My writing

Shopping in India- then & now, a little nugget on what this could mean in the age of AI and clear evidence that I’m getting old.

The era of you-had-to-be-in-the-know of shopping Shopping in India in the 90s and early 2000s was this colorful, thrilling and exhausting experience; especially for weddings and festivals. You had to know the right markets and specific shops (who did the best silk, trendiest lehengas, or stocked the most range of fabric) to have a good shopping trip. People in the know held these secrets close to their hearts. There were endless shops, hidden entrances, underground markets tucked behind tiny doorways. It was easy to get lost or overwhelmed. Once you got in, you had to deal with a salesperson, pray they were in a good mood, air gesture at the wall-high clothes stack behind their back till they located the specific coordinates of your potential outfit and try out clothes in a tiny changing room where the lock didn't work. Even if you made it this far you could walk away with nothing if the price haggling didn’t end in a win for your team.

The era of easy shopping Then came the organized mega retail stores. I remember buying clothes at ‘Max’ for the first time and being mesmerized by the ease of it all. Everything was displayed openly. You scanned, picked, tried, bought and still had time for a bite before heading home. Shoppers Stop, Lifestyle, Westside — they took the stress out of shopping. The stores didn’t fully crack the extreme variety & customization of small retailers and bazaars, but the ease meant it didn’t matter, at least for a while.

The 180 to being in the know — shops hidden in the alleyways of Instagram When everyone has access to the same stores, twinning becomes an unfun recurrence. The sameness of fabrics gets to you. The desire for custom fit returns. So after a decade of big-box retail, you scroll on Instagram hoping to find your next dream outfit. You tell the algorithm what you’re looking for; the more you scroll and deep-dive into accounts, the better it learns your taste. You look for smaller accounts in search of niche, emerging aesthetics.

This age of Instagram shopping feels a lot like shopping back then; Instagram’s alleys hide vendors you know nothing about. The algorithm is your doorway to this underground market. You have to play the scrolling game to discover a unique find and enter WhatsApp conversations where the back and forth continues. Same same but different no?

As a product researcher, this fascinates me. You make things easy, predictable and accessible, and people will come. But in fashion and products, scale and sameness aren't always good from the customer's perspective — people start to crave organic, authentic, niche, small-batch experiences. They say AI will make it possible to scale and still stay hyper personalized. Maybe, we’ll see.

Real pictures from my very first smartphone - Xolo - taken during a Sari shopping trip.

View on LinkedIn ↗

How do you onboard to a company in a domain you have no idea about?? Here are some recollections and observations from across my career but if you have a crazy onboarding story or a secret sauce to do it well- I am all ears.

In 2017, I joined an Ads team at Google and I had zero knowledge of the Ads space. Nothing. Nada. But before UX research I had been a developer and a good chunk of AdMob's user-base were developers; developers integrating ads. That gave me some firm ground to kickoff from but there was a lot of new stuff to learn. I think I started a project the week after joining. So there was no lead time to understand everything.

Somehow I figured it all out- ROAS, eCPMS, CTRs, match rates and fill rates. Ads Ops and developers. The publishing business. I learned the domain while I learned the job. Bit by bit- sometimes fumbling, sometimes soaring.

Then at the start of the pandemic, I moved to Meta and my world turned upside-down. I knew the space- it was another side of ads but the scale of the work changed. And the pace rocketed. There were a lot more disciplines to collaborate with. I was swimming in change that brought so many emotions with it. I also missed my old team at Google. I almost went back.

And then somehow I put my head down, gave up control and showed up with vulnerability to get good at the job before me. I think at this point I've worked on 10 different teams(at least) and had about 10 different managers. I have worked on everything from how people buy stuff to why won't people use genAI. My onboarding muscle is actually a ripped 6-pack(help me out with a better analogy please)!

I ramble, here is my advice on onboarding (would love to hear yours too):

Meeting with as many people as will give you time, asking lots of questions and being ok with feeling dumb.

Learning the balancing act of stating 'I don't know (yet)' and 'based on what I do know here is what I'd do'.

You are going to be both a novice and expert for a little while and you're going to have to embrace the discomfort.

And the most important of them all- seek feedback from those around you and be willing to listen to it.

