marketfeed. Wealth
May 2025 – Mar 2026 · Product Designer
marketfeed app (iOS / Android)
A 0→1 mutual fund product that opened marketfeed to a new audience of first-time investors, starting at ₹1,000, not ₹4 lakh.
Impact
- ~90%
- Onboarding completion
- 2 min
- Acc. opening time (95% autofilled)
- 250+
- Signups
- ₹14L
- Assets under management in beta
Role
Product Designer. Planned all seven flows end to end and owned the UI, the branding, and every icon and illustration in them.
Team
Sharique Samsudheen (CEO)
Sooraj E (CTO)
Vivek Krishna (Head of Design)
Nikhil DS (Product Design)
Smarak Das (Sr Engineer)
Ijas Ahammed (Front-end Developer)
Problem
The audience we were turning away
marketfeed's flagship product, Automated Trading, served 2,500 active users at a ₹4 lakh minimum ticket size. But the top of our funnel looked nothing like that: 2.3M+ YouTube subscribers and 741K+ on Instagram, mostly first-time investors with ₹1,000–₹10,000 to start. Feedback calls with our most active traders surfaced the same request repeatedly: manage our long-term wealth too. Users who trusted us with lakhs in trading wanted a place for the rest of their money, and the audience who couldn't afford trading wanted a way in.
Mutual funds closed both ends: SIPs from ₹1,000, regulated and trusted, and aligned with where Indian retail is heading under AMFI's Viksit Bharat 2047 vision.
The business model shaped every design decision
marketfeed Wealth earns ~0.6% commission p.a. on AUM through a regular-fund advisory model.
Constraints
- One designer, nine months, seven flows — planning each flow end to end, then the UI, the branding, and every icon and illustration in it.
- Regulated domain: question content, disclosure rules, and consent requirements owned by compliance/PMS, not design.
- Two external vendors in the critical path — Decentro for KYC, Fintech Primitives as the mutual fund distributor — and either can fail mid-flow.
What actually got built
Before the decisions, the shape of the thing. Wealth is seven flows sitting inside an app that already existed, each one handing off to the next, and each one able to fail on somebody else's infrastructure. One pass through the shipped product, start to money moving:
Dashed pill — a handoff to a system we don’t control.
- Discovery — an in-app pitch page that explains what a plan is, what it costs, and who is behind it.
- Risk profiling — six questions, one per screen, ending in an investment style.
- Plan curation — the matched plan, its funds, its allocation, and the two you weren't matched to.
- Account opening & KYC — PAN, address, income, declaration, bank, nominee, eSign.
- Purchase — amount, mandate, OTP, registration.
- Dashboard — portfolio when it works, recovery surface when it doesn't.
- Redeem — standard, regulated, deliberately unremarkable.
The vendor layer is the part you can't see. Two external systems sit between a user and a registered SIP: Decentro for KYC and Fintech Primitives as the mutual fund distributor, with the AMC and fund data behind every number on screen. Either can time out, half-succeed, or return an error we don't control — which is why the last two sections of this study exist.
Design
Five chapters, in the order a user meets them. Each one is a problem, the options, what I shipped, and the board it came out of — all out of one component file, built as the flows were drawn.
1 · Getting in
The problem. The first screen was called Risk Assessment. People read that as a form to fill in, so they left before the product had shown them anything.
What I shipped. I renamed it Find your investment style and laid it out as 3 simple steps. Same six questions, same regulatory purpose — it just stopped sounding like work. ~90% of people who reach the intro now finish it.
The page behind the button. A first-time investor is deciding whether to trust us, not which fund to buy. So the intro is a full page, and it answers four questions in order: what a plan is, why us and not a broker, what it costs, and who is on the other end. On cost it shows ₹999 struck through, ₹0, then says plainly that we earn a commission from the fund house and it is already priced into the fund's returns. We make money on AUM. Saying so is better than being found out.
Then the six questions. One per screen, clear progress, no scrolling form, ending in a result screen that names the style rather than scoring the user.
2 · Matching, not gating
Where the six questions land. They end in one of three investment styles, and each style maps to one plan: Conservative to Flow, Moderate to Rise, Aggressive to Surge. The app names the style, explains what it means, and takes the user straight to the plan built for it.
ConservativeMatched to Flow — safety over high returns, minimal risk exposure.
The problem. That last part is where it gets difficult. Regulation requires us to show every plan to everyone — a Conservative user must be able to see Surge. So the profile can never be the only thing deciding what a user buys. Two obvious answers, both wrong:
- Show everything and gate nothing. The profile becomes decoration, and someone buys risk they were just told to avoid.
- Block anything that doesn't match. The profile becomes a cage, and people work around the product instead of using it.
What I shipped. Show all three plans, and put a consent step in front of any purchase that doesn't match the profile. A Conservative user can buy Surge — they just have to say so first. The profile sets the default; it doesn't make the decision.
The same page, three times. Every plan page carries the identical structure — recommendation, funds with expense ratio and allocation, the composition donut, why it suits your profile, who curated it, then the other two plans. Nothing is hidden from a Conservative user that an Aggressive user can see; only the recommendation changes.
And the gate is a matrix, not a screen. Three mismatches are possible, and each sheet names both the profile the user was given and the plan they picked, so the consent is about their specific choice.
