If you run marketing inside a SaaS company, you've felt the tension between what you can measure today and what actually moves the market over months and years. Performance campaigns report clean numbers. Brand work usually doesn't, and that asymmetry decides budgets.

On a recent episode of the SaaS Stories podcast, I spoke with Stephanie Clapham, Director of Research at Latana, about how next-generation brand tracking is closing that gap. Rather than write a guest profile, here are the practical takeaways — whether you're the first marketer at an early-stage startup or running a mature B2B team across regions.

Listen to the full episode of SaaS Stories with Stephanie Clapham

Why classic brand tracking keeps failing modern teams

Most brand trackers were designed for a world of slow research cycles, email panels, long surveys and tidy quarterly decks. They break in a market that moves weekly, where decision-makers live on a phone. The cost isn't only money — it's missed signal. When your tracker can't reliably reach the audience you care about, or every wave arrives as an isolated event, you spend the meeting debating data quality instead of deciding anything.

Stephanie's critique is blunt: traditional systems make brand measurement feel like an art because the methods are outdated. It shows up as three chronic symptoms. Access to the right respondents is limited. Samples are too small or too arbitrary to support segmentation. And confidence intervals get ignored, so teams overread noisy movement. The result is a neat dashboard that can't tell you where to invest.

Start with the foundations that predict growth

Every category differs, but a brand's basic needs are familiar. Track awareness, perception, consideration and preference among the specific audience you serve. That audience choice matters as much as the metrics — it decides whether your numbers reflect your market position or the opinions of whoever was cheapest to reach.

Don't try to measure everything in one instrument. Long trackers are the enemy of good data and of respondent goodwill. Lock your core health metrics to a consistent cadence, then answer bigger questions with targeted dips. The modular approach keeps the pulse steady while letting you test new hypotheses without bloating the base study.

Treat reliability as a requirement

Teams cling to round sample sizes. Saying we need 400 completes per market feels concrete. The better question is what margin of error you're willing to live with on the decision you're about to make.

If you plan to slice by role, company size and region, the headline sample number is irrelevant unless you can show acceptable confidence at segment level. Build the plan around margins of error, and label them on every chart. It changes the quality of conversation in the rooms where budget gets allocated.

Trust is now a metric and a method

Consumer trust has eroded across sectors for years, which affects how buyers receive your message and how your team should treat research inputs. If stakeholders doubt the method, they won't act on the insight.

That's why collection design is strategic rather than operational. Meeting respondents where they already are, instead of coaxing them through a twenty-minute panel survey, increases authenticity. Cleaning data with behaviour-based quality scores reduces fraud while preserving real voices. When the inputs are trustworthy, you get to talk about brand truth instead of caveats.

Brand tracking isn't only for acquisition

In SaaS the revenue story is dominated by expansion and retention, and brand shows up there too. Perception, consideration and preference inside your installed base are leading indicators of renewal risk and upsell potential. A tracker that can't see movement among your own customers is missing the arena where most of your long-term revenue lives.

Deep segmentation is the unlock, and it requires combining multiple traits in one view without collapsing reliability. Multilevel regression with post-stratification — the Bayesian workhorse behind Latana's approach — is one way teams are solving it.

How to begin when budget is tight

You don't need a complex program to get value. Define the audience that matters, fix a clean set of health metrics, and include your closest competitors for context. Keep the instrument short, mobile-native and consistent.

Working with a partner? Prioritise their ability to reach your audience, quantify uncertainty, and adapt as you learn. Doing it in-house? Be honest about setup load — sampling frameworks, survey design and quality control consume more time than expected, and errors compound quickly. Early discipline beats a bigger budget that arrives later.

Where AI helps, and where to stay sceptical

AI isn't a magic wand, but it earns its place in three spots today. Modelling, where treating waves as related rather than isolated stabilises estimates and forms a coherent narrative over time. Data quality, where machine learning scores respondents across behavioural signals and improves the balance between overcleaning and letting bots through. And speed to insight, where natural-language retrieval against your own dataset cuts the time from fieldwork to action.

Stay sceptical in two places. Fraud is rising because bad actors have the same tools, so static quality systems lose ground every quarter. And instant answers only help when the inputs and the uncertainty are visible. Fast and wrong is worse than slow and right.

What changes over the next three to five years

Panels will represent less of the population brands need to reach — Gen Z isn't sitting in an inbox waiting for a survey invitation. Ad-based sampling that meets people inside their natural media patterns becomes the norm. Demand for near-real-time reads will push teams to automate insight generation and treat trackers as living systems rather than quarterly rituals. And verifying respondents will move from a checkbox to an ongoing arms race.

Through all of it, the brands that win will pair better methods with better questions.

A practical way to apply this next week

Audit your brand health reporting against four prompts. Are we measuring the audience we sell to, or the one we can afford to reach? Are awareness, perception, consideration and preference tracked regularly, with margins of error on every view? Is the instrument short enough that you'd happily take it yourself on a phone? Do you know which segments you can cut reliably and which you can't?

Pick one gap and fix it in the next cycle. Replace sample-size targets with margin-of-error targets for your primary audience. Move questions out of the base study into a dip. Add competitor reads if you lack context.

None of that requires rebuilding your research program. It just makes it useful.