#Lifeintech #lifeofaUXR #onboarding #productbuilding #AdTech

View on LinkedIn ↗

Err the heights of AI-pocrisy?

I tried using Muse to dispute an Uber charge. Uber made it difficult for Muse every step of the way. Wouldn't let it enter the code (muse has email access to grab it) and then it put up the prove you're a human puzzle.

Here's the thing- Muse handed over the puzzle to me and I failed. I failed three times. At being human.

And then I give up and try to dispute directly in the Uber app and guess what- Uber's 'help' is a fully AI chatbot. And it gave me the refund in 3 questions.

If your customer support is a bot, why can't your bot just talk to my bot. Why do I need to sit there and click buttons?? 😂

#ai #uber #product

View on LinkedIn ↗

A few thoughts on the importance of silence- in work and in life.

In qualitative research, embracing silence is a key skill. If you can get good at it, you can get really good at gathering good product feedback. It's really simple. You ask a question and then you wait. Don't rush in to help them answer, don't complete their sentences. Don't make up their opinion for them. Why is this silence so important?? People take time to craft their responses. They're remembering. They're trying to articulate something difficult in words. They're warring in their head about what to share, how much to share and what is ok to say in this setting. Let them. That time between question and response is important.

It can be really uncomfortable if you're usually the one who takes the lead in driving conversations. Two things that can help- becoming aware of your behavior around interrupting silence can actually be immediately helpful in reducing it. Another tool is to practice asking a question and then counting to a number in your head (sloww count to 10) before you allow yourself to say anything. If you just did a double-take to the idea of waiting 10 whole seconds without a response- you are precisely the person who needs to practice this more.

This little research expertise nugget connected with another aspect of life recently. An ongoing advice in parenting circles is to let our kids be more bored. The prevailing knowledge says this time being bored sparks better creativity in play and builds focus. I've noticed this to be true when I've experimented at home. It's true for us adults too. The more distraction free our household is (phones away, TV off) the more fun our time together is.

We have all read articles on this. We're doing too much. Filling every micro-moment (the commute, the queues, the toilet seat) with content and productivity. Our phones were the old villains. Now it's AI. And a long time ago I remember it being the TV (anyone remember it being called the idiot box?). There is always something to fill your time and keep your mind running. But turns out we need spaces to do nothing, to arrive at free time with no plans to fill it.

So silence isn't just important during customer interactions. You need silence in your personal life too. To let your brain rest. So let's be ok with saying less, listening more and doing nothing from time to time.

#productthinking #research

View on LinkedIn ↗

It's been amazing to play around with Muse!! I didn't work on Muse during my time at Meta but I did work on lots of AI stuff and Meta AI --> the thing I learned and every researcher in this space keeps learning is that everyday folk don't want to learn about prompting and looping; they want stuff done and proactivity is a huge plus. They cracked it with Muse! It's been a joy to hook up email access for one task, forget about it and have it proactively remind me that I'm due to send my eldest with some fall leaves for a school activity- I'd completely missed this buried in the daily school report. Another time I was at a conference and it told me about an email that was super timely for me to respond to.

That little thumbs up acknowledging everything I ask for is super satisfying.

Some stuff still needs work, but it's a great, great first start. I keep saying this to people- it's way too soon to call it in this AI race. The best players could possibly be the ones yet to arrive.

The cute and fuzzy mascot is a brilliant move! Albeit, I have to now think for a beat before finding the app- I'm always typing Kit (my chosen mascot name) first, not finding it, and then searching Muse. Many personalized names might mean a less sticky brand name?

I'm also not fully sold on the main chat vs side chat aspect. As an AI power user, the break from chats to this is kind of disorienting but maybe everyone else doesn't care?

The biggest hurdle has been getting Muse to log into stuff that doesn't provide connectors. Hopefully there's a solution for that soon.

#ai #product #meta #muse

View on LinkedIn ↗

Once, a sales lead asked me to lower a number on a survey results deck. It was a % lift in revenue that would get people to try a competitive product. The sales team was worried it was too high, it would either become an unreasonable sales target or force the leadership team to pivot away from building that product.

