Stock broking app marketing strategy: the launch plan for India in 2026

India has roughly 4.57 crore NSE active clients and the number is falling — down 7% across FY26, per NSE data reported by Angel One. The top five brokers hold over 67% of them, Groww alone 29%, and discount-broking revenue is contracting because SEBI's F&O measures gutted the product that paid for everything.

The August 2026 picture, from NSE data reported by Entrackr:

Groww has roughly double Zerodha's active base; Paytm Money is the only name going backwards.

The incumbents are large — Groww's FY25 revenue was ₹3,901.7 crore (about $443 million), Angel One's ₹5,238.3 crore (about $595 million) — but both rest on a product line Inc42 reports historically supplied 70–90% of discount-broker revenue: futures and options. That line is under direct regulatory attack. F&O notional volumes fell 29% in FY25; Zerodha reported a 40% revenue decline in Q1 FY26, Angel One 19%, Groww's fees line 17.4%. SEBI's own data shows more than 90% of individual F&O traders lose money — the reason for the crackdown, and the reason performance advertising here is dead.

The structural fact that changes your strategy: you are launching into a contracting market where the incumbent revenue engine is being dismantled and the regulator has proposed banning referral incentives and install rewards. You cannot buy your way in with a bonus. You win on cost, product breadth and onboarding friction — conveniently, the only creative territories left legal.

This post plans India only. US figures appear later purely as benchmark contrast, because the one genuinely good Apple Ads panel is American. Both stores classify broking apps as Finance.

You are acquiring someone who will spend three to seven days and real friction becoming a usable customer, and the event that predicts revenue is the first funded trade — not install, not registration, not even KYC completion. KYC is the expensive step, funding the intent step, the first trade the money step.

The full ladder:

install
  → mobile / OTP signup
  → PAN entry
  → KYC initiation
  → Aadhaar e-KYC or DigiLocker
  → bank linkage + penny-drop verification
  → e-sign
  → account opened / demat activated
  → first fund transfer in
  → first trade executed
  → second trade within 30 days
  → active client (traded at least once in 12 months)

Two things about it reshape the whole plan.

First, the drop-off is front-loaded into a regulated step you cannot redesign away. Business of Apps' finance benchmarks, updated 7 January 2026 from their Fintech App Report 2025, put activation at roughly 14% by day 30 across finance. For broking, read that as the KYC funnel: roughly 86% of installs never become usable accounts. Better onboarding copy, DigiLocker over manual upload and drop-off messaging move that number, but nothing gets it to 60%, because the steps are statutory.

Read that 14% as a borrowed proxy, not a broking benchmark. It is a cross-finance average spanning payments, neobanks, insurance and lending, and "activation" means different things inside it. For a payments app, activation is one successful transaction, which a motivated user completes in ninety seconds. For a broker it is PAN entry, Aadhaar e-KYC, bank linkage with a penny-drop and an e-sign — a statutory identity chain with four independent failure points and a multi-day tail. The definitional mismatch biases the borrowed number high for broking, while the same source's note that stock trading holds the highest retention in finance and payments the lowest pulls gently the other way. Net of both, the honest planning range is roughly 8% to 18% of installs reaching kyc_complete by day 30, with 14% used below as the midpoint because it is the only figure anyone publishes. If your real rate lands at 8%, every cost-per-account figure in this post roughly doubles, so run the model at both ends before you commit a budget.

Second, the last rung is what the market judges you on. NSE defines an active client as one who traded at least once in the trailing period, so mirror that internally: report trailing-12-month traded accounts by acquisition cohort, not registered users, and accept that it matures 90 days after the spend that created it.

Retention beats the rest of finance without being good: Business of Apps notes stock trading has the highest average retention within finance and payments the lowest, against a finance D30 of 4.2%, exact split paywalled. Adjust's retention set — undated: the source page states no measurement period and the figures match 2022–23-era releases — puts fintech D1 at 22% against a 26% all-category median. Treat both as a directional floor, never as a 2026 measurement and never as a target.

The practical read: bid on kyc_complete early and move to first_trade only when volume clears the platform threshold. Bidding on install buys KYC abandoners at scale.

Four things, one of which has a lead time measured in weeks: your registration numbers, your ad platform financial-services verification, your event taxonomy and your attribution stack. Start the verification clock before you start the build.

