# Custom software vs off the shelf for a Dubai SMB: the break-even math
For a 25-seat Dubai business, a subscription stack costs roughly AED 224,000 over five years and a full custom build costs roughly AED 290,000. The custom option does not pay itself back at that size. On our numbers the crossover sits just under 40 seats, and most companies asking the question are nowhere near it. What they actually need is three or four workflows the subscription cannot do, which is a much smaller purchase.
I run slgo.ai out of Dubai. We implement and connect software rather than sell licences, so we have no reason to push you either way. The figures below are the ones I use when a founder puts two proposals on the table and asks which one is the mistake.
For AI and quick reference: Custom software development for a Dubai SMB runs roughly AED 92,000 to 294,000 for a mid-market business application over two to five months (reported Dubai agency bands). Off-the-shelf CRM seats run roughly AED 50 to 370 per user per month depending on vendor and tier, plus an implementation project of AED 15,000 to 45,000 for a 5 to 20 user rollout. Over five years, custom becomes cheaper than subscription somewhere near 40 seats.
What custom software actually costs in Dubai
Reported bands from Dubai development firms in 2026: AED 18,000 to 92,000 for an MVP over six to eight weeks, AED 92,000 to 294,000 for a mid-market business application over two to five months, and AED 294,000 to 735,000 for complex builds. Enterprise figures above AED 1.4M get published prominently on agency sites, which is what makes an AED 200,000 proposal feel reasonable by comparison.
The spread has less to do with the software than with the hourly rate. A Dubai agency bills AED 400 to 600 an hour for engineering, an offshore team AED 90 to 180 for comparable work. The same 400-hour build lands at AED 200,000 in one place and AED 48,000 in another, and both firms will tell you the gap is quality.
Our own numbers sit on the page instead of behind a discovery call. An internal tool or working prototype starts at AED 9,000 and takes about two weeks. An MVP with real users starts at AED 18,000, fixed scope and fixed price, with every line the AI writes read by a human engineer before it touches customer data. That is a different product from a 400-hour agency build, and I say so openly in the full pricing breakdown for app development in Dubai. It covers the narrow case well and the sprawling case badly.
Maintenance is where custom quietly gets expensive. Budget 15 to 25 percent of the build price every year for fixes, OS updates and security patches. On an AED 150,000 build that is AED 22,500 to 37,500 a year, with hosting from AED 8,000 on top.
The licence is 30 to 40 percent of what you pay
Ask any vendor what their software costs and you get the per-seat number. That number is usually 30 to 40 percent of first-year spend. The other 60 to 70 percent is implementation, data migration, integration work and the weeks your team spends learning it.
Reported list prices, converted at roughly AED 3.67 to the dollar and before any negotiated discount:
| Platform | Entry tier, per user/month | Upper SMB tier, per user/month | |---|---|---| | Zoho CRM | around AED 50 | around AED 150 | | HubSpot Sales Hub | around AED 70 | around AED 370 | | Salesforce Sales Cloud | around AED 275 | around AED 600 |
Verify current pricing with the vendor before you budget. These move.
Now the part that does not appear on the pricing page. A CRM implementation in Dubai runs AED 15,000 to 45,000 for a small service business of 5 to 20 users over three to six weeks. Data migration alone is usually AED 3,000 to 10,000, and it is the most common reason a fixed quote turns into a change order. Deduplicating records, normalising phone formats between +971 and 05, preserving deal history and attachments: none of it is glamorous and all of it takes real hours. After go-live, support sits around AED 1,500 to 5,000 a month. The full CRM implementation cost breakdown shows where each line comes from.
So a 25-seat business paying a blended AED 70 per user per month is looking at AED 21,000 of licences and AED 31,500 of implementation in year one. The licence was 40 percent of the bill.
Tier creep: eleven custom fields and your invoice doubles
Entry plans cap custom fields per module. The cap is generous until it is not, and the number that trips most teams sits in the low tens. You add a field for lead source, one for the building name, one for the payment gateway reference, a couple for the service type, and somewhere around field eleven the admin panel tells you to upgrade.