Another time I was working with a local insights team (made up on non-researchers) in a key non-US market & noticed they gathered data about local behavior around our product and only reported stuff that was unique to their market and could lead to product changes that better suited the local users. They were also incentivized to influence new product features so they focused more on reporting findings that could lead to new features. They were not usually reporting ALL of their learnings. So when leadership heard their feedback- they assumed local behavior in that market was vastly different from the US AND that we needed a lot of new unique features to survive in that market. In my own research, I found this to be untrue- the full picture was always more nuanced.

In both of these cases, my research methods training & expertise helped me see and report data as I found it without altering facts or cherry-picking insights. But it was also a role thing. I'm incentivized and trained to produce good data about user behaviors. That is my role. It's ok if my data brings good news or bad news or ok news, convenient or inconvenient. It is literally my whole job to come back and say what I find.

Data about users can be political(a product in one org can be objectively terrible compared to another similar one owned by a different org), inconvenient (you need to unship an idea) or inconclusive (these tests results aren't pointing to a winner). It becomes extremely important to have dedicated people that are skilled in working with data & not have those responsibilities rolled up into an AI layer which seems to be a hot trend these days.

I'm all for making everyone more research capable but this does not replace the need for dedicated Data scientists and UX researchers. When you roll up insights functions into other jobs you lose that objectivity. The insights get tangled up in incentives of those other roles. It is incredibly hard to see this until teams move on to do other things and individuals retrospectively realize what happened.

#goodresearch #product #tech #data

View on LinkedIn ↗

2 thoughts on AI slop writing and websites. 2 antidotes that could fix it.

I wasn't joining the this is made by AI witchhunt because everyone has their own deal.. their English is bad, they need to get content out but their real life is difficulting them right now, they just want to voice opinions but ain't got no time to craft it. Whatever, no judgement from me. I write because I love writing. And I used em dashes before it became a sin to do it. But...even as non-judgemental as I am it's so so easy to spot Claude writing now. The sameness of it all makes my eyes glaze over. Give me all your raw thoughtsickles with bad grammar any day. Puhleeze sprinkle it with your own personality. It will be 100% better, I promise.

And we need to talk about the- this is my personal website- trend. Look I get it, coding your website with the AIs is mad fun. I built my own with Claude and codex (tokenmaxxing 🙌) over one weekend. Anyhow, lucky for me I'd spent time reviewing a dozen or so portfolios made with AI help before I made mine.... If someone didn't tell you- Claude designs websites that look a lot like Claude's design. Same with ChattyG. Has no one else picked up on this?

The backgrounds are all beigy grayish. The standout font is orang-ey. Or like it'll give you the serious designer aesthetic with grayscale and eggshell white and a very serious looking font. It'll play with whitespace and font sizes effects in the same exact way. After you see a few it's impossible to ignore the sameness. For most people this is probably fine. But if you need the design to stand out for your profession (like erm if you're a designer)- please use a dedicated separate chat session for designing the site. Give it a moodboard of sites and images that you love and brand guidelines (who you are, your values, your aesthetic preferences, images of rooms you love, clothing preferences. You can go wild with this). Create a dedicated design . md where all this lives once finalized so you can update it quickly without breaking your site. That's it. There's probably a lot more you can do to get fancy with it but this will do to make it not the same as every other AI generated website.

Again, no judgement from me but my two cents if you want to tackle some slop.

#betterwriting #betterwebsites

View on LinkedIn ↗

There's a new kind of founder being born in the AI era and that is the frustrated customer.

They were your loyal customer. They felt passionate about your product. Then they experienced frustration. Which they expressed in whatever channel they had access to. You didn't listen. You didn't mean to not listen- you prioritized ruthlessly and their feedback just didn't make the cut.

Used to be that when you were frustrated with a product you either sighed and shrugged or you shopped around. Today you set a weekend aside, roll up your sleeves and build.

I've spoken to a few different founders who's primary motivation to become the competition to their favorite product is- I've given them feedback, I've asked for this very thing and they couldn't put it on their roadmap. Or they took weeks to prioritize my issue. Now I can have an update ready in a couple hours, a day at most.

Good customer service and spending time understanding what your users actually need AND getting it shipped quickly is table stakes today. Yet teams with those exact skillsets (ahem UX researchers) are shrinking.

