A programme manager gets the green light: replace an order management platform that has grown in layers for twelve years and now talks to the ERP, a warehouse management system, three carriers across two countries and an e-commerce storefront. The resource plan has a line that just says “analyst”. Procurement asks what kind. Someone from IT says “systems analyst, because of the integrations”. Someone from the business says “business analyst, because the processes are a mess”. An outside adviser answers diplomatically: “You need both.”

That last answer sounds the safest and is the most expensive. Two analyst roles for the full length of a project is an annual line item a programme manager will struggle to defend to the CFO unless they can say what each person will be doing in month three. And in month three it often turns out that one of the analysts has run out of work, or that both are doing the same job under different names.

This article defends a position that may seem contrarian, including against what we have written ourselves before: both kinds of analysis always have to happen, but who you hire first is decided by where the project’s biggest unknown sits, not by job titles or by a “both, equally” rule. On projects that change an existing systems landscape, which describes most projects in established companies, the systems analyst should usually come first, and a good part of the business analysis should sit with a product owner on the client side. Below is how to recognise which case you are in, how to put numbers on it and how to test candidates.

Why does “you need both” not help a programme manager?

Because it answers a different question. “You need both” is true at the level of work to be done: someone must understand the business problem and someone must specify the system that solves it. Our earlier piece on business analysis versus systems analysis described both roles and how they collaborate, and that description still holds. But a programme manager is not asking which kinds of work exist. They are asking whom to hire in week one, for how long and at what cost.

To that question, “both” is not an answer but a postponement. Two people from day one means double cost in the phase where the project knows least about what it actually needs. It also creates a coordination problem: two people interview the same stakeholders, each with their own questions, and by the third meeting the stakeholders start giving shorter answers. And the boundary between the two roles often sits in a slightly different place in each analyst’s head, so the requirements documents overlap.

The second weakness of the advice is the assumption that business analysis must be done by an external analyst. Many organisations already have a person who knows the processes, has authority to decide and owns the business outcome: a product owner, a department head, a process owner. If that person has time for the project, part of the business analyst’s job is already staffed. Hiring a full-time role next to them means paying twice for the same “why” question.

The third weakness is the most practical. “Both” says nothing about sequence, and sequence determines what the project knows after the first month. If the biggest risk is that nobody knows how the old platform really behaves and what data leaves it, a month of process workshops will produce beautiful future-state maps that break against the first integration.

How do the standards separate business analysis from systems analysis?

Start with definitions, because the market uses both titles loosely. The International Institute of Business Analysis (IIBA), in its BABOK Guide, defines business analysis as the practice of enabling change in the context of an enterprise by defining needs and recommending solutions that deliver value to stakeholders. BABOK organises the work into six knowledge areas: business analysis planning and monitoring, elicitation and collaboration, requirements life cycle management, strategy analysis, requirements analysis and design definition, and solution evaluation. Note the starting point: change in an organisation, not a system.

Systems analysis has no single BABOK equivalent, but its craft is well described. The International Requirements Engineering Board (IREB) runs the CPRE certification programme for requirements engineering. Its foundation level covers eliciting, documenting, validating and managing requirements, and its higher levels specialise in areas such as requirements modelling and requirements engineering in agile settings. The US Bureau of Labor Statistics describes computer systems analysts as people who review an organisation’s existing computer systems, propose ways to make them more efficient and use techniques such as data modelling to plan new systems. BLS reported a median annual wage of $103,790 for the occupation in May 2024. Starting point: the system and its surroundings.

The cleanest dividing line comes from ISO/IEC/IEEE 29148, the international standard for requirements engineering. It distinguishes several requirements documents at different levels: the business requirements specification (BRS), the stakeholder requirements specification (StRS), the system requirements specification (SyRS) and the software requirements specification (SRS). The first two are the natural output of a business analyst. The last two are the natural output of a systems analyst. Our guide to the software requirements specification shows what a good document at the lowest of these levels looks like, and the glossary entry on requirements analysis covers the basic vocabulary.