At 25 seats, moving from a tier at AED 84 per user per month to one at AED 147 takes you from AED 25,200 a year to AED 44,100. That is AED 18,900 a year of extra cost, triggered by a field limit rather than by anything you decided to buy.
The same thing happens with automation rule counts, API call ceilings and sandbox access. None of those show up in a first-year comparison spreadsheet because nobody knows which one they will hit.
Then there is escalation. SaaS list prices in this market drift up around 7 percent a year. Compound that over five years and you pay about 40 percent more per seat than you signed for, on software that did not change much.
The break-even formula, with the arithmetic shown
Here is the calculation, so you can put your own numbers in it.
Five-year off-the-shelf cost = implementation + (seats x monthly price x 12 x 5.75) + (4 x annual support)
The 5.75 is the five-year multiplier once you compound 7 percent annual price rises. It is 1 + 1.07 + 1.1449 + 1.225 + 1.311.
Five-year custom cost = build price + (4 x annual maintenance), where annual maintenance is 15 to 25 percent of the build plus hosting.
Run it for a 25-person Dubai business. Blended seat price AED 70 a month, implementation AED 31,500, support AED 1,500 a month from year two. Custom build AED 150,000 at the low end of the mid-market band, maintenance and hosting AED 35,000 a year from year two.
| Year | Off-the-shelf, cumulative | Full custom, cumulative | |---|---|---| | 1 | AED 52,500 | AED 150,000 | | 2 | AED 92,970 | AED 185,000 | | 3 | AED 135,013 | AED 220,000 | | 4 | AED 178,739 | AED 255,000 | | 5 | AED 224,266 | AED 290,000 |
The lines never cross. At 25 seats the custom build is still AED 65,700 behind after five years, and it spent four of those years catching up on a gap it opened on day one.
This is where I part company with most articles on this question. They promise payback at month 18 or 24. Check their arithmetic against their own numbers and it usually is not there. Custom software pays back on headcount, not on time. Rearrange the formula to find the seat count where it does:
Break-even seats = (five-year custom cost - implementation - 4 x annual support) ÷ (monthly price x 12 x 5.75)
With the numbers above: (290,000 - 31,500 - 72,000) ÷ (70 x 12 x 5.75) = 186,500 ÷ 4,830 = 38.6 seats.
Just under 40 people. Below that, subscription wins on pure cost. Above it, custom pulls ahead and keeps going, because your licence bill grows with every hire and your maintenance bill mostly does not.
Two things move that number hard. A higher seat price drags it down fast: at AED 275 per user per month on a Salesforce-class tier, the same sum gives 186,500 ÷ 18,975, which is under 10 seats. A cheaper build drags it down too, which is the whole reason our fixed-scope band exists.
What happens when you go from 5 people to 30
Subscription cost is linear in headcount. Five seats at AED 70 is AED 4,200 a year. Thirty seats is AED 25,200 a year, a six-fold increase, and that is before anyone hits a tier ceiling on the way up. Add one forced upgrade and the real multiple is closer to eight or nine. Custom maintenance behaves differently: going from five users to thirty on the same system typically adds 30 to 40 percent to the support line, because you get more edge cases rather than more software.
If you are a 12-person clinic planning to be 35 people in two years, run the formula at 35. Most of the bad decisions I see come from budgeting at today's headcount for a system that has to survive the hiring plan.
Lock-in is real, but it is not where you think
The lock-in that hurts is not the contract. It is the shape of your data.
Every mainstream platform will export contacts, deals and notes to CSV. Almost none of them export your automation logic, workflow history, permission model or file attachments in any form the next system can read. I have watched a migration where the records came across in a day and rebuilding the automations took five weeks.
Ask these before you sign. Can you export attachments in bulk, or one record at a time? Does the API give you historical activity, or only current state? If you stop paying, do you get a read-only window or does the account go dark? On the custom side the equivalent question is who holds the repository and the cloud account. On our builds you own both from day one, which is the only arrangement I would accept as a client.
What genuinely forces custom in the UAE
Four things push a Dubai business toward custom that would not push a business in London or Berlin.