Deep and regular user understanding paired with quick action is necessary now more than ever. Invest in these skills! Hire people to listen to your frustrated customers or they will turn into your competition ;)

View on LinkedIn ↗

Maybe you're a founder or a designer or a PM without a researcher. Someone who is in a customer facing role. If listening to people who use a product is part of your role, keep reading.

Getting really good at listening to people AND observing them will give you a strong leg up in gathering product insight & uncovering differentiated value. These skills are also useful for anyone trying to do anything- it's going to make you a better co-worker and perhaps even better at your relationships!

So you have 30 mins with someone and you want to gather some product feedback.

Warm up first. Don't dive right in. Establish connection first, it will be worth it. People need to trust you and feel like they can share with you before they can open up.

Become better at listening. Get uncomfortable with silences so they can actually get their point across. Don't rush in to complete their sentences or provide explanations. It makes people switch off and become passive in the conversation. Let them lead. Jump & bring it back on topic when it truly veers off onto a tangent.

And then go beyond just listening. Combine it with observing. This is very powerful. People will say one thing and quietly turn around & do the opposite thing. Follow up your verbal questions with- would you mind just showing me how you do that. If it isn't a digital experience, get into the same space as them as much as possible. Zooming out and observing people use the product in their environment is eye opening.

Help people remember, so you're not relying on memory alone. Memory fades. Memory modifies. Put people back in the context and the moment as much as possible.

#userresearch #productdiscovery #product #qualitative

View on LinkedIn ↗

My turn to be annoyed with Claude. It's gotten so verbose and condescending. 3 months ago, I loved Claude over Chatgpt and now I switch away to chatgpt for every other task.

I just asked for weekend hike recommendations and I get this-

Let's be honest, this isn't about the weekend. It's about this other thing you didn't ask me about that I'm still gonna give you paras and paras of text on sprinkled with lots of judgement and somewhere in there is stuff about the hike that I'm too tired to read.

I don't even want to get started with the more complex workflows- walls of texts, lots of preamble and prose and judgement on the task being set.

I saw someone on LinkedIn say they've saved to preferences- please Claude just shut up and do the thing I ask you to do. I am starting to see why.

Anyone at Anthropic know what's going on?

View on LinkedIn ↗

To the tired, overworked, hopefully positively challenged, Tech worker-

May your users love your product and your metrics skyrocket, May your day be few in meetings and full of deep work pockets, May all your stakeholders be aligned and your projects in the green, Whatever your individual strife, I really hope you feel seen.

May your blockers be challenges that make you better than yesterday, And eventually be a great story that sends that promotion your way. May all the feedback you receive be constructive & precise, The kind that ends up being, a blessing in disguise.

May everyone you encounter understand what it is that you do, and if they try to do your job for you, may that they at least understand that they suck at it more than you.

May your manager be an ally and if not, at least there’s a mentor who’s got your back, May your 1:1s be fruitful & joyful; full of gossip or learning or enlightenment. If they are all a bunch of updates on status tales, this is your permission to turn that meeting into an email.

May your HC requests get approved & your team grow full & strong, May you find the perfect person. With this job market, there is no way it takes too long.

If you’ve got company stock I wish that it falls just before grant, falls even more at vest, And when you want to sell I hope, it gets to its very best.

May the AIs in your life be productive & purposeful, May you dream of what you wish to create and not worry about prompts and tokens, May everything that you ship with it stay strong & unbroken (well, I just needed something to rhyme with token).

May your time on LinkedIn be brief & mentally stimulating, I hope no social comparison gets you down today, And I hope this made you smile a little on this Monday!

Drop your manifestation, and then go do stuff!

(People have been reaching out reading my stuff here on LinkedIn, I love those conversations. If you're doing something in Tech- product, design research, whatever- reach out, let's talk.)

#techlife #product

View on LinkedIn ↗

At some point last year my toddler at our weighing scale and asks how does it work? I was distracted; 'Figure it out', I said- my usual response to help her learn through exploration. She then proceeds to say, 'Hey weight machine, how do you work?!' and looks at me expectantly because I usually rephrase if our smart home devices don't get what she is saying.

Somehow her little brain has come to associate Tech with voice because that is her world. I looked at her dumbfounded and thought, well why doesn't our weighing machine talk? Would be nice!