In short, a business analyst answers “why” and “what must change in the organisation”, while a systems analyst answers “how must the system behave” and “how must it work with the rest of the landscape”. The line is not sharp, because stakeholder requirements flow into system requirements. But it is clear enough to decide which kind of work your project has more of, and which is harder.

Why does the job title on a CV settle nothing?

Because titles drift across companies and countries. In Central Europe the combined “business/systems analyst” profile is very common, and our own available experts page lists one. In the UK and the Nordics “business analyst” often covers what a German or Polish employer would call a systems analyst. In the US, “systems analyst” can mean anything from an infrastructure-minded generalist to a requirements specialist inside an ERP programme. Behind any of these labels you may find a former developer who reads an API specification fluently but struggles in a workshop with senior management, or a former operations lead who runs stakeholder sessions brilliantly but leaves the data model to the architect.

Title inflation adds noise. In some organisations a “systems analyst” writes detailed use cases from requirements someone else has already agreed. In others it is the person who designs integration contracts and argues with the architect as an equal. Likewise, a “business analyst” may be a strategy facilitator or someone who turns meeting notes into backlog items. A CV title will not tell you which one you are hiring.

So reverse the order of thinking. Instead of starting from the job title, start from the work that is hardest on your project and choose the profile to match it. What the contract calls the role is secondary. What matters is that the person is strong exactly where the project is weak. It also sharpens the brief to vendors: “business/systems analyst, five years’ experience” gets you a dozen CVs that are hard to compare, while “someone who in month one maps the data flows between the order platform, the ERP and the warehouse system and proposes contracts for the new interfaces” gets you fewer CVs, each of which you can actually evaluate.

Where does your project’s biggest unknown sit?

This is the question everything else depends on. Every project has two axes of uncertainty. The first is about the problem: do we know what the organisation needs, who benefits, how processes will change and how success will be measured? The second is about the solution: do we know how the current systems landscape works, what data we have and in what quality, which interfaces must be built or changed, and what performance, security and regulatory constraints apply?

Here is a simple test. Ask the team five questions about the problem and five about the system. Problem questions: Who owns the business outcome? Can we draw the current process in one diagram that every department agrees with? Do we know the success metric and its current value? Do stakeholders agree on priorities? Does the change affect one department or several? System questions: Is there current documentation for the existing interfaces? Do we know which system is the source of truth for which data? Do we know volumes and peak loads? Do we know who, on each external system, can change an interface and how quickly? Are the non-functional requirements written down, or do they live only in people’s heads?

Count the “we don’t know” answers. If they cluster on the problem side, start with a business analyst, even if the project is technically complex. If they cluster on the system side, start with a systems analyst, even if the project sounds like a business initiative. If both sides are full of unknowns, you have a high-risk project where two people from the start may genuinely be needed, but then it is worth running a short discovery phase first rather than staffing up fully on day one.

The test has limits. Answers depend on whom you ask: the IT team will underrate business unknowns, and the business will underrate system unknowns. Ask both sides separately and compare. The disagreement is itself information. If the business believes the process is perfectly understood while IT says nobody can describe it, the analyst’s first job is to resolve exactly that gap.

When should the business analyst come first?

When the biggest risk is building a correctly functioning system that solves the wrong problem. Four situations are typical.

A new product or a new line of business. The company is entering a market it has not served before, or building a digital service for a new customer segment. There is no process to map, because the process does not exist yet. You need someone who will run conversations with prospective users, separate hypotheses from facts and help decide what goes into the first release. Integrations are usually simpler at this stage, because a new product touches few systems at first.

A process change spanning several departments. The project changes how sales, customer service and finance work at the same time. Each department has a different picture of today’s process and different expectations. The main work is agreeing one picture and one set of priorities, and that is a business analyst’s craft, not a systems analyst’s.

No product owner with real authority. If nobody on the client side can make scope and priority decisions, the business analyst often ends up carrying part of that role, preparing decisions for a steering committee. It is not ideal, but it is common. On such a project, hiring a systems analyst first means getting very precise specifications of requirements nobody has approved.

A regulatory change whose interpretation is open. A new reporting obligation, or a regulation such as the EU’s Digital Operational Resilience Act (DORA) for financial entities, first requires deciding what exactly the organisation must do differently. Before any system requirement exists, someone has to translate the rule into process, together with legal and compliance.

