Straight answers, nothing gated

What software actually costs in the UAE

Published pricing is a starting position. These are the numbers that decide the bill.

How much does CRM implementation cost in the UAE?

For a 10 to 50 seat company, licences commonly run US$25 to 150 per user per month depending on tier, and first-year implementation frequently costs more than the licences. The driver is data migration and integration complexity, not seat count, which is why two companies of the same size get quotes that differ by a factor of three.

How much does ERP implementation cost in the UAE?

Budget one to three times the annual licence spend in year one. Process complexity and data migration drive it, not headcount. A finance-only rollout and a full manufacturing deployment are different projects wearing the same name.

How much does an AI consultant cost in Dubai?

Decision and scoping work is usually a fixed fee against a defined question rather than a day rate. Build work should be quoted only after the workflow is specified. A number produced before anyone has mapped the process was invented to fill a proposal.

Why is the implementation quote higher than the software?

Because the software is a product and the implementation is a project. Migration, integration, configuration, testing and training are people-time, and people-time does not have a price list. This ratio is normal. It is also the number to negotiate, and almost nobody does.

What is a fair discount to expect on enterprise software?

It depends far more on timing and term than on how hard you push. Vendors discount against quarter and year end, against multi-year commitments, and against a credible alternative. Asking for a discount with none of those in play gets you the standard concession everyone gets.

The long version: Why the licence is the smaller invoice.

How long things take

Timelines quoted in a sales meeting describe installation. These describe the project.

How long does it take to choose a CRM?

Three to six weeks from discovery to a decision, mostly determined by how many people have to agree and how fast vendors schedule demos. Faster than that usually means the shortlist was inherited rather than built.

How long does an ERP implementation take?

A finance-only rollout commonly runs two to three months. Full scope across finance, inventory, manufacturing and payroll typically runs six to nine, because competent programmes overlap workstreams rather than running them end to end. Under two months for full scope describes a software installation.

How long before a renewal should we start work?

Twelve months to build the calendar, six to decide direction, three to open the conversation. If your notice window is 90 days and you start thinking about it at 60, you are confirming a price rather than setting one.

How long does an AI project take to reach production?

Two to four weeks for a decision engagement. Six to twelve for a first production workflow. The variable is rarely the model. It is integration surface, error handling, and agreeing what the system should do when it is unsure.

The long version: ERP selection and implementation.

Before a vendor demo

The demo is the vendor's best hour, rehearsed. These questions move it onto your ground.

What should we ask in a software demo?

Ask them to run your process, not theirs. Bring three real scenarios, including the awkward one, and ask to see each performed live. Then ask what it cannot do. A vendor who cannot name a limitation is either new to the product or not being straight with you.

How do we stop a demo becoming a feature tour?

Send the scenarios in advance and say plainly that the session is scripted against them. Good vendors welcome it, because it shortens their sales cycle too. A vendor who insists on their standard deck first is telling you something about how the implementation will go.

Who should be in the room for a software decision?

Whoever will use it daily, whoever owns the budget, and whoever will be blamed if it fails. Those are often three different people and they frequently have not agreed on the problem. Finding that out during evaluation is far cheaper than finding it out after signature.

Should we take the vendor's reference calls?

Take them, and treat them as marketing. Then ask for a customer who left, or one in your industry at your size that the vendor did not choose. What comes back, including a refusal, tells you more than the curated call does.

The long version: How to buy enterprise software without regretting it.

Signs you are overpaying

None of these require an audit to spot. All of them are common here.

How do we know if we are paying for unused licences?

Pull seats provisioned against users active in the last 30 days, per system. If you cannot produce that number, that is the finding. Industry research puts unused or underused licences at roughly a third of SaaS spend in the average organisation.

We have the same tool in three group entities. Does that matter?

It is one of the most reliable sources of overspend in GCC groups. Three entities buying separately means none has the volume to negotiate and nobody is looking at the combined number. Consolidating the commercial relationship rarely requires disrupting the operational one.

Our contract is in US dollars but our budget is in dirhams. Is that a problem?

It means a flat renewal is still a movement on your books, and multi-year dollar commitments carry an exposure nobody priced. Worth raising at renewal rather than discovering at year end.

Is a percentage-of-savings consultant a good deal?

It looks free and it is not. That model pays the consultant to cut aggressively whether or not the cut hurts the business, and gives them a reason not to tell you when a tool is worth keeping at full price. A fee aligns the advice; a share of savings aligns the cutting.

The long version: You are paying for seats nobody sits in.

When not to buy

The cheapest recommendation is often the one nobody is selling.

Do we need a CRM at 15 people?

Often not yet. The question is whether deals are being lost to forgetfulness rather than to price or product. If the real problem is that nobody has agreed what a qualified lead is, a CRM will encode the confusion rather than remove it.

Do we need an ERP, or better accounting software?

If the pain is reporting and month-end close, modern accounting software plus a reporting layer is usually cheaper, faster and sufficient. ERP earns its cost when inventory, production or project costing must share records with finance in real time.

Should we build a custom AI tool or use what we have?

Exhaust the AI features already bundled into software you pay for first. They are the cheapest route and they improve without you doing anything. Build where the workflow is specific to how your business operates and no vendor covers it.

The long version: AI consulting, including what not to build.

Every answer above changes once it meets your actual stack.

The long versions: CRM selection, ERP selection, SaaS spend audits and AI consulting. Longer pieces sit in the writing, and the three ways an engagement starts are on services.