Context
An India-first crypto derivatives exchange — futures and perpetuals, INR deposits and withdrawals, registered with the national financial-intelligence unit, high leverage on the headline products. Two people sit in front of it: a trader who already runs equity F&O and wants depth and execution, and a salaried first-timer who came for the domestic on-ramp and is frightened of the leverage. My scope was the public site.
The screens below are not images. Every one is live HTML I wrote, rebuilt from scratch with invented data — and every one is the direction I proposed, never the product as it stood. There is no “before” on this page, because a before would be someone else’s live product annotated without their say-so.
The problem
Adopt what leadership admired about the reference, without the result looking like a copy. That was the brief, and the difficulty in it was not visual.
The benchmark encoded a different product. A global spot exchange sells scale to everyone; this was a leveraged-derivatives platform operating under Indian registration, where the homepage carries an obligation as well as a pitch. Borrowing the reference’s confidence would have imported the wrong standard of proof — and the brief, as written, made that borrowing the goal.
Deepening the resemblance would have satisfied the brief and made the actual problem worse.
What I found
Three audiences, and only two that mattered: the leverage trader who wants density, and the first-timer who needs reassurance. Every serious decision on the page turned out to be a negotiation between them, and the aspirational benchmark resolves that badly — it simply overwhelms the newcomer.
The real competitive set wasn’t global. It was the domestic derivatives platforms: same INR rails, same registration story, same customer. Against them the product already owned two things peers struggle for — a distinct identity and an India-first compliance story. What it wasn’t doing was spending them. The registration story sat at the foot of the page rather than where a nervous first-timer decides; the first screen carried no live market data; the figures were set in a face built for display rather than for scanning. And a leveraged product has to carry its risk framing at the point of the leverage claim — the principle I ended up designing hardest to.
The finding that changed the project came from reading the site end to end. It felt familiar. Not like a product with a point of view — like something I had already seen somewhere else. The competitor teardown confirmed it: the resemblance was there before any redesign started.
And what I couldn’t see, I wrote down as not seen. No analytics access and no test account, so KYC, deposit and first trade — the highest-stakes screens in the product — went into the audit marked not assessed rather than guessed at.
No analytics access · no test account
Why this page exists
The four steps on the right are where a first-timer actually decides whether to stay. They are also the four I could not reach: no analytics to show me where people fell out, and no test account to walk the path myself.
So they are recorded as not assessed, rather than assessed from the outside and written up as findings.
What that costs the recommendation
Everything proposed for the public site stands on evidence I gathered. Nothing proposed for the funnel does, because there is none — which is why the strategy asks for analytics access and a test account before phase two rather than proposing screens for steps 03 to 06.
- The limit is stated before any finding is. No analytics and no test account, so the audit could reach the public site and nothing behind the sign-in — and a reader knows that before they read a word of what I concluded.
- The coverage rule stops dead where the evidence stops. Two steps of six were assessed; the rest of the bar goes dashed rather than running on and quietly implying it was covered.
- Verify, KYC, INR deposit, first trade. The four steps where a first-timer actually decides whether to stay, and the four I could not reach — drawn as not assessed rather than left off the diagram, because a funnel with the hard half missing is the more flattering picture.
The decision
I rejected the benchmark in the brief.
Not the admiration behind it. The reference does data density and live-market feel better than anyone, and those are industry conventions nobody owns. What I rejected was the benchmark as the standard of success — because the site I had been asked to improve already read as borrowed, and that was the defect.
So I wrote the real competitive set into the document, named it as the set we would be judged against, and drew the consequence: for a leveraged product under a registration regime, the homepage’s job is proof, not spectacle.
Live market data in the first screen, because a venue showing no prices reads as illiquid. Regulatory and fund-security cues promoted out of the footer, because that is the one advantage an offshore competitor cannot copy. A financial-grade type system with tabular numerals — figures a person cannot scan are figures a person cannot trust. And risk framing placed next to the leverage claim, designed rather than disclaimed.