In none of these cases does the systems analyst disappear. They join later, once the problem is described, usually when the team starts designing integrations and the data model.

When should the systems analyst come first?

When the business problem is relatively clear and the risk lies in systems, data and integration. This is more common than it looks, because most projects in long-established companies do not build something from scratch. They change something that already exists.

Replacing or modernising a legacy system. The company knows what the system does, because it has used it for years. What it does not know is exactly how: which rules are buried in stored procedures, which fields are actually used, which undocumented data exchanges run overnight. The first job is reconstructing the current system’s behaviour. Our legacy system modernisation checklist walks through that assessment; on an incremental replacement, the systems analyst is the person who sets boundaries and contracts for each slice.

An integration project. Connecting a new platform to the ERP, a warehouse system, a payment provider and a data warehouse. The business problem fits in one sentence (“an order must flow through without being re-keyed”), but every connection has its own formats, limits, maintenance windows and owners. A full-time business analyst has little to do after week two.

Data migration. Moving customer, contract or transaction history into a new system. The questions are which fields map to which, what to do with records that fail the new rules, and how to prove nothing was lost. Under GDPR it also means knowing which personal data you are moving and why you still need it. This is work on data and models, not on processes.

Demanding non-functional requirements. The system must handle specific peaks, meet security requirements or availability targets. Such non-functional requirements have to be written down, quantified and turned into architectural decisions, as our guide to functional and non-functional requirements explains. The systems analyst is the architect’s natural partner in this work.

Business analysis is still needed on these projects, but it often fits into a few workshops at the start and regular priority reviews. If the client has a competent product owner, they can run those sessions, with the systems analyst asking the challenge questions. This is the core of our thesis: on projects that change existing systems, the technical unknown is usually larger than the business one, and the reflex of “business first, technology later” puts the risk in the wrong order.

When is one person enough, and when is it a trap?

An analyst who covers both roles makes sense on many projects. It works when three conditions hold. First, the project has one main business area and one decision-maker. Second, there are a handful of integrations with known, documented interfaces. Third, the delivery team is small, up to about eight people, so one person can keep up with refining requirements and answering developers’ questions.

In those conditions a single person with experience on both sides is often better than two specialists, because nothing is lost in hand-offs. The same conversations that establish the goal also settle integration details. Stakeholder fatigue drops too: they talk to one person, not two.

The trap appears when the person is genuinely strong in only one half. An analyst with a technical background will cut process conversations short and jump to the data model, because that is where they feel at home. An analyst with a business background will write requirements like “the system should integrate with the ERP” and leave details “for later”. Either way the weaker half of the analysis does not vanish. It moves to where it is most expensive: into developers’ heads during implementation, or into acceptance testing.

How do you spot it early? Watch which questions keep coming back from the team. If developers ask “why do we need this feature” and “what happens in this edge case”, business analysis is missing. If they ask “where does this data come from”, “what format” and “what do we do when the ERP doesn’t respond”, systems analysis is missing. Two or three sprints of that observation tell you more than the best CV.

Who does the analysis when there is no analyst?

Someone always does. That is the single most important point for a programme manager weighing a project with no analyst, or one instead of two. Analysis is not optional work. It is work someone will do either deliberately or by accident. The only questions are who, and at what cost.

Missing business analysis is usually absorbed by the product owner, the project manager or the developers themselves, who make small decisions about system behaviour during implementation. Each decision is reasonable on its own. A few hundred of them add up to a system that does something slightly different from what the business intended, and that surfaces in acceptance testing or after go-live.

Missing systems analysis is absorbed by the architect, the tech lead or the most experienced developer in the relevant area. On a small project that can be enough. On a large one it means your most expensive people spend their time pinning down field formats and chasing owners of external systems instead of designing and building.

PMI’s 2014 Pulse of the Profession report on requirements management, based on a survey of more than 2,000 practitioners, gives a sense of scale. PMI found that 47% of unsuccessful projects failed to meet their goals because of inaccurate requirements management, and that 5.1% of every dollar spent on projects and programmes was wasted for that reason. The data is more than a decade old and self-reported, but it shows the order of magnitude: gaps in analysis are not a side cost, they are one of the main reasons projects fail.