Why it works both ways:
- The regulator gets what it needs: every plan visible, and consent recorded at the exact point the risk transfers.
- The user keeps the decision: friction sits on the risk choice itself, and nowhere else in the flow.
3 · Account opening: when the vendor is the design problem
The problem. Beta funnel data showed drop-off concentrated at one point: after plan discovery, at account opening. Users completed onboarding, finished risk profiling, found their plan, and died at KYC. Our vendor's (Fintech Primitives) KYC step was long, manual, and tiring, and no amount of screen-level polish on our side could fix a flow we didn't control.
Options considered.
- Redesign our wrapper around the existing KYC flow → cosmetic; the fatigue was inside the vendor's steps.
- Accept the drop-off as a cost of compliance → contradicted the entire retention-based business model.
- Switch the KYC vendor post-launch → painful, but the only option that addressed the cause.
What I shipped. In the months after launch, I used the beta drop-off data plus the design read on user fatigue to make the case to swap the KYC layer to Decentro, while keeping Fintech Primitives as the underlying MF infrastructure where it was strong.
Result: ~2-minute account opening, 95% of data autofilled from just a phone number.
The vendor swap bought the data; the design decides what to do with it. A prefill of 95% is only worth something if the interface stops asking. So KYC isn't a form — it's four collapsed, already-ticked cards: Personal Details, Income, Address, Declaration. The label reads Review your pre-filled details, not Enter your details. The user's job is to confirm, not to type.
Then the 5%. A prefill system is only as good as its failure mode, and the fetch degrades in three distinct ways — no Aadhaar name and number, no personal/income/address/declaration, no bank details. Each tier keeps the same page and opens only what's missing, with the field marked and the keyboard already up.
4 · Purchase: a calculator instead of a default
The problem. SIP minimum is ₹1,000. Hard-defaulting to it would anchor users low and teach us nothing about what first-time investors actually start with.
What I shipped. Free entry for any amount above the minimum, quick-pick pills at 2×, 5× and 10× the minimum (₹2,000 · ₹5,000 most popular · ₹10,000), and a live SIP calculator that opens the moment the user types, projecting 10-year returns in real time.
The calculator turns an abstract question ("how much should I invest?") into a concrete one ("₹5K/month for 10 years = ₹X"). And letting users choose freely was a research bet: we learned what first-time investors actually start with, which now shapes every future default.
Then everything after the amount. Frequency, SIP date, mandate, OTP, registration — the part of the flow where the user has already decided and just needs it to finish.
5 · Dashboard: the thing we got wrong, and how we fixed it
What we got wrong. The MVP dashboard assumed the happy path: you invest, you see your portfolio. Real beta behaviour broke that assumption immediately. Users killed the app mid-onboarding, emandates failed, purchases partially succeeded and stuck at "loading." They landed on a portfolio dashboard with nothing in it and no way forward. In a flow that hands off to KYC, to a distributor, to a bank mandate and to an AMC, most users were on an unhappy path at some point, and we'd designed for none of them.
What I shipped. Rebuilt the dashboard as a recovery surface:
- 3-step progress card: detects wherever the user dropped off (risk profile → account opening → purchase), shows sub-step progress inside account opening, and gives one clear CTA to continue.
- Action Center: surfaces every stuck state — failed payments, emandate failures, partial purchases, and loading-stuck transactions — each with a plain-language status and a resume action.
And once they're through, it has to be worth arriving at. Portfolio value, per-plan performance, the funds inside it, every transaction, and the fund-level detail one tap deeper — including the third-party fund data users check us against.
Why it mattered. This was post-launch iteration driven by observed behaviour, not assumptions, and it's what kept beta users from abandoning entirely when something (inevitably) failed.
The unhappy path is the product
Two external vendors, a bank mandate and an AMC sit between a user and a registered SIP. In beta, a majority of users hit at least one failure — a mandate that didn't set up, a registration that half-succeeded, an OTP that never arrived, a fund list that wouldn't load. None of those are edge cases at that rate; they're the product.
So every failure screen follows the same three rules: name what failed in the user's language, never in the vendor's; say what is and isn't lost — a partial SIP registration lists which funds registered and which didn't; and carry exactly one action, always a retry or a resume, never a dead end with a support number.
Where I chose not to design
Risk profiling questions are owned by compliance and the PMS team; the content is dictated by what's legally needed to categorise a user. Re-litigating the questions would have burned weeks for marginal gain. My lever was length and pacing: pushed the count down to 6 questions, one per screen, clear progress, no scrolling forms.
Redeem is a standard, well-understood, regulated flow. It worked in beta without surprises. We shipped it clean and moved on.
The time saved on both went into the dashboard rebuild, which is where the product actually needed a designer.
What I learned
Reframing beats redesigning. The highest-impact change was a rename. Users finish what they started; they don't start what feels like work.
Vendor choices are design choices. When a third party's UX is killing your funnel, the design fix is replacing the vendor, and the designer should be the one making that case with data.
The unhappy path is the product. In regulated, multi-vendor fintech, most users hit a failure state eventually. Recovery surfaces like progress cards and action centers are where the real UX work lives.
What's Next
Investment Baskets
Curated ETF baskets users can buy in the app, then pledge for a second return.
