Why SAAS marketing audience research pays off before you spend

What to take away

Why audience research comes before spend

Most software companies write copy, pick channels and set budgets before they know who they are selling to. That order is backwards. Audience research is the cheapest part of the programme and the one that determines whether everything after it works.

For a UK software business, the audience question has three layers. Who exists in the market. Who can buy what you sell. Who will actually move this quarter. Labour market data answers the first, your own customer base answers the second, and sales conversations answer the third.

The payoff is narrow targeting. A team that knows its buyers are operations managers in mid-sized logistics firms can ignore most of the market and still hit its number. A team that targets "businesses in the UK" will spend the same money and learn nothing.

If you want the wider planning picture before you start, the companion guide on SAAS marketing trends and outlook for England in 2027 sets out how budgets and channels are shifting.

Employment data as a market map

The Office for National Statistics publishes monthly employment and employee types data, including counts by industry and working pattern. The release covering May to July 2026 put UK employment at 33.41 million people. That figure is not your market, but it is the denominator you reason from.

The release to cite is the ONS UK labour market statistical bulletin, which appears monthly alongside these dataset tables.

The useful move is to break the total down. Which industries employ the roles your product supports. How many of those roles sit in firms above your minimum contract size. Which regions cluster them.

ONS industry breakdowns let you sanity-check assumptions. If you believe accountancy firms are a core segment, you can see how many people the sector employs before you build a campaign around it.

Keep the output small. A one-page market map with three to five segments is enough to brief a campaign. More than that and nobody uses it.

Five research methods and when each one works

Desk research

Start with published data, trade bodies and your own analytics. Desk research costs time, not money, and it should answer the size and shape questions before you speak to anyone.

Customer interviews

Interviews reveal language, triggers and objections that no dataset contains. Ten to fifteen conversations with recent buyers will surface more usable insight than a hundred survey responses with weak questions. The method is set out in detail in the guide to SAAS marketing customer interviews in England, which covers recruitment, question design and how to avoid leading the witness.

Three questions that earn their place: "Walk me through the last time this problem came up." "What did you try before you looked for a tool?" "Who else had to agree before you could buy?" Record the calls and pay for transcription if you can; Otter.ai or your meeting software both handle it.

Surveys

Surveys work when you already know the options and need to size them. They are poor at discovering new motivations. Keep them short and expect a low response rate from busy B2B buyers.

A tool such as Typeform, SurveyMonkey or Google Forms is enough. Two questions to include: "Which of these four problems is most urgent for you this quarter?" and "Who signs off the budget for tools like this?"

Sales-call review

Recorded calls are an underused dataset. Listen for the words buyers use in the first five minutes, and for the point at which they ask about price. Tag ten calls with the objection raised and the trigger mentioned, then count the tags.

Competitor and review mining

Reviews of rival products show what buyers value and what they complain about. Treat this as secondary evidence and check it against your own interviews.

For teams that want formal training in this work, the Chartered Institute of Marketing runs data and insights training courses covering research design and analysis.

Building the audience profile

An audience profile for a software company is not a personality sketch. It is a set of attributes you can filter on or ask about in a form.

Firmographic attributes include industry, employee count, revenue band and technology stack. Role attributes include job title, seniority and function. Behavioural attributes include what they searched for, which pages they read and whether they have used a competitor.

Trigger attributes matter most and are hardest to find. A new compliance deadline, a funding round, a change of finance system, a merger. These are the events that convert a latent need into a live project.

Write the profile as a table with attribute, source and confidence. Confidence should be high, medium or low. A four-column template, the same three fields plus the date last checked, is enough. Keep it in a shared spreadsheet, such as Google Sheets or Notion, so the team edits one version.

This stops a hunch from being treated as a fact three months later.

Once the profile exists, the guide to SAAS marketing target audience in England shows how to turn it into a working segment definition with inclusion and exclusion rules.

Where competitor research fits

Audience research tells you who to serve. Competitor research tells you who already serves them and how. Running the two together stops you building a profile for a segment that is already saturated.

Look at which segments rivals ignore, which they over-serve and which they serve badly. A gap in service quality is often more valuable than a gap in the market. The method for mapping this is covered in the guide to SAAS marketing competitor research in England, including how to score positioning claims against evidence.

