Skip to main content

Buyer workspace · this week · ranked by community votes

Sign in
List your SaaS

Agree Your Requirements Before You Look at SaaS Tools

Buying Process · 11 min read ·

Teams that start with demos buy what dazzles. Teams that start with requirements buy what fits. How to write, rank and agree a short list first.

Illustration: A calm indigo checklist panel with a compact data grid: columns Requirement, Why, Owner, Priority, with must-have rows marked by small solid squares

The usual route to a new tool goes like this. Someone is frustrated, a colleague mentions a product, a demo is booked, the demo is excellent and a purchase follows. Six months later, half the team has never logged in and the other half is working around the tool's gaps.

The problem was not the tool. It was the order of events. The team looked at solutions before they had agreed on the problem. This guide offers a better order: write down what you need, rank it and agree it, before you look at a single product. It takes a few hours and saves months.

Why demos mislead

A demo is a performance. The vendor chooses the data, the path and the pace. Features that work beautifully in the demo may not match your reality. And a smooth presentation makes every feature feel essential.

Without a prior list of requirements, teams tend to evaluate on the following.

  • What they were shown, not what they need.
  • Who presented best.
  • What the loudest person in the room liked.
  • What is new and shiny.

A written list shifts the conversation from "do we like it?" to "does it meet the things we said we needed?"

What requirements are

Requirements analysis is the process of determining the needs or conditions a product must meet, taking into account the needs of its different users. For software purchases, that means writing down what the tool must do, how it must behave and what constraints it must respect.

Notice that requirements describe needs, not solutions. "Notify the right person when a task is overdue" is a requirement. "Has a bell icon" is a feature.

Start with the problem

Before any list, write the problem in a paragraph.

  • What is happening today? Describe it plainly.
  • What does it cost? In time, money, errors or frustration. A rough estimate is fine.
  • Who is affected?
  • What would success look like? In a sentence or two, observable.
  • What have we tried? Including existing tools and manual workarounds.

If you cannot write this paragraph, you are not ready to buy. If members of your team write different paragraphs, you have found the first thing to resolve.

Gather the right people

Requirements come from several perspectives. Involve:

Daily users. They know the real work and the real pain.

Administrators. They will set it up, manage access and handle problems.

Managers. They need the tool to support reporting and outcomes.

Finance. They care about cost, contract terms and renewal.

Security and data protection. They care about where data lives, who can see it and what risks arise.

IT or technical staff, who understand integrations with existing systems.

A decision-maker, who will say yes or no and needs to trust the process.

Not everyone needs a seat in every meeting. But every group should have a chance to contribute before the list is frozen. Missed stakeholders are the commonest source of late surprises.

Collect requirements

A few simple methods work well.

Interviews. Ask users to walk you through a typical day with the current process. Ask what is slow, what goes wrong, what they work around.

Observation. Watch someone do the job. People rarely describe everything they do.

Document review. Look at current forms, spreadsheets and emails. They reveal hidden requirements.

A workshop. Put users, admins and managers in a room and have them write requirements on notes, then group them.

A survey, for large groups. Short, with a few open questions.

Capture requirements in a plain format, one per line, in the form: "The tool must let [who] do [what] so that [why]." The "so that" matters. It stops you from collecting solutions in disguise.

Sort into categories

Group requirements into types.

Functional: what the tool does. Create, share, report, approve.

Data: what it stores, imports and exports.

Integration: what it connects to.

Security and privacy: access control, data location, auditing, deletion.

Performance and reliability: speed, uptime, support hours.

Usability and accessibility: ease of learning, support for assistive technology.

Administration: user management, permissions, billing.

Commercial: cost, contract length, exit terms.

Support and training.

For security, the UK National Cyber Security Centre's cloud security principles are designed to help you choose a cloud provider that meets your security needs, and they apply to software as a service as well as to platforms. You do not need to adopt them wholesale, but they offer a ready-made checklist for thinking through what to ask.

Rank them

Not all requirements are equal. Rank each.

A familiar method is MoSCoW: must have, should have, could have and will not have this time.

  • Must: without this, the purchase fails. Keep this list short. If everything is a must, nothing is.
  • Should: important, but a workaround exists.
  • Could: nice to have.
  • Will not this time: explicitly out of scope.

Test your musts: "If a tool did everything else perfectly but lacked this, would we still reject it?" If not, it is a should.

Write each requirement so it can be tested

A requirement you cannot test is a wish.

Weak: "Easy to use."

