PASS, and compliance with the Personal Data Protection Law (Federal Decree-Law No. 45 of 2021, with full compliance expected by January 2027) all require localisation that global vendors often retrofit poorly. An app development company UAE that builds around these requirements from the start ships faster and cheaper than a global product that bolts them on. The pattern holds across our work: where the regulation is local and specific, the custom path is frequently the lower-cost one over five years.
Avoiding the Most Common Custom Software Pitfalls
Custom done badly is worse than off-the-shelf done adequately, and the failure modes are predictable. Scope creep is the most common: a six-month project becomes an eighteen-month one because requirements grew without budget recalibration. The same dynamic shows up in ERP, where cost overruns compound the longer a project runs; Panorama Consulting research puts average ERP overruns near 189 percent. The fix is rigorous scope discipline at kickoff and a real change-control process that puts every new requirement through the same evaluation as the original ones.
Maintenance debt is the second. Custom systems built without documentation, automated tests, and clear ownership become unmaintainable within two to three years, at which point the business faces a worse decision than the one it started with. Budget for maintenance from day one, treat documentation and tests as production deliverables rather than nice-to-haves, and assign long-term ownership before the build team disbands.
Vendor dependency is the third, and it can be worse than SaaS lock-in. A custom project built with a single partner who holds all the institutional knowledge is its own trap. Insist on full code ownership, comprehensive documentation, and a transition plan another team could follow. Custom software vs off-the-shelf is not a binary, forever choice. It is a series of sequential bets, and good architecture preserves the option to revisit each one.
Making the Right Call
The companies that get build-vs-buy right have no fixed bias. They evaluate each workflow on its own merits using a structured framework that weighs strategic fit, five-year total cost of ownership, and risk profile. They build where the workflow is the differentiator, where integration cost would otherwise dominate, or where a vendor roadmap is blocking a change they need now. They buy where commercial products are mature and customisation would add no advantage.
The right posture is portfolio thinking. Treat the application estate as a deliberate mix of custom and commercial components, optimise each for its specific job, and revisit the mix every two to three years as costs, capabilities, and competitive dynamics move. The AI era has shifted the equilibrium toward more custom than was justified in 2022, but it has not removed the case for off-the-shelf in the categories where commercial products genuinely dominate.
Frequently asked questions
How much does custom software cost versus an off-the-shelf licence?
There is no fixed number, because cost is driven by the workflow, the integration surface, the user volume, and the compliance requirements. The useful comparison is not custom price against licence price, it is five-year TCO against five-year TCO, with the SaaS renewal increase modelled and the integration and internal-staff costs on both sides counted. Off-the-shelf usually wins year one; custom often wins by year five for differentiating workflows. The honest way to size it is a parallel evaluation with real quotes. Book a discovery call and we will scope the specific workflow before either path is recommended.
How long does a custom software build take?
For a well-scoped workflow, a first production release is typically a matter of weeks to a few months rather than the year-plus that the old model assumed, because AI-assisted development has compressed delivery for focused builds. Larger, multi-system platforms take longer. The variable that moves the timeline most is requirements clarity at kickoff, not engineering speed. Projects that skip the documentation step are the ones that slip, which is why we insist on it before quoting a timeline.
What is the ROI of building custom software?
ROI depends on the inputs, so a fixed multiple would be dishonest. The return comes from three places: avoided SaaS renewal increases over five years, avoided integration cost where a custom platform replaces a stitched-together stack, and the value of owning a differentiating workflow outright rather than renting parity from a vendor. For commodity workflows the ROI of building is usually negative, which is the whole point of deciding per workflow. We model the specific numbers for your case during the assessment.
Does AI change the build-vs-buy decision?
Yes, in two directions. AI-assisted development lowers both the build cost and the maintenance burden of custom software, which moves more workflows into the “build” column than two years ago. At the same time, AI features are a leading reason SaaS prices are rising, which raises the five-year cost of staying on commercial products. The net effect is that the threshold for building has dropped, but the decision is still per workflow, not a blanket shift to custom.
How do we avoid getting locked into a custom-software vendor?
Lock-in on the custom side comes from a single partner holding all the institutional knowledge. Mitigate it contractually and architecturally: insist on full code ownership, require comprehensive documentation and automated tests as deliverables, and ask for a transition plan another team could pick up. Done this way, a custom build preserves more optionality than a deeply integrated SaaS contract, because you control the code and the data model rather than a vendor’s roadmap.
Can we mix custom and off-the-shelf in the same system?
That is the recommended approach, and it is what most disciplined estates look like. Buy the commodity components, build the differentiating ones, and connect them with clean integration. The goal is not ideological purity toward one model, it is the lowest five-year TCO and the most optionality across the whole estate. Reviewing that mix every two to three years keeps it aligned as prices and capabilities change. See how we approach this across engagements on our case studies page.
Run an honest build-vs-buy assessment before you commit
We scope the workflow, model five-year TCO on both paths, and recommend custom or off-the-shelf based on the numbers, not a sales target. Reach us at hello@thekentro.com or book a call.