Deposit in INR from a bank you already use. Withdraw the same way. No offshore transfer, no stablecoin round trip, no third currency.
- Live prices, above the headline. A venue that shows no market data reads as a venue with no volume — so the strip sits in the first screen rather than three scrolls into it.
- The claim and what it costs are one component. Split them and the number ends up on the homepage while the risk ends up in the footer. Here the sentence about a 2% move sits at the same optical weight as the number that attracted you, and the account opens at 2× by default.
- Two audiences, two doors. A first-timer starting at ₹500 and an F&O trader who wants the book are not one average visitor, and a single call to action serves neither of them.
- Registration, rails, segregated funds, tax — out of the footer. This is the one advantage an offshore competitor cannot copy, so it belongs where a nervous first-timer is actually deciding.
| Market | Last price | 24h change | Funding | 24h volume |
|---|---|---|---|---|
| BTC / INRBitcoin · spot | 58,42,100.00 | ▲ +1.84% | — | ₹184.2 Cr |
| BTC-PERPPerpetual · INR margined | 58,39,900.00 | ▲ +1.79% | +0.0091% | ₹96.4 Cr |
| ETH / INREthereum · spot | 2,71,480.00 | ▼ −0.92% | — | ₹61.8 Cr |
| ETH-PERPPerpetual · INR margined | 2,71,205.00 | ▼ −0.97% | −0.0042% | ₹38.1 Cr |
| SOL / INRSolana · spot | 14,206.40 | ▲ +3.51% | — | ₹12.7 Cr |
| MATIC / INRPolygon · spot | 38.94 | ▼ −2.16% | — | ₹4.9 Cr |
| XRP / INRXRP · spot | 182.65 | ▲ +0.44% | — | ₹3.4 Cr |
| DOGE / INRDogecoin · spot | 9.71 | ▼ −5.08% | — | ₹2.1 Cr |
- Buy by amount, not by quantity. A first-timer thinks in rupees; asking them to enter 0.00085586 BTC is asking them to fail. The quantity is worked out for them, and shown to eight places so nothing is quietly rounded away.
- One weight, one alignment, all the way down. Every figure column is tabular and right-aligned, so the rupee, the lakh and the paisa land in the same place on every row. Figures a person cannot scan are figures a person cannot trust.
- Lakh and crore, not million and billion. A customer converting the site’s numbers in their head is a customer being asked to do the exchange’s work.
- Direction is stated three times — glyph, sign and colour — because colour is the cue a reader is least likely to be able to use. And the site’s accent is kept off this table entirely: the accent is a warm red and a falling market is a red, and one screen cannot carry two meanings of the same colour.
What I rejected to get there: the straightforward version of the job. Restyle the reference in our colours and ship it. It was what had been asked for, and it was weeks of work rather than months.
The trade-off
Time, and the shape of it. The weeks went into documents rather than the interface I had been asked for. The governance I proposed — one accountable approver, gates signed in order, bounded revision rounds — assumed a review process that didn’t exist.
And on the question that decided everything — who the site is primarily for — I didn’t take the decision. I put it to leadership with a signature line under it, which left the work depending on someone else’s answer.
What happened
Nothing was reviewed. Eight documents and several rounds of design went up, no response came back, and the earlier landing page is still live today.
So there is no outcome to claim here — no conversion figure, no adoption number, nothing measured. The six decisions the strategy asked for were never taken, which means the primary-user question is still open.
What I’d do differently
The first redesign I produced was assembled, not discovered. I took references from Pinterest and Dribbble, kept the existing content exactly as it stood, restyled a few sections well and presented that as a redesign.
When I corrected course, the correction was still desk work — heuristics, competitor teardowns, three personas written from the product’s positioning rather than from anyone I had met. On a product about people risking money with leverage, I never spoke to a single person who trades. The audit was right about the interface and silent about the user.