When New York elected Zohran Mamdani as mayor, most of my peers were celebrating – a rare surge of hope after years of political fatigue. But not everyone was cheering. A certain section of society reacted like someone had swapped their morning coffee for decaf: confused, jittery, slightly offended.

Because Mamdani wasn’t ‘supposed’ to win. Not according to the pundits. Not according to the polls. Not according to the stereotypes quietly baked into the way people talked about him – his background, his religion, his age, his politics.

And yet… there he is, the newly sworn-in mayor. The plot twist nobody saw coming, but half of New York City (and plenty beyond) cheered for.

For those of us in market research, this moment isn’t just political theatre. It’s a mirror, and not a flattering one. The same dynamics that shaped how Mamdani was framed are the ones we risk reproducing in our questionnaires, sampling frames and ‘representative’ claims.

Here are five lessons we should probably tattoo onto our industry consciousness before the next big cultural moment rolls around.

1. Language does more work than you think

Every question, every word choice, every example we give carries invisible baggage. Mamdani’s campaign showed how quickly identity cues get weaponised. If a single stray adjective in your questionnaire can nudge respondents toward a stereotype, congrats: you’ve recreated a mayoral smear campaign in miniature.

The solution is to use our curiosity as researchers and interrogate our work along with the structure within which this work occurs: seek out those who do not present as you do to challenge your outputs.

Are we auditing our work with colleagues and peers who have different lived experiences or are we marking our own homework by only checking our work with those who look and think like us?

2. Our samples should reflect the diverse landscape we exist in

Representation matters. Mamdani’s voter base wasn’t your typical online panel – it was multilingual families, immigrants, young renters, broke millennials, politically homeless suburbanites. Basically, the exact groups traditional samples tend to forget exist.

If your sample can’t meaningfully reach these communities, your ‘representative’ claim lacks structural integrity.

We need community recruitment, in-language outreach, mixed-method data collection, fair incentives and quotas that actually account for lived diversity – not just the same demographics we all default to.

3. Identity data is powerful – and politically flammable

Here’s the flip side: with representation comes responsibility. The debates after Mamdani’s win – his religion, his citizenship, who gets to lead a ‘real’ American city – underline how sensitive identity can be.

If you’re collecting that data in research, minimise it, secure it, give opt-outs, and explain why you need it. In 2025, ‘because the client asked’ doesn’t cut it anymore. As agency-side researchers, it is our job to challenge our clients. As clients, it is our job to represent our consumers truthfully. Diversity without data ethics isn’t progress; it’s a liability.

4. Moderators and analysts need cultural range, not just warm smiles

When respondents bring up topics like religion, migration, racism or systemic bias, the worst thing a moderator can do is smile politely and move on. Mamdani’s campaign reminded us that identity is always part of the conversation – even when people pretend it isn’t.

Equip moderators with cultural literacy training, escalation protocols and the confidence to sit with discomfort and complexity. We need to stop flattening whole populations into one axis of identity in our analysis. Intersectionality isn’t optional anymore. Again, the heart of our industry is in our curiosity, so we should all lean into discomfort and seek to understand the underlying ‘why’ so we can truly drive business and societal outcomes that matter.

5. Social listening is useful… until it isn’t

Mamdani’s momentum was undeniably fuelled by online energy: grassroots videos, creator amplification and digital organising. Social listening can catch that energy, but only if you know what you’re looking at.

Social listening should be treated as a signal amplifier, not a truth machine. Think of it like a messy roommate – occasionally helpful, often loud and definitely not reliable without context. Filter the bots, pile-ons and misinformation. Use social signals to spark questions, rather than declaring answers.

So, what does this all mean for DE&I in research?
Elections reveal the stories communities tell about themselves. Mamdani’s campaign revealed something else, too: trust is a currency, and communities who have been historically ignored won’t magically open up because you emailed them a 15-minute survey.

If brands, agencies and public bodies want to understand a genuinely diverse public, they can’t rely on the same old pipelines. DE&I isn’t the soft, fluffy bit – it’s the hard infrastructure of truth-finding. It is work that takes ongoing time and effort. Systems created over centuries take time to unpick and change.

Annabelle Jones is senior research manager at Simpson Carpenter and a member of the steering team at CORe (Colour of Research)