Neither Meta nor Google publishes a turnaround time for financial-services verification, so any number here is an estimate rather than a documented SLA. The planning figure that has held up: four to six weeks from first submission to a spendable account, so start at least eight weeks before your intended launch date.

Keep that duration separate from what a delay costs, because they are different quantities and conflating them makes the risk look smaller than it is. The duration is four to six weeks. The cost is a month of committed fixed spend — creative already produced, agency retainers already running, a launch date already communicated — burning against a media start you cannot make. Verification is not expensive. Missing it is.

The verification gates

Meta has required SEBI verification for securities and investment ads targeting Indian users since 2025 — announced 26 June 2025, enforced from 31 July, per PPC Land. It covers Facebook and Instagram, catches global campaigns that reach India, and applies to agencies running on a client's behalf. It requires SEBI registration of the entity that both benefits from and pays for the ad, and in-scope ads must carry a public disclaimer naming that beneficiary and registration number, retained in the Ad Library for up to seven years. Read that as a marketing constraint: your ad account, your billing entity and your SEBI registration must all reconcile before you can spend.

Google Ads gates financial-services advertising in India behind a documented verification, and the policy defines four advertiser paths, which most write-ups collapse into one:

A SEBI-registered broker takes the authorized path, verifying against its own SEBI registration and member codes. An agency or an affiliate publisher running ads for that broker takes the approved-third-party path and needs the broker's written approval on file before it submits. Get this right at submission: switching paths means starting again.

Most paths require third-party verification through G2 Risk Solutions first — obtain verification, receive a unique code, submit it to Google. The wording is unforgiving: business information "must exactly match with the business details available on the relevant financial services registries," and false information means verification revoked with immediate effect plus possible account suspension. A June 2026 rollout extends the model across 24 EEA markets, explicitly covering brokerages and day trading. Put both on the launch plan as hard dependencies with no float.

Play's financial features declaration

Every Play developer must complete the financial features declaration, including those with no financial features. Stock trading sits in Trading & Funds; your category must be Finance, with clear terms disclosure and a privacy policy matching your Data Safety form. Google also restricts what a financial app may access: contacts, photos, location, phone numbers and external storage are prohibited — stricter than general Play policy, so kill any contact-list referral flow now. If you attach personal lending, your app must appear on the RBI's list of digital lending apps deployed by regulated entities, with every partner named.

And the production-access gate that catches first-time Android publishers. A personal Play Console account created after 13 November 2023 must run a closed test with 12 testers who opt in for 14 continuous days before it can apply for production access. Organization accounts are exempt. Testers who opt in for fewer than 14 days and then opt out do not count, and if a tester opts out and comes back, the clock restarts. Review is typically up to seven days on top of that. Older guides say 20 testers — outdated. Start recruiting testers at least six weeks before your intended launch date: this gate has no appeal path and no expedite.

Event taxonomy

Define these once, before either SDK goes in:

Seven events, one definition each. The common failure is Firebase calling it account_opened, the Meta SDK CompleteRegistration and the MMP af_complete_registration, then three weeks of arguing about which number is real.

Attribution

Google-only, Firebase is free and sufficient. The moment a second paid network exists an MMP pays for itself — Firebase does not feed Meta, and two SDKs give two incompatible definitions of the same event. With roughly 31% category fraud exposure, the fraud detection alone justifies one in a way it would not in, say, habit trackers. Mark key events in Firebase before launch: GA4 caps you at 30 per property and marking is not retroactive.

Copy the instrument-noun stack. Indian broking leaders write no taglines; they jam every high-volume instrument name into the title and short description, because those are the searched terms. Then they carry registration disclosures in the description, because SEBI treats store metadata as advertising.

Groww's live listings, India:

No verb, no brand promise, nine product nouns in one Play short description. Each is a search term with real volume.

India head terms from leader metadata: demat account, demat account opening, stock market app, share market app, trading app, mutual fund, SIP, IPO, F&O, options trading, intraday, nifty, sensex, plus brand defence on the three leaders. These are extracted from leader listings rather than a volume tool, so treat the ordering as a hypothesis to test.