Better: "A new user can complete the main task without help within fifteen minutes."

Weak: "Secure."

Better: "Supports single sign-on with our identity provider and two-step verification for all users."

Weak: "Good reporting."

Better: "Can export a monthly summary of tasks by owner as a spreadsheet."

Each testable requirement can be checked in a demo or a pilot, which makes evaluation fair.

Include constraints

Some requirements are not features but boundaries.

  • Budget range, as a ceiling or band, kept in your own records.
  • Timeline. When must it be live?
  • Data location and legal needs. Regulatory or contractual restrictions.
  • Existing systems it must work with.
  • Staff capacity to set up and run it.
  • Risk tolerance. How much would failure cost?

Writing these down avoids spending weeks on a tool that was never feasible.

Freeze, then guard

Once the list is agreed, freeze it before you look at any product. Share it with all stakeholders and get a clear yes.

During evaluation, new ideas will appear, often triggered by demos. Handle them calmly.

  • Add only by agreement.
  • Say what it displaces or what priority it has.
  • Note the source: is it a real need or a reaction to a flashy feature?

A list that grows with every demo ends up describing the most feature-rich product, not the right one.

Use the list

With requirements agreed, you can do things that were impossible before.

Search with purpose. Use the must-haves as filters when browsing directories and categories.

Build a shortlist of tools that plausibly meet the musts.

Send the list to vendors and ask them to respond against it. Request evidence, not just yes or no.

Run demos to your script. Ask vendors to show how their tool meets specific requirements with your sample data.

Design a pilot around the musts.

Score fairly. Evaluate each product against the same list.

Justify the decision. If anyone asks why you chose this tool, you can answer with the list.

Government guidance on buying and building technology, such as the UK's Technology Code of Practice, places weight on meeting user needs and making things open and reusable. The thread is the same: understand needs first, then choose.

Keep it proportionate

Not every purchase needs a long process. Match the effort to the stakes.

  • Low cost, easy to leave: a one-page list and a free trial.
  • Moderate: a short workshop, a written list and a pilot.
  • High cost, deep integration or sensitive data: a thorough list, security and legal review and a formal pilot.

Spending a week on a tool you could swap out in a day is as foolish as spending a day on one you will be stuck with for years.

A worked example in words

A marketing team of eight feels overwhelmed by content requests arriving by email, chat and hallway conversation. The lead writes the problem: requests are lost, priorities are unclear, and reporting is manual. Success means one place for requests, clear owners and a monthly summary.

A one-hour workshop with the team, a designer, the finance lead and the IT manager produces twenty-two requirements. After ranking, six are must-haves: a single intake form, assignment to an owner, status visible to requesters, comments on requests, export of a monthly summary and single sign-on. Four are shoulds, including calendar view. The rest are could or will not. Constraints: live within six weeks, data in the UK or EU, no per-seat cost above a ceiling.

Using the list, the lead filters a directory to a shortlist of five, sends the six musts to each vendor and runs demos against the team's own sample requests. Two tools fail the single sign-on requirement and drop out. The final choice is made in a short meeting by reading the list, not by who gave the nicest demo.

Common mistakes

  • Starting with demos.
  • Missing stakeholders.
  • Collecting features, not needs.
  • Everything is a must.
  • Untestable requirements.
  • Letting the list grow during evaluation.
  • Over-engineering a small purchase.
  • Not writing down the problem.

On this site

The categories and search pages help you filter products by what they do, using your must-haves as a guide, and the launches page shows new entrants. The submit page is where vendors list a product, and the blog has further guides for buyers and teams.

The short version

Write the problem first, gather users, administrators, finance, security and a decision-maker, collect requirements in the form "who must do what so that", group and rank them with must, should, could and not this time, make each one testable, record constraints, freeze the list before demos and add to it only by agreement. Then shortlist, demo and pilot against the list. Fit beats dazzle.

Questions and answers

Why write requirements before looking at tools?
Because demos are persuasive. A prior list of what you need keeps the team judging fit instead of polish.
Who should be involved?
People who will use the tool, people who administer it, finance and security or data protection, and a decision-maker.
How long should the list be?
Short enough to use: usually ten to twenty requirements, ranked by importance.
What is the difference between a need and a want?
A need is something the tool must do for the purchase to be worthwhile. A want is helpful but optional.
How do I keep requirements from growing?
Freeze the list before demos, and add new items only by agreement, with a note of what they displace.

Sources

Related guides

Need help?