Keep this work separate from your own audience profile. Mixing the two produces a document that describes your competitors' customers rather than your own.

Turning findings into personas

The persona is the output format, not the research method. A persona built before the research is fiction with a stock photo.

A usable B2B persona has four parts. The role and its remit. The metrics that person is judged on. The objections they raise. The evidence that triggers a search.

Keep personas to three or four. Each one should map to a segment you can target and a message you can write. If two personas would receive identical ad copy, merge them.

The guide to SAAS marketing buyer personas in England covers how to write them so sales and marketing use the same version, and how to retire one when the evidence changes.

Before and after: a worked example

This table shows a fictional B2B software company, a team of 12 selling scheduling software to UK healthcare providers, before and after a six-week research programme. Figures are illustrative.

To set your own baseline, take one quarter of acquisition spend and divide it by the leads sales accepted as qualified. Repeat the sum month by month so a seasonal peak does not flatter the figure. Change one variable at a time and keep the formula fixed.

The spend did not change. The definition of the audience did. That is the usual pattern: research reallocates effort rather than adding cost.

In this illustrative example the research programme took about 45 hours of product marketing time across six weeks. For most teams a first pass is 40 to 60 hours over two to four weeks. The only cash cost is usually survey or transcription tooling, and the total depends on seat count and plan.

Platform targeting in practice

LinkedIn Ads lets you advertise to its audience of professionals using firmographic and job-title targeting, including company size, industry, seniority and function. That maps directly onto the firmographic and role attributes in your profile.

Google Ads works differently. Its account structure separates campaigns, ad groups and keywords, which means audience definitions have to be expressed as keyword themes and audience layers rather than a single filter. That documentation explains how the parts fit together.

Platform audience size estimates are a check on your market map, not a substitute for it.

Build the audience in a document first. Then translate it into platform settings. Doing it the other way round means your strategy is whatever the interface allowed you to click.

Privacy and data sourcing

The Information Commissioner's Office sets out what organisations must do when they use the marketing services of data brokers. The duties include having a lawful basis for processing, being transparent with the people whose data you hold, and honouring objections.

For most software companies, bought lists are a poor trade. The data is often stale, the lawful basis is often unclear, and the reputational cost of a complaint outweighs the reach.

First-party data avoids most of this. Form fills, product usage, email engagement and sales notes are all data you collected with a clear purpose. Use them, and document what you will do with them.

If you do buy data, check the contract for the broker's assurances on consent and suppression. Keep a record of the checks. A breach of the electronic marketing rules can lead to enforcement action, and "the list provider said it was fine" is not a defence.

Running research as a cycle

Audience research is not a project with an end date. Markets move, job titles change, and your product adds features that attract new buyers.

A practical cycle is quarterly. One week of desk research to refresh the market map. Two or three customer interviews. A review of closed-lost deals. An update to the persona documents with a note on what changed and why.

Assign an owner. Research without an owner becomes a folder nobody opens. The owner does not need to be a dedicated researcher; a product marketer with two days a quarter is enough.

Record what you decided not to do as well as what you did. The exclusions are often the most valuable part of the document, because they stop the same debate recurring.

How to brief an agency

If you bring in a software marketing agency, the brief should contain your market map, your persona documents and the questions you have not answered. An agency that starts with channel tactics before reading those has skipped the work.

Ask what evidence they will collect, how they will store it and who owns it at the end. You should own the raw data, not just the slide deck.

Set a review point at four to six weeks. By then you should have interview notes, a revised audience profile and a first set of campaign changes. If the deliverable is only a presentation, the engagement has not started.

Common questions

How long does audience research take?

A first pass takes two to four weeks for a small team, covering desk research, eight to twelve interviews and a written profile. The quarterly refresh afterwards takes two or three days.

Do we need primary research if we already have customer data?

Usually yes, but less of it. Your data tells you who bought. Interviews tell you why, and why is what the copy needs.

Can we target the whole UK market and let the platform optimise?

You can, but the platform optimises towards whoever converts, which may not be whoever renews. Start from the segment you can serve and defend.

What should we do with the findings?

Turn them into three or four personas, a segment definition with exclusions, and a short list of changes to messaging and targeting. Then review in 90 days.