Store conversion in finance is unusually good: Business of Apps puts page-view-to-install at 32.8% on iOS and 19.7% on Play, against dating at 18.2% and 20.0%. That iOS number argues for disproportionate effort on your iOS screenshots.

The metadata compliance trap

Groww's opening description line reads: "Groww is India's No.1 Stock Broker, trusted by 4 Crore+ Indians." That is a comparative superiority claim in a regulated category, and under SEBI's proposed code — draft, not notified — comparisons that disparage competitors would be banned outright. Store metadata is advertising. Do not copy that construction without substantiation your counsel will defend.

Do copy the disclosure block. Groww's App Store text carries its SEBI registration number (INZ000301838), member codes (BSE-6699, NSE-90187, MCX-57420), segments, registered office address and digital-lending-app disclosure. Build it once, use it in both listings.

Which disclosures are mandatory on a broking ad?

The risk warning, the retail F&O loss statistics, and investor-grievance routing — alongside the registration block above. These are in force today, not proposed, and they are the part of broking creative that marketing writing most often skips.

Every Indian broking advertisement carries the standard market-risk warning. The line, verbatim:

"Investment in securities market are subject to market risks, read all the related documents carefully before investing"

Reproduce it exactly as issued rather than tidying the grammar. It belongs on store screenshots, video end cards, static creative, the website and your own onboarding — anywhere the material functions as an advertisement.

Three more that belong in the same module:

• Retail F&O loss statistics. SEBI separately requires brokers to display the published statistics on individual traders' losses in the equity derivatives segment. This is where the ">90% of individual F&O traders lose money" figure used earlier in this post comes from — and it is not merely market colour explaining why performance creative is dead. You are required to display it. Plan the disclosure surface into the creative template from the first storyboard.
• Investor grievance routing. Name your investor-grievance email address and the escalation path: SEBI's SCORES portal for complaints, and Smart ODR for online dispute resolution. These sit in the same block as the registration numbers.
• Registration and member codes, as in the Groww template above.

Build the whole thing once — risk warning, F&O loss disclosure, grievance and SCORES/Smart ODR routing, registration and member codes — as a single counsel-approved module, then drop it into the store listing, the site footer and every ad format. Design creative around a persistent disclosure strip from the storyboard stage. Retrofitting it into a finished six-second vertical is how brands end up with unreadable eight-point type over the last frame. The exact wording and placement rules are set by circular and they move, so confirm the current form with your compliance counsel before you ship.

Run Apple's four-campaign structure — Brand, Category, Competitor, Discovery — with exact match and Search Match off on the first three, and expect Finance to be one of the most expensive categories on the platform. AppTweak's 2025 panel of roughly 3,500 apps, 50,000 campaigns and $1 billion in spend across 38 countries puts US Finance CPI at $8.44 on search results — roughly double the US all-category median of $4.06, behind only Sports and Games, with cost per tap at $3.55 against a $1.91 median. In the US, Apple Ads is a high-intent, low-scale channel.

India is far cheaper: CPI $0.89 against a $1.80 global median, tap-through 5.2%, conversion 48%. AppTweak publishes no India Finance cut, so a planning number has to be constructed — and the construction has a defect you should see before you lean on it.

Why the ₹163 planning CPI is a range, not a number

The two figures in the ratio are not measured the same way. The $8.44 Finance CPI is measured on search results specifically. The $4.06 US figure is the all-category, all-placement median, blending in Today tab, search tab and product pages, which convert very differently — AppTweak's own tap-through spread across placements is 7.4% to 0.29%. So the 2.08 ratio mixes a Finance category premium with a placement-mix difference, and it is then applied to India's $0.89, which is itself an all-placement number. You are multiplying an all-placement India figure by a ratio that is partly measuring placement mix.

The two honest ends of that:

This post uses ₹163 throughout, and you should read it as "somewhere between ₹78 and ₹165, budgeted at the top." Plan at the ceiling: planning at the floor under-budgets every campaign against a documented platform minimum, and you find out in week two with a campaign that never exits learning. Every threshold and all three budget columns in this post rest on this one constructed number, so the first job of week one is to replace it with a measured CPI and re-run the tables.