The conclusion for a programme manager: deciding not to hire an analyst does not save analysis work. It moves it onto people with other jobs and into later phases where fixes cost more. Sometimes that is a deliberate and sensible choice. But it belongs in the project plan, with the name of the person who will take the work on.

What does a decision matrix for typical projects look like?

The table below pulls the reasoning together. It does not replace the unknowns test, but it lets you check quickly whether your instinct matches the typical distribution of risk.

Project typeWhere the biggest unknown sitsHire firstWhen to add the second roleCan one person cover it?
New digital product for a new customer segmentProblem: who, why, what goes into release oneBusiness analystWhen the first integrations and data model appearOften, with a small team
Process change across several departmentsProblem: agreeing process and prioritiesBusiness analystOnce the target process is approvedRarely, if processes are complex
Legacy system replacementSystem: hidden rules and dataSystems analystFor short target-process workshopsYes, with a strong product owner
Integration project (ERP, warehouse, payments)System: interfaces and system ownersSystems analystUsually never; business analysis sits with the product ownerYes
Data migrationSystem: mapping and data qualitySystems analystFor business-rule decisions on inconsistent dataYes
Regulatory change with open interpretationProblem: what the rule means for processBusiness analyst with legal inputOnce the interpretation is signed offDepends on how many systems are affected
Large programme with several workstreamsBoth axesShort discovery, then both rolesFrom the programme’s startNo

Two notes on the table. First, “hire first” does not mean “hire instead”. It means: whose work in the first month reduces uncertainty most. Second, the last column assumes a product owner on the client side with time and authority. If there is none, shift every answer one step towards two people.

What does an analyst cost, and how do you know it pays off?

Start with rates. According to ARDURA’s 2026 benchmark, set out in our article on body leasing rates in Poland, a mid-level business analyst costs EUR 205–300 per day and a senior EUR 300–410 per day. The benchmark has no separate row for systems analysts. As a reference point, we suggest assuming their rate falls between the senior business analyst range and the senior backend developer range, which in the same benchmark is EUR 335–485 per day. That is an estimate, not a price list entry: it depends on whether the role includes data modelling and designing integration contracts, or mostly writing use cases.

An illustrative example. A senior business analyst at the middle of the range, EUR 355 per day, over 21 working days costs about EUR 7,450 a month. Four months of their work comes to roughly EUR 29,800. A second analyst for the same period doubles that. On a project with a budget of several million euros that is a rounding error; on a project of half a million it is a sum that needs a reason.

How do you judge whether it pays off? Not by comparing the analyst with a developer’s rate, but with the cost of the mistake the analyst prevents. Ask three questions. What does one week of the whole delivery team cost? How many such weeks will the project lose if one integration is designed wrongly and has to be rebuilt? What does a one-month delay to go-live cost in lost business benefit? If a team of five costs around EUR 40,000 a month, one month of rework costs more than four months of an analyst. That is arithmetic a CFO will accept, because it can be checked.

The same arithmetic works the other way. If the unknowns test shows that almost everything on the problem side is clear, a second analyst prevents no expensive mistake and their role is hard to defend. That money is better spent on another developer or tester.

Should you hire an analyst permanently or bring one in through staff augmentation?

Analysis work has a distinctive shape over time. There is most of it at the start of a project and at every major scope change, and less in the middle of implementation, when requirements are written and the team is delivering. The shape differs between the two roles: business analysis usually tails off faster, while systems analysis lasts longer, because every new integration and every slice of a replaced system needs fresh work.

That leads to a practical recommendation. If the company runs many projects that change the same systems, a permanent systems analyst who knows the landscape and its history is an investment that compounds. Knowing how the ERP really behaves, and who at the vendor changes its interfaces, is hard to rebuild and is lost with every departure. A permanent business analyst makes sense when the company changes its processes regularly and needs someone who knows the organisation and its politics.

