<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Divyanshu Seth on Medium]]></title>
        <description><![CDATA[Stories by Divyanshu Seth on Medium]]></description>
        <link>https://medium.com/@divyanshuseth?source=rss-e8c860664971------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*DFWfCDNLFHABRpY6.</url>
            <title>Stories by Divyanshu Seth on Medium</title>
            <link>https://medium.com/@divyanshuseth?source=rss-e8c860664971------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 23:20:05 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@divyanshuseth/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Scaling a Credit Card from Zero to 50,000 Issuances in 90 Days — What I Learned About Growth…]]></title>
            <link>https://medium.com/@divyanshuseth/scaling-a-credit-card-from-zero-to-50-000-issuances-in-90-days-what-i-learned-about-growth-system-92114bd08b45?source=rss-e8c860664971------2</link>
            <guid isPermaLink="false">https://medium.com/p/92114bd08b45</guid>
            <dc:creator><![CDATA[Divyanshu Seth]]></dc:creator>
            <pubDate>Tue, 05 May 2026 13:58:55 GMT</pubDate>
            <atom:updated>2026-05-06T05:41:02.635Z</atom:updated>
            <content:encoded><![CDATA[<h3>Scaling a Credit Card from Zero to 50,000 Issuances in 90 Days — What I Learned About Growth Systems</h3><p><em>By Divyanshu Seth</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QkVNZPaJs8SOhc6CKK_FBA.png" /></figure><h3>The Setup</h3><p>In October 2024, I joined slice as Performance Marketing Lead. slice had just received its banking licence and was transitioning from a fintech app into a full-stack consumer bank — launching credit cards, savings accounts, and fixed deposits alongside its existing borrow (unsecured loan) product.</p><p>By the time I took on the credit card mandate, I had already scaled two other products at slice — grown Borrow (a mini unsecured loan with its own application and approval gate) 2x to ₹3.8Cr/month at a CAC under ₹1,600, and launched and scaled Digital Savings Accounts to ₹3Cr/month while holding account-open CAC under ₹400 and add-money CAC under ₹1,200. Both had approval gates, risk dependencies, and the same core challenge: optimise for the issued product, not just the application. Each sharpened my thinking on signal quality, audience sequencing, and what efficient scale actually looks like in a regulated financial product.</p><p>The credit card was harder than both — higher ticket, stricter credit policy, longer funnel from application to issuance. And the one I’m most proud of: <strong>CAC halved in 30 days, then 30x spend scale over the next 90 — without CAC moving.</strong></p><h3>The Challenge (Situation)</h3><p>Launching a financial product from zero is fundamentally different from scaling an existing one.</p><p>There’s no historical data to anchor bids. No proven creative format. No signal on which audience segment converts at what approval rate. The algorithm has nothing to learn from.</p><p>But the harder constraint wasn’t data — it was the nature of the product itself.</p><p>A credit card isn’t just an app install or a lead form. The acquisition funnel has two distinct gates: <strong>application</strong> (top of funnel, marketing-owned) and <strong>approval</strong> (bottom of funnel, risk and credit policy-owned). Optimising for one without understanding the other is how you end up with great CTRs and terrible unit economics.</p><p>The goal was clear: scale to meaningful monthly spends while keeping CAC at or below ₹3,000 per issued card. The implicit constraint was equally clear: don’t just drive applications — drive <em>approvable</em> applications.</p><p>With no prior data, no proven playbook, and a 90-day informal timeline to demonstrate traction, the pressure was real.</p><h3>What We Tried First (and Why It Wasn’t Enough)</h3><p>The instinct in any new product launch is to run broad acquisition — wide audiences, multiple creatives, let the algorithm figure it out.</p><p>We did that in week one. The results were predictable: high volume, high variance, and no clarity on what was actually driving approved cards vs. rejected applications.</p><p>The core problem: we were optimising for applications, but the business metric was issued cards. Without connecting those two events in our tracking setup, we were essentially flying blind.</p><p>This is where most performance marketing teams stop — they report on what they can measure and hope the downstream metrics follow. I didn’t want to do that.</p><h3>The Real Work (Task + Actions)</h3><p>We restructured around three parallel workstreams:</p><h3>1. Build the Signal First</h3><p>Before scaling spends, I worked with our tech team to implement event-level tracking through MMP (AppsFlyer) and Firebase — specifically whitelisting two events: <strong>application submitted</strong> and <strong>card approved</strong>.</p><p>This sounds basic. It isn’t. Getting risk, credit policy, and tech teams aligned on what constitutes a “quality” event — and then actually firing those events reliably from the backend — took two weeks of coordination that most marketing teams skip entirely.</p><p>Once we had clean approval signals feeding back into Meta and Google, the algorithms had something meaningful to optimise against. CAC started dropping almost immediately.