The same caution applies to cost per tap, on the single best-measured metric in mobile advertising. AppTweak puts US CPT at $1.91; Adapty's panel of more than a million ad groups across 90 countries puts the same metric at $1.58. A 20% spread between two disclosed first-party panels is the right calibration for how much precision any of these numbers carry. Neither publishes an India Finance cut.

The structure Apple documents:

Add every Brand, Category and Competitor keyword into Discovery as exact-match negatives, so Discovery only spends on genuinely new terms. Budget splits like 30/35/30/5 are convention, not Apple guidance.

Competitor is where the money is and where discipline matters most. Bidding on "groww" or "zerodha" is normal, but your creative and Custom Product Page must not make comparative claims about those brands, because SEBI's proposed code would catch comparisons that disparage competitors. Bid the term; do not name the competitor in the copy.

Custom Product Pages are your highest-leverage asset: up to 70 per app, Marketing role in App Store Connect, search-results ad groups only, editable without a new app version. Build one for demat-account-opening terms leading on zero account-opening cost, one for F&O and intraday leading on the ₹20 (about $0.23) per order flat fee, and one for mutual fund and SIP leading on breadth.

The 2026 bidding tension. Apple launched Maximize Conversions on 25 February 2026 — automated bidding to a target CPA and daily budget, with documented guidance to budget for at least five conversions a day and run at least two weeks before judging. Search Match is mandatory on the automatic ad group, it is unavailable for pre-order campaigns, and target CPA is a weekly average rather than a per-query ceiling. That mandatory Search Match conflicts with the exact-match discipline the four-campaign structure depends on, and Apple has published nothing reconciling them, so in practice you run the manual structure or a Maximize Conversions campaign alongside it in separate reporting lines. Five conversions a day at a broking cost per account-opened is a real commitment — do the sum before switching it on.

Launch App Campaigns for Installs on target CPI, respect Google's documented daily budget floor of at least 50 times your bid, and do not touch the campaign before 100 conversions have registered. Validate the full funnel on Android first, because Android attribution is deterministic and iOS is not.

That is the most important structural decision here after the funnel arithmetic. Android Privacy Sandbox was cancelled in October 2025 — GAID persists and deterministic user-level Android attribution continues. iOS runs on AdAttributionKit and SKAdNetwork: aggregate, delayed, crowd-anonymity-gated. These are not the same channel with a bid multiplier applied.

Google's documented budget floors, which practitioners get wrong constantly:

The learning threshold is a conversion count, not a time window: changes before the first 100 conversions have registered may disrupt learning. Google also advises against selecting more than one action for target CPA, suggests iOS bids around 1.5 times Android, and recommends bidding roughly 20% above baseline target CPI when targeting in-app-action users.

Two silent failure modes cost a fortnight each. Your robots.txt must allow AdsBot-Google and AdsBot-Google-Mobile to reach apple-app-site-association and assetlinks.json. And Android installs need deferred deep links enabled in your measurement SDK or post-install routing silently fails.

On buying burst installs

You will be pitched a burst: 24–72 hour windows, $35,000–50,000 budgets, claimed movement of 5–20 positions, gains fading within two weeks. The independent evidence is worse than the pitch — an ACM IMC 2020 study of real incentivised campaigns put top-chart appearance at 3.1% baseline, 7.5% for vetted incentive platforms and 2.5% for unvetted ones — the cheap tier produced no detectable ranking benefit at all — and MIT's 2025 shutoff experiments found an organic halo of about 8%.

The compact version you cannot skip. Apple added clause 5.6.3 "Discovery Fraud" in its February 2026 Guidelines revision, under the Developer Code of Conduct, where the stated remedy is termination of the Developer Program account rather than app rejection: "Manipulating any element of the App Store customer experience, such as charts, search, reviews or referrals to your app, erodes customer trust and is not permitted." The Section 3 preamble catches agencies too — "or engage with third-party services to do so on your behalf." Google separately states it filters incentivised installs out of ranking systems, escalating to top-chart removal then store removal, so the burst can be silently zeroed while you still pay for it — and the account consequence cascades: "any related Google Play developer accounts will also be permanently suspended." And for finance specifically, AppsFlyer's State of Ad Fraud 2026 puts the finance vertical at roughly 31% fraud, against 11.7% on iOS and 14–15% on Android overall — the highest vertical figure in the dataset as reported — which makes bought installs disproportionately risky here.