Anyhow, cut to 2026 and I'm constantly tinkering and building with AI. And I love doing it with Voice. For someone who spent the last decade working on screens and helping teams optimize flows on screen, clicking around for an outcome feels so archaic now. I want to do everything with my voice, as much as possible.

- it's soooo much faster than typing - typing requires well-formed thoughts and I am now in the habit of transferring my stream of consciousness into a session before giving the raw thoughts some shape. - and my goodness, this is my favorite- the ability to just stay in one window and shoot out voice commands to go ask codex to do stuff in other places while I focus on the thing at hand instead of getting lost in tabs on tabs on tabs has been life-altering. I stand by this hyperbole.

Wispr Flow is amazing. Voice within Codex is a game-changer. I haven't played around with other voice integrations- any you recommend?

#buildwithAI #productwork

View on LinkedIn ↗

Product researchers are often told to learn about the business; speak the language of your key decision makers; learn about what the company cares about and ladder your work up to it. Failure of research projects is often mapped to a failure in understanding the business & stakeholders.

It's fair feedback but thrown around way more often than is valid. Even when people do all the legwork- understand the business, stakeholders, accurately define what is unknown, use the right language- research projects can still fail. Why?

- the right questions but the wrong stakeholders. If your project is about to kill your PMs idea, it might need to be more visible to their boss to have any impact.

- the right study but the wrong timing. The headcount got approved but it took time to find the right person. The problem understanding moved forward anyway. Before the insights land- the decision makers moved on without you knowing.

- the right insight but the wrong leadership. Research insights can sometimes be uncomfortable. It might mean pausing investments or redirecting resources. It's healthy when leaders interrogate findings but it's equally important for them to amplify insights when the results are sound. Good research can only go so far without fearless leaders who will back the work and make change happen.

Landing impact from a research project is a complex sport- the timing, the stakeholders, the leadership, the stars and the suns- they all need to align. Learning the business is important, but its not the only lever.

If you're working with a researcher and their work isn't having impact- be kind to them. They're probably stressing out about learning the business and knowing all the right metrics when their head is full of useful insights that can help you crack the next product problem.

#productresearch #product #research

View on LinkedIn ↗

I am an AI-pilled person and still I say this: AI does not belong in every step of the workflow.

One part of my job I wouldn't fully hand over to AI is data analysis.

“Pick the top 5 themes and cite at least 3 quotes to support your analysis.” Nope. Don't do it.

Getting familiar with the data is where you start to see the story. Spending time with it is how you get from a bunch of datapoints to true insight. Many can do the “pick the themes from the transcripts” bit. My moat is my years of expertise in understanding human behavior and my lived experience, applied to the data to get to a unique observation.

AI helps me stay fully present in the data as it comes in, without worrying about double-checking everything in real time. I want to see the raw signal: the patterns, exceptions, contradictions, and language participants use. That means directly moderating or actively listening in, and reading as many participant entries as I can.

I still use AI to make sense of the shape of the data. Like understanding at a high-level on participant sentiment- summarizing participant responses succinctly; applying themes that I can use later for secondary analysis. But, the first pass on the data is for me. AI comes in later. Once I have my first draft of synthesis, I often get it critiqued: what did I miss? What did I overstate? AI adds useful checks, critiques and improvements before another human receives my work and I'm glad to have that support.

This isn't unique to UXR work, btw. I was speaking to a data expert and they shared the same thing: looking through the data helps them understand what it's saying. A junior engineer shared a similar sentiment. He stopped using the CLI to go straight from Jira ticket to committed code because it was keeping him from learning how the code actually worked.

This can look different for every project, but the headline is the same: don't fully hand over your moat to AI. Your involvement will increase quality and deliver proportionately higher value- to the project but also to you.

What's something you don't use AI for that you could?

#UXResearch #AI #ResearchInsights #ProductResearch

View on LinkedIn ↗

Well-intentioned product teams can do their research, add meaningful value, and still lose everyday use to a less capable alternative.

But first, a tale of two kitchen tools: the nifty pressure cooker and the powerful Instant Pot. The adjectives probably give this away, but read on.

The Instant Pot can do so much. It has more settings. The pressure system is friendlier. You can avoid the scary whistles with natural steam release. It can slow cook, make yogurt, and ferment dosa batter.