</p><p><strong>Learning:</strong> In financial products, the most important work a performance marketer can do is upstream of the campaign — in the tracking and signal architecture.</p><h3>2. Warm Audiences Before Cold</h3><p>The second unlock was counterintuitive: instead of going wide with cold acquisition immediately, we activated slice’s existing user base first.</p><p>slice already had millions of users on its borrow and savings products. These users had completed KYC, had a relationship with the brand, and had demonstrated financial behaviour. Their approval rates were structurally higher than cold audiences.</p><p>We built CRM campaigns across Push Notifications, WhatsApp, and RCS to this base — with messaging tailored to their product usage. A borrow user got a different pitch than a savings user. The CAC on this cohort was 40–50% lower than cold acquisition, and approval rates were significantly better.</p><p>This gave us two things: early volume at low CAC, and a clean benchmark to measure cold acquisition efficiency against.</p><h3>3. Creative Velocity as a Scaling Lever</h3><p>Once we had signal quality and warm audience coverage sorted, we pushed aggressively on creative testing to unlock cold audience scale.</p><p>We built a three-format creative pipeline:</p><ul><li><strong>Meme/relatable content ads</strong> — high thumb-stop, 2.5x CTR vs baseline, lowest CPI. Used for top-of-funnel scale. The trade-off: click-to-application rate was lower because the audience wasn’t always high-intent. Used these to buy cheap reach.</li><li><strong>AI-generated variants</strong> — used to rapidly test hooks, formats, and messaging angles at 3–4x the velocity of traditional design. Most were average, but winners emerged with 15–20% lower CAC than baseline.</li><li><strong>Agency-shot UGC (script-led)</strong> — highest-quality, trust-building format. Lowest CAC at scale. Best approval rates because the storytelling pre-qualified intent. Became our highest-spending ad group by month two.</li></ul><p>The insight: different creative formats serve different funnel jobs. Memes buy reach cheaply. AI ads accelerate testing. UGC converts at scale. Running all three simultaneously — rather than picking a “winner” — gave us both efficiency and volume.</p><p>We ended up with 200+ static creatives and 50+ videos in active rotation, with 20+ influencers onboarded for social proof and reach amplification.</p><h3>The Outcome (Result)</h3><p>By end of March — roughly 4 months from first spend — the numbers told a clear story:</p><ul><li><strong>December (launch, MTD):</strong> ₹15L in spends, CAC at ₹6,600+ — no signal, no system, algorithm learning from scratch</li><li><strong>January:</strong> ₹34L in spends — infrastructure month. Signal built, warm audiences activated. CAC brought down to ~₹3K</li><li><strong>February:</strong> ₹1.68Cr in spends — cold acquisition kicked in with clean signals. CAC held stable</li><li><strong>March:</strong> ₹4.5Cr in spends — <strong>30x in 90 days</strong> (from ₹15L monthly run rate at start of January), 50,000+ overall card issuances. CAC remained at ~₹3K</li></ul><p>The two-act structure matters: first we fixed the CAC, then we scaled without breaking it.</p><p>Going from ₹6,600 CAC at launch to ₹3,000 CAC in 30 days wasn’t luck — it was the tracking infrastructure, warm audience activation, and early creative learnings all clicking into place simultaneously. Everything after that was about holding that efficiency while pushing volume aggressively. Scaling 30x without CAC regression is only possible if the underlying system is genuinely working — not just the algorithm getting lucky on a good week.</p><h3>What I’d Do Differently</h3><p>Two things:</p><p><strong>Start the tracking work on day one, not day five.</strong> We lost the first two weeks of January because I underestimated how long backend coordination would take. In a new product launch, the first month should be infrastructure-heavy — signals, tracking, baseline measurement. Campaigns can wait. The ₹15L we spent in January while still fixing tracking was largely wasted learning budget.</p><p><strong>Involve the risk team earlier in creative briefing.</strong> Some of our early creatives drove high application volume from audiences that had structurally low approval rates — users with thin credit profiles or mismatched income signals. If we’d looped in risk during the creative strategy phase (not just campaign analysis), we’d have qualified intent earlier and reduced wasted spend in weeks 3–4.</p><h3>The Broader Pattern</h3><p>Every time I’ve scaled a product efficiently, the work has followed the same pattern:</p><ol><li><strong>Fix the signal before scaling the spend</strong> — garbage in, garbage out</li><li><strong>Use warm audiences to build a benchmark</strong> — don’t go cold-first if you have an existing base</li><li><strong>Run creative formats in parallel for different funnel jobs</strong> — don’t optimise for one metric across all creatives</li><li><strong>Make downstream teams (risk, product, CRM) part of the growth loop</strong> — acquisition doesn’t end at the click</li></ol><p>These aren’t frameworks. They’re the things I’ve had to learn by getting them wrong first.</p><p><em>Divyanshu Seth is a performance marketing professional with 10+ years of experience scaling consumer products across fintech, edtech, and consumer apps in India.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=92114bd08b45" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>