The legitimate version produces the same install curve: pre-registration cohort, email waitlist, PR embargo, creator posts and paid campaigns all landing the same day. Same velocity, real users, no policy exposure.

Meta works well for broking volume and badly for broking creative, because every in-scope ad carries a permanent visible disclaimer naming your verified beneficiary and SEBI registration number, archived for up to seven years. Design around the overlay, and clear verification before you plan spend, because it gates the account, not the ad.

Optimization goal. Start on App Installs, Meta's recommended default, then move to App Event Optimization once App Events are flowing. Meta discourages Link Click Optimization if the SDK is installed, and Value Optimization is documented as available "on a limited basis to partners and advertisers" — not universal, so do not plan around it.

The learning phase number. Around 50 optimization events per ad set per rolling seven days is the universally cited threshold — consistent across trade sources but not primary-verifiable, because Meta's Business Help Center is robots.txt-blocked and nobody can cite Meta directly for it. Budget, creative, audience and optimization-event changes all reset it. Worth internalising: raising budget purely to escape the "learning limited" label typically yields only 5–10% improvement, and consolidating into fewer, larger ad sets is the structural fix. In broking, where deep events are scarce, resist running six ad sets when two would clear the threshold.

iOS and the eight-event model. Up to eight conversion events per verified domain, ranked by priority; only the highest-priority event in a session is reported, and reordering pauses affected ad sets for a 72-hour cooldown. A 2026 change most playbooks miss: the standalone Aggregated Event Measurement tab and manual ranking step were removed for many web accounts, but the eight-event model still governs iOS app campaigns. For broking, rank them first_trade, first_funding, kyc_complete, kyc_started, registration_complete, and set it once — every reorder costs three days.

Conversions API for App Events belongs in the launch build. Mandatory parameters: action_source = "app", advertiser_tracking_enabled carrying ATT state, extinfo, and event_id for deduplication. Install events get automatic 90-day deduplication; post-install events do not — and in a category where the money event lands five days after install, that is the gap you close yourself.

Here is the arithmetic that decides your media plan, and it has a budget threshold in it rather than a flat verdict. Below roughly ₹25 lakh a month, optimising to first_trade is not inadvisable but arithmetically out of reach — the event is too rare and the platform floors too high. At ₹25 lakh it is borderline, on one channel, on Android. At ₹75 lakh it is comfortable on Android. On iOS it does not become viable at any budget in this post, for measurement reasons set out below. Under about ₹25 lakh you optimise to kyc_complete and accept it; the budget table at the end scopes each scenario against that line.

The only published anchor is Business of Apps' finance activation figure of roughly 14% by day 30 — your KYC-complete rate off installs. Everything below is a planning model you replace with measured data by week three, stated openly rather than dressed as a benchmark, because no credible India cost-per-KYC figure exists.

Working model from 10,000 installs:

So roughly 5.9% of installs become first trades, and 86% never become a usable account at all.

The app_open rung is in that table deliberately. India is an Android market with a long tail of low-storage devices and metered data, and a material share of installs never open the app once — auto-installs delivered by pre-registration, downloads that finish on a dead battery, and users who install three broking apps in one sitting and open one. Leave app_open out and that loss silently inflates every rate below it: at 85% open, an install-to-registration rate you report as 55% is really 65% of everyone who ever saw the product, and your onboarding looks worse than it is while your acquisition looks better. Instrument it before you draw conclusions about either — and measure your own figure, because 85% here is a model assumption, not a benchmark.

Cross that against the documented platform gates:

At a planning CPI of ₹163 (about $1.85, derived above and explicitly not a published figure), clearing Meta's threshold on first_trade needs about 850 installs a week per ad set — roughly ₹6 lakh a month (about $6,800), per ad set, on one channel. Clearing it on kyc_complete needs about ₹2.5 lakh a month (about $2,840); on registration_complete, about ₹64,000 (about $730).

Google's target CPA floor makes the point more brutally. If your cost per first trade is ₹163 ÷ 5.88% ≈ ₹2,772 (about $32), a target CPA campaign on that event needs at least ten times the bid daily — ₹27,720 a day, roughly ₹8.4 lakh a month (about $9,500) — for one campaign, before Meta or Apple.