In other cases a flexible model is more sensible. With staff augmentation you bring the analyst in for the period their work is needed and release them when the shape of the work changes. At ARDURA the notice period is 30 days, and if a specialist is not the right fit within the first two weeks, we replace them at no extra cost; if they leave mid-engagement, we provide a replacement within five business days. For an analyst role, where fit with stakeholders matters as much as skills, that guarantee has practical weight. For clients in Western Europe, a nearshore analyst in a nearby time zone also makes live workshops realistic, which matters more for analysts than for most developers.

There is a middle option too: an external analyst for the intensive opening phase, tasked with handing knowledge over to an internal person who takes the role into run mode. This has to be planned from the start, with documentation, joint sessions and a period when both people overlap. Without that, the knowledge leaves the project with the last invoice.

How do you test an analyst candidate in an interview?

Certifications tell you something, but not much. CBAP, IIBA’s certification for experienced business analysts, requires 7,500 hours of business analysis work in the last ten years, so it confirms that the candidate has done the job for a long time. IREB CPRE at foundation level confirms familiarity with the vocabulary and techniques of requirements engineering. Neither tells you whether the candidate is strong in the half of analysis your project needs most.

So use short exercises, different for each role. For a business analyst: describe a situation where two department heads want conflicting things, and ask for a plan for the first conversation. A good candidate starts with questions about the goal and the success measure, not with a solution. A weak one proposes a functional compromise before understanding what the dispute is about.

For a systems analyst: show an excerpt from an API specification, or the schema of two tables from an existing system, and ask for the three questions that must be answered before designing the integration. A good candidate asks about idempotency, error handling, volumes, the source of truth for the data and who owns changes to the interface. A weak one lists which fields need to be copied across.

For a candidate for a combined role, run both exercises and compare the quality of the answers. If one is clearly weaker, you know which half of the analysis on your project will go unattended. That need not rule the candidate out if someone else can cover that half, but you must know before the contract is signed, not after the third sprint.

How should you plan an analyst’s first three months?

A work plan for the analyst is the best test of whether you are hiring the right person. If you cannot write down what must exist after the first three months, you probably do not yet know whom you need.

For a systems analyst on a platform replacement, a sample plan looks like this. Month one: an inventory of the current system’s interfaces, a data flow map, a list of undocumented behaviours found in interviews and in the code, and a list of external system owners. Month two: the target data model, contracts for the first interfaces, and non-functional requirements written down with values. Month three: detailed requirements for the first functional slice, a data migration plan for that slice, and acceptance criteria agreed with the test team.

For a business analyst on a new product, the plan looks different. Month one: a stakeholder map, conversations with prospective users, and a list of hypotheses ranked by risk. Month two: target process descriptions, prioritised scope for the first release, and agreed success measures. Month three: a first-release backlog with acceptance criteria, and a plan for measuring benefits after launch.

The plan has a second purpose: after three months it lets you judge whether the role is still needed at the same intensity. If the business analyst has delivered everything and the remaining work is refining stories, the product owner or the systems analyst can take that on. If the systems analyst is still uncovering undocumented behaviour in the old platform, the role should run longer than planned.

What are the key takeaways for a programme manager?

“Business analyst or systems analyst” is a good question only when it is about sequence and proportion, not about choosing one kind of work over the other. Both kinds of analysis must be done. What you have to decide is who does it in month one, who takes it over later, and whether it needs a separate role at all.

Remember six things. First, start with the unknowns test: ask five questions about the problem and five about the system, to both sides of the project, and count the “we don’t know” answers. Second, on projects that change existing systems, the systems analyst should usually come first, because that is where the larger risk sits. Third, check whether the client side has a product owner with time and authority, because they can carry much of the business analysis. Fourth, one person covering both roles works on small projects, but only if they are strong in both halves, and you check that with an exercise, not a CV. Fifth, judge an analyst’s value by the cost of the mistake they prevent, not by comparison with a developer’s rate. Sixth, write into the plan who does the analysis no analyst is doing, because someone always will.

If the unknowns test has told you whom you are looking for, we can put forward business analysts, systems analysts and combined profiles through our staff augmentation services, typically with first candidate profiles within 48–72 hours. If you would rather talk through the team composition and the order of roles first, get in touch and we will help you lay out the first three months of analysis work for your project.