For a long time, I loved it and used it all the time.

It gave me real value: I could set dinner up, leave the house, and come back to it cooked. That is why I bought it.

But as our family has grown, my brain space for the drudgery of everyday cooking has shrunk. I do not have time to think. Cooking has to feel like an extension of my hands.

And so, about 75% of the time, I reach for my small pressure cooker.

Not because the Instant Pot is worse at cooking. It is better in many ways.

But it is bulky. It feels like it has more parts to wash. It takes more effort to take out, set up, clean, and put away.

I still reach for it when I need to set dinner up and leave the house. In that moment, the extra effort is worth it.

But for an ordinary weeknight meal, it usually is not.

Both have earned their spot in our kitchen, for now. But our kitchen has limited space. At some point, I will need another drawer for the kids’ things. When that happens, I will not sit down and compare every feature.

The things we use every day will stay within reach. The things we use less often will go to the garage; making access even harder and reducing use even more.

The Instant Pot innovated. It created real value. But it also made some parts of the experience more complicated.

As my life changed, that friction started to matter more.

A product can win someone over with what it makes possible. But whether it remains part of their everyday life depends on the full experience around that value.

That is what product teams can miss when we evaluate a feature only by the value it creates.

If you are a product builder thinking of a product-feature fit: - Have clarity around who you're building it for- the novice or the expert? - Evaluate the whole experience, not just the feature; observe what changes before, during, and after someone uses it. Does it remove friction or add another decision, setup step, or cleanup task? - Simplify. Aspire to reduce cognitive load, especially for the things people do often.

Because value that takes too much effort to access is value I may not use. Innovation can get you into someone’s life. Simplicity is often what keeps you in their everyday routine.

#productprinciples #UXR #productthinking

View on LinkedIn ↗

I am an expert in qualitative research and I believe AI-moderated qual is here to stay. It's not amazing (yet) and it's not replacing us (never). But it is a tool in our toolkit.

I know- we bring the human to research and AI moderators are not nearly as good as an experienced researcher. I didn't like it either. But after months of tinkering, this feels like the inevitable, better future.

Hear me out.

It makes the everyday tradeoffs of research suck a little less. We say we pick the methodology based on the research question. And we try to. But perfect scoping does not exist in industry research. It's defined by time, research ops, energy, strengths, and how accommodating (or not) your stakeholders are. When you need insights fast, or at scale, AI-facilitated research helps you deliver on time. Where speed is of the essence, being late with more depth makes you irrelevant, not credible.

There's a lot of research that needs doing (ahem, tactical work) and none of us want to do it. I can't count the number of times research teams wanted the unforgiving tactical work prioritized but no one wanted to do it. It can be boring, especially in a zone you've done lots of studies in already. It can feel repetitive and draining to repeat the script 8-10 times (this isn't exploratory qual where you can meander). It might not be your strength if you're a mixed-methods researcher who leans qual. And it's not work that gets you promoted beyond the first couple of rungs, especially in big tech. AI moderation empowers all ICs, junior or senior, to do this work without it taking up all of their time.

There's never enough time to do all the work we WANT to do. I would like to include fringe users in our sample but I only have time to focus on the target sample and even that barely. I would like to do that quick study for marketing with the hyper specific question I know I can answer. It doesn't merit its own study, so it doesn't happen. I would like to drip insights to my team between sessions but am drained by moderation. Include that extra segment. Do that mini study. Mobilize the insights when you have confidence, not when you have the final report. Do it by leaning on the AI tool you're hesitant about.

And yes, the D word: democratization. Lines are blurring across knowledge work. With time back from not moderating, scale yourself beyond your research skillset; author the first draft of the strategy rooted in your insights and circulate for feedback. Bring your designer a UI iteration 2.0 that folds in the feedback and cut the back and forth.

I'm not saying every study is a fit for AI moderation. It never will be. But it deserves consideration for a section of your projects. Sharing your core job with AI, not just parts of it, is an identity shift.

If you haven't tried an AI moderator yet (Outset, Listen Labs, Strella, Genway AI): what's your biggest fear or concern? If you have: what's your experience?

#UXResearch #UXRtools #AIModeratedResearch

View on LinkedIn ↗