Those are the numbers you came for, so state them rather than leaving them inside a floor derivation:

The ranges are wide because they are honest. Both inputs are constructed — a CPI extrapolated from an Apple Ads panel that publishes no India Finance cut, and an activation rate borrowed from a cross-finance average that defines activation differently for each sub-vertical. Nobody publishes either number for Indian broking. Quote the midpoint internally, quote the range to whoever approves the budget, and replace both with measured figures by the end of week three.

Then the iOS problem, which most playbooks omit entirely. AdAttributionKit gives three conversion windows from first launch: days 0–2, 3–7 and 8–35. Fine conversion values exist only in the first postback and only at crowd-anonymity Tier 2 or 3; windows two and three are always coarse, and at Tier 0 they produce no postback at all.

A broking first trade does not happen in the first 48 hours. Your money event lands in window two or three, where you get coarse values at best and nothing at Tier 0 — and a new app at launch volume is exactly the profile sitting in the low tiers. On iOS, at launch, first_trade is not merely expensive to optimise toward; it may be structurally unmeasurable regardless of settings. That is why every scenario in the budget table below caps iOS at kyc_complete even where Android has moved on to first_trade — the constraint is measurement, not money, so a bigger budget does not fix it.

So: encode kyc_started or kyc_complete into your fine conversion value inside the first two days, because that is the only signal you will reliably receive. Validate the full ladder on Android, then use the Android-measured KYC-to-trade ratio to value your iOS KYC conversions.

The sequencing that works: launch on install or tCPI, clear 100 conversions on Google or roughly 50 a week on Meta, move down one rung, re-clear, repeat. Never skip a rung. Most broking launches stop at kyc_complete for the first quarter, and that is the correct answer, not a failure.

The returns territory is closed, the incentive territory is closing, and what remains is cost, speed, product breadth and education. Build on those four and you will still have a campaign after the SEBI code is notified.

What is in force today, and what is only proposed?

This is the distinction the category's marketing writing gets wrong most often, so take it slowly. SEBI's Common Advertisement Code is a draft. The comment window closed 14 July 2026 and the code carries a six-month transition from notification, per Medianama. As of September 2026 it has not been notified, so nothing in it binds a stock broker yet, and provisions may change before it does. What is in force is a separate, narrower advertisement code binding investment advisers and research analysts — not brokers.

The testimonial question, stated precisely, because it is the one most often garbled: testimonials are restricted today for investment advisers and research analysts under the April 2023 code, and proposed for stock brokers under the draft CAC. A broker running testimonial creative in September 2026 is not in breach of a broker advertising code — it is running a format the draft would remove, on a timeline of notification plus six months. Plan as though it lands. Do not tell your team it already has.

What would the proposed code change about creative?

If the CAC is notified in its current form, these are the provisions that reshape what you can make. Every one is conditional — none binds a broker today, all would take effect notification plus six months, and the draft may change before then.

• Guaranteed, risk-free or assured-return claims would be banned outright, with a narrow online-bond-platform exception.
• Testimonials would be banned for brokers, as would comparisons that disparage competitors — closing the user-testimonial UGC format that works so well elsewhere.
• Incentives would be banned — vouchers, coupons and perks used to push trading, revive dormant accounts or drive app downloads. That would end refer-and-earn and install rewards, currently standard in Indian broking growth.
• Performance claims could cite only ratings from a recognised Past Risk and Return Verification Agency, with same-frame disclaimers whenever you name a specific security or scheme. Short formats such as SMS and push could use abbreviated disclosures linking to the full version.
• Process would change from pre-clearance to post-publication reporting, with every ad uploaded to a central portal within 24 hours of running. Violations could trigger withdrawal, penalties and a halt on client onboarding — an existential penalty for a growth team rather than a fine.
• The celebrity definition would widen sharply: the top 50 of a national celebrity index, influencers with more than 5 lakh followers on any single handle, national-team sportspersons, TV hosts with a season or 10+ episodes, and explicitly "a virtual character, meaning a fictional avatar with lifelike human traits." AI avatars would count as celebrities, all would need prior approval, and they could promote the entity but not make claims about specific products.
• The code would bind you to the CCPA Dark Patterns Guidelines 2023: no false urgency, no forced action, no subscription traps hiding cancellation.