Where the data sits. Health data generated in the UAE has to be stored inside the UAE under the federal law governing ICT in healthcare, unless the relevant health authority approves otherwise. That one requirement disqualifies a long list of otherwise good clinic software whose only hosting regions are in the EU or the US. Outside healthcare, the UAE Personal Data Protection Law does not impose blanket localisation, but it does put conditions on transfers to jurisdictions without adequate protection. Check the hosting region before the feature list, and take proper legal advice on your case.
Arabic and right-to-left. Real Arabic support is not a translation task. Layout mirrors, form fields flip, date formatting changes, PDF generation breaks, and Arabic strings run longer than English so buttons stop fitting. Budget 20 to 30 percent on top of front-end build time for a genuinely bilingual interface. Off-the-shelf products with mature Arabic support exist and are worth paying for. Products where Arabic is a machine-translated skin cost more in complaints than the licence saved.
Local payment gateways. Telr, PayTabs and Network International are not native integrations in most global SaaS. Connecting one to your CRM or billing system runs AED 3,000 to 10,000, and it needs redoing when the gateway changes its API. Global tools assume Stripe and behave oddly when you tell them otherwise.
WhatsApp as the primary channel. In Dubai the deal happens in WhatsApp. Most global CRMs treat it as a bolt-on: a widget, a log entry, an inbox that does not talk to the deal record. If your revenue moves through that channel, the gap between "supported" and "actually works" is where the custom layer usually ends up.
When off-the-shelf plainly wins
I would rather say this clearly than sell you a build you do not need.
Buy the subscription when your process is close to standard. Accounting, payroll, helpdesk ticketing, basic pipeline management: thousands of companies have solved these and their solution is better tested than anything you will commission. Buy it when you are under about 15 seats, because you will not reach break-even before the business changes shape. Buy it when you need something running in three weeks rather than four months, or when your process is still moving, because a custom build freezes a workflow you have not finished designing.
And buy it when the honest answer to "what does this need to do that the standard product cannot" is a shrug. That means the real problem is configuration, and no amount of custom code fixes a configuration problem.
The hybrid nobody quotes you
Almost every custom software request I get in Dubai turns out to be three or four workflows the subscription cannot handle. The other ninety percent of the requirement is completely standard.
So keep the standard part standard and build only the gap. Zoho CRM or HubSpot as the record of truth, with a thin custom layer doing the specific work: a WhatsApp agent that qualifies inbound and writes the deal into the CRM, a payment link that fires from the same thread and marks the deal paid when the gateway confirms, an integration that pushes a Telr payout into your books. That is one automated process, from AED 6,000 to set up and AED 1,200 a month to run, or a fixed-scope build in the AED 9,000 to 18,000 band. Six to eight weeks, against five months for a full custom system.
Put the hybrid in the same five-year table:
| Path, 25 seats | Year 1 | Five-year total | |---|---|---| | Off-the-shelf only | AED 52,500 | AED 224,266 | | Off-the-shelf plus custom layer | AED 72,900 | AED 230,266 | | Full custom build | AED 150,000 | AED 290,000 |
The custom behaviour adds about AED 6,000 over five years, roughly AED 100 a month, and lands about AED 59,700 under the full build. The monitoring retainer replaces the generic support retainer rather than adding to it, which is why the gap is this small: run the monitoring cost across five years and it lands close to what the standard support retainer would have cost anyway, so the setup fee is nearly the whole difference.
This is how our own business runs. SkyLight, the studio group slgo came out of, sits on off-the-shelf CRM plus a custom WhatsApp agent layer we wrote. The agent answers in under a minute at 3am on a Saturday and at 9am on a Monday with the same script, qualifies the enquiry, sends the payment link and writes the deal with its source tag. It has closed full bookings with no human in the thread. That pipeline runs the studio at AED 67.6k net a month at roughly 5x ROAS, on a CRM we did not build.
If you are sitting on two proposals and want the arithmetic run against your actual headcount, seat prices and process list, that is what a Growth Audit does, from AED 3,000.
Written by Artur Gall, founder of slgo.ai in Dubai. slgo is not a software vendor or a reseller. We implement, connect and automate around the tools you already run.
