Overview

An AI-powered data analytics product, one of the bigger bets at the time in the team's roadmap, had already been through several rounds of evaluative research. The reception was positive but shallow: people found it interesting, even "cool," but struggled to say what problem it actually solved for them. That's a hard place to build a go-to-market from.

This project is a different kind of study than most in this portfolio. Rather than simply discovering user needs from scratch, it set out to test and prioritise value propositions, the actual messages (or perhaps more accurately the intent) a business would use to sell and position the product. That meant working far more closely, and far earlier, with Product and Design stakeholders than a more typical discovery project: defining the candidate propositions to test was as much a research activity as testing them with customers.

It ran as two related phases across two studies: a qualitative round to explore how different value propositions landed, followed by a quantitative round to prioritise them, and the specific needs, problems, and product properties behind them, at scale.

Note: In this page, specific results are are described directionally, instead, in line with ServiceNow's confidentiality commitments. I seriously understand that I am omitting a lot in these descriptions, but be mindful of the strict confidentiality that I have to adhere to. I'd be happy to clarify any questions you might have on this regard.

Team

•  1 Staff UX Researcher (yours truly)
• Multiple Product Managers, Product Design, and Engineering stakeholders

My Contributions

Led User Research across all stages:
• Reframing broad stakeholder ask into focused research goals;
• Running stakeholder workshops to define candidate value propositions ahead of customer validation;
• Designing both the qualitative (interviews) and quantitative (survey, MaxDiff) phases;
• Synthesising findings into positioning and messaging recommendations, shared continuously with stakeholders;

Timeline

• Q3-Q4 - 2025

The Process

For this research project, I followed a 6-stage process: Kickoff, Planning, Preparing, Running, Analysing, and Sharing. In the next sections, I'll provide a summary of each stage. As with any exploratory research, some stages ran in parallel or looped back on each other rather than strictly in sequence.

Kickoff

Prior studies on this product had a recurring theme: positive reactions, but no clearly understood problem being solved. This was a risk for long-term adoption and relevance. After some period of "lobbying", we gathered support from stakeholders to chase two main goals: identify which value propositions resonate most with customers, and why, and prioritise the user needs and problems those propositions should be built around.

Because the object of study was messaging and intent itself, not just user behaviour, this project needed close, ongoing collaboration with Product Management and other Customer-Experience-focused stakeholders from day one — not just at the readout.

Planning

The research questions driving this study were roughly:

• Which value propositions resonate most with customers, and why?

• Which underlying needs, problems, and product properties should those value propositions be built around — and how should they be prioritised against each other?

Answering both meant pairing depth with scale, so I structured the work as two connected phases, each with a distinct methodological role:

• Phase 1: Qualitative exploration: a stakeholder workshop to define candidate value propositions, followed by moderated customer interviews combining structured ranking exercises with open discussion, to see how each proposition landed and why.

• Phase 2: Quantitative prioritisation: a survey applying MaxDiff analysis (a trade-off methodology built to counter the "everything sounds nice" bias of simple rating scales), to rank propositions and the needs, problems, and product properties behind them, across a meaningfully larger and more varied group than interviews alone could reach.

Together, the two phases were designed to answer not just what resonated, but why it resonated and how much it mattered relative to everything else on the table; the difference between a preference and a priority.

Preparing

For Phase 1, I ran a workshop with Product, Design, and Engineering stakeholders to define the value propositions worth testing, and mapped each one back to the problems, goals, and tasks it was meant to address. I also designed the interview's ranking exercises: jobs, needs, problems, and goals; so customer priorities could be captured before any concept was shown, avoiding the bias of ranking against propositions we'd already put in front of them.

For Phase 2, after the customer feedback of Phase 1, we refined and translated the sharpest four propositions into a MaxDiff design across the underlying product properties. These were put against an eight-dimension perception scale, with rotation to control for ordering effects. Again, I emphasize, we're not looking for simple preference metrics (X% like A), we're gathering evidence on what value propositions truly resonate with customers and why.

Running

Phase 1 sessions followed a consistent arc: warm up on the customer's own role and day-to-day context, then a structured exercise ranking their real tasks, needs, and problems in data analytics. All this before any concept was shown, so their priorities wouldn't be anchored by what we were about to pitch. Only then did we introduce four draft value propositions, one at a time in rotated order, each followed by open reaction and a closing forced-rank across all four. Sessions ran with ServiceNow customers spanning industries and seniority. These included primary decision-makers to hands-on analytics users. We used Miro to support the live ranking exercises alongside the conversation.

Phase 2 broadened this into a self-serve survey, translating this comparable logic into a quantitative instrument: a MaxDiff task presenting customers with repeated small sets of product properties and asking them to pick the most and least valuable in each. This design forces genuine trade-offs rather than letting everything score highly. This was paired with a forced-choice ranking of the four value propositions and an eight-dimension perception rating of whichever one each customer preferred most. This reached a meaningfully larger and more diverse group than interviews alone could support, across industries, seniority, and tenure with the product.

Analysing

Phase 1 analysis worked from two data sources side by side: the structured rankings (surfaced what people prioritised) and the discussion around each value proposition (which surfaced why). Triangulating the two, rather than reading rankings and quotes in isolation, is what let me flag which top-proposition carry real risks - e.g., language testing well, but under completely wrong expectations. Lower-ranked propositions still earned a place in the discussion, as they kept surfacing in conversation even when it didn't top the ranking.

Phase 2 anlaysis relied partially on modelling the MaxDiff responses to estimate each property's relative share of preference, exactly to avoid the "everything is nice" flattening you might get from simple rating scales. Those models were then compared/correlated/analysed... by various segments, such as industry, tenure, ServiceNow usage frequency; to check whether the overall pattern held everywhere or masked meaningful differences underneath. While it did held, mostly. We've seen relevant differences in participants' responses from regulated industries, leaning towards properties others largely ignored. Combining these property-level MaxDiff results with the perception ratings on each customer's top proposition is also what pointed to the real barriers to adoption.

Sharing

Findings went out as two related readouts rather than one: a preliminary Phase 1 summary shared quickly to keep pace with decisions already in motion, then a full Phase 2 report, inc. executive summary, supporting detailed customer quotes and insights behind it. Stakeholders were kept in the loop between phases too, so positioning conversations didn't have to wait for a final report to start.

Outcomes

This study's findings had a direct line to decisions well beyond a typical roadmap conversation, i.e., into how the product would be positioned, sold, and potentially packaged and licensed. Specifically, it pushed the team to:

• Consolidate messaging around the ideas that actually resonated with customers;

• Prioritise trust-building over differentiation in go-to-market plans;

• Consider audience-specific messaging, since different identified segments didn't all prioritise the same thing;

• Flag specific industries as segments worth a dedicated follow-up, given the different priorities that surfaced.

More broadly, this project showed the value of treating positioning itself as a testable hypothesis, not just as a marketing afterthought, and it's opened the door to further studies validating messaging before it ships, rather than after (something that stakeholders were not yet used to).

Thank you

Recommended Next

OR
See all related projects below:

© Tiago Gonçalves, 2026