Genuine educational content would be exempt where branding is minimal, which makes education your compliant top-of-funnel — but "minimal branding" is a real constraint, so treat it as brand building rather than direct response. Check the draft's current status with your compliance counsel before you build a plan around any of it; this post is a snapshot of September 2026.

What that leaves you. Groww's listing copy is the model: free demat account, ₹0 on delivery, ₹20 (about $0.23) per executed order, instrument breadth, social proof by scale rather than testimonial. Cost transparency and onboarding friction are the surviving direct-response angles. "Open a demat account in under ten minutes with DigiLocker" is legal and converts. "Make ₹5,000 a month trading options" ends your campaign and possibly your onboarding.

No published creative-intelligence study for Indian broking apps exists, so format discipline carries over from general mobile UA: short-form video, write to the first three seconds, UGC-style production over polish. Note the tension — UGC style is fine, UGC testimonials would not be. Use founder and staff explainers, screen recordings and product walkthroughs, not customers describing their gains.

How many creative variants can a broking team actually run?

Twelve to twenty net-new assets a month, not the 50–80 this series recommends everywhere else. Broking is the one category where the general variant-volume discipline does not carry, and the reason is the proposed 24-hour portal filing obligation: every ad uploaded to a central portal within a day of running, with a halt on client onboarding among the possible penalties. Fifty to eighty filings a month is a compliance function, not a growth tactic, and no launch-stage team has one.

The operating model that makes 12–20 work:

• A module library, cleared once. Approved hook lines, approved product claims, the disclosure strip, the risk warning, the F&O loss statistic, three end cards. Counsel reviews the modules, not the permutations.
• 12–20 net-new assets a month, each assembled only from cleared modules, so per-asset review is a checklist rather than a legal read.
• One named compliance-ops owner who files the day's live assets in a single batch inside the 24-hour window and keeps the filing log. That is a fraction of a role at Lean and a whole one at Scale — it is the compliance-ops line in the budget table below, and it is why that line exists.
• Anything outside the module library goes to counsel individually and does not launch that week.

If your growth model needs 80 variants a month to work, broking is the wrong category. That is a finding, not a workaround.

Three scenarios, all India, all monthly, using the ₹163 (about $1.85) derived planning CPI and the 14% KYC rate. Recompute against measured numbers at the end of week two — a starting shape, not a forecast. Media is split by operating system rather than only by channel, because Android and iOS are not the same channel with a bid multiplier applied and the deepest viable event differs between them.

Media only in the budget rows; creative production and compliance ops sit on top of it, not inside it. Both are real cash and both are usually missing from launch plans in this category — the second one especially, because it is the cost of the 24-hour filing obligation described above.

On the iOS split. India iOS is roughly 4–6% of devices, and these scenarios put 29–31% of media there. That is a deliberate over-index, not an oversight, and it rests on three things: finance store conversion is 32.8% on iOS against 19.7% on Play, Apple Ads is the only channel with any published India CPI at all, and AppsFlyer's APAC finance data has iOS at a record 16% regional install share with India and Indonesia supplying 58% of APAC iOS finance revenue. A funded broking account on iOS is worth more than one on Android. Re-test the split at week four against measured cost per funded account by OS — if iOS is costing more than roughly 1.6 times Android per funded account, cut it back toward the device share.

On first_trade scoping. At Lean it is out of reach on every channel and both platforms. At Standard, ₹25 lakh, it is borderline: a single Google target-CPA campaign bidding ₹2,772 needs ₹27,720 a day, about ₹8.4 lakh a month, which the ₹10.5 lakh Google Android line covers — but only if you concentrate nearly the whole Android budget into one campaign and accept roughly 30 conversions a day. At Scale it is comfortable on Android. On iOS it is never viable in this table, because a broking first trade lands in AdAttributionKit window two or three, which return coarse values at best and nothing at Tier 0. kyc_complete is the iOS ceiling at every budget here, and more money does not change that.

At Lean you run one Meta ad set, not four. At Scale you cross 50,000 cumulative installs within a couple of months — the gate for App campaigns for Engagement, where remarketing to KYC abandoners becomes a channel rather than a CRM exercise. With 86% of installs stalling before account opening, that pool is your largest