The Future of IT Consulting: Trends to Watch in 2024

The strategy that never met the constraint

The sequence is familiar. Leadership commissions a technology strategy. The document is well argued, properly challenged and duly approved. Eighteen months later the operating model is unchanged, and the strategy is quoted in board packs as evidence that the thinking has been done.

Very little in the document was wrong. It was written by people who would not be in the room when the first real trade-off arrived: the one between the architecture as designed and the migration the finance team could actually absorb alongside year-end. Somebody made that call under pressure, without the authors, and the plan bent around it.

Almost every meaningful change in how consulting is bought is a response to that failure mode. It is not a fashion cycle but an economic correction: when advice stops at the recommendation, the residual risk transfers to the client at exactly the point the work becomes difficult.

AI became a business design question

The interesting artificial-intelligence conversation inside a scaling business is rarely about model capability. It is about which decisions the organisation is genuinely willing to change. Automating an existing process is a tooling exercise. Deciding whether that process should exist in its current form is a design exercise, and that is where the value concentrates.

What blocks the second exercise is usually not the technology. It is that nobody can say authoritatively where the customer record is mastered, which of four working definitions of revenue the system should learn from, or who is accountable when an automated decision turns out to be wrong. AI does not create data ownership problems; it converts them from irritating into expensive.

The practical consequence is unglamorous. In a mid-sized business, an AI and automation programme is mostly a data ownership, process and controls programme with a model at the end of it. Anyone presenting it the other way round is selling the easy half.

ERP stopped being a platform choice and became a boundary choice

Selection debates ask which platform is better. The more consequential question is narrower and harder: what is the ERP the system of record for, and where exactly does the boundary sit between it and commerce, warehousing, CRM, pricing and planning?

Get the boundary wrong and the business pays twice: once for customisation that fights the platform’s own model, and again for integrations whose real job is reconciling two versions of the same truth. Boundaries are also the least reversible part of the architecture, because reports, month-end routines and job descriptions grow around them within a year.

This is why integration and digital architecture has stopped being a technical detail and become a scalability constraint leadership should understand in its own terms. The question is not how many systems are connected. It is which system is allowed to be wrong about what.

Controls moved earlier, and that genuinely costs something

Designing access control, segregation of duties, logging and audit evidence into a programme from the outset is now the standard recommendation, and it is the right one. It is also slower at the start, and honest advisers say so rather than pretending governance is free.

The first twelve weeks of a governed programme look less impressive than the first twelve weeks of an ungoverned one. Fewer screens, more argument about who may approve what. What you buy with that patience is the avoided retrofit, and retrofitting controls after go-live means re-testing processes people already depend on, usually against an investor, customer or certification deadline that will not move.

The honest problem with continuous partnership

The case against project-shaped consulting is sound: projects end at the moment the organisation starts learning. But the case for continuous engagement carries its own failure mode, and it is rarely named by the people selling it. A long relationship can quietly become an annuity in which the adviser’s knowledge compounds and the client’s does not.

One test cuts through the rhetoric. Across the life of the relationship, is internal capability increasing? Named owners who hold their systems, run-books written by the people who operate them, decisions taken without escalation. If the honest answer after two or three cycles is no, the arrangement is a dependency, whatever the roadmap says.

An adviser worth retaining will define what “no longer needed here” looks like, and be comfortable with an engagement that shrinks by design.

Commercial models are the trend nobody puts on a slide

Outcome-based pricing is attractive in the abstract and fragile in practice. It requires a number that is measurable, attributable and not confounded by everything else the business did in the same quarter. Where those conditions do not hold, outcome pricing simply relocates the effort into arguing about attribution.

What tends to hold up is a hybrid: a fixed, accountable scope for the change itself, with a smaller variable element tied to something both sides can observe without dispute: a cut-over date met, a control tested, a manual routine retired. The pricing model matters less than the underlying signal, which is whether the supplier is prepared to be measured at all.

What to ask when you are buying

Most buying questions test whether a firm has done this before. The useful ones test whether it will still be there when the plan meets reality.

• Which of today’s recommendations will you still be accountable for in twelve months, and in what form?
• What did you advise a business like ours not to do? An adviser with nothing they declined is describing a catalogue.
• Which named individual is in the room at the trade-off, as opposed to at the pitch?
• What would have to be true in our business for this plan to fail? A firm that cannot answer has not modelled the risk it is asking you to carry.
• What do we own at the end: credentials, source, documentation, the ability to run it without you?

None of this lengthens procurement. It separates an adviser buying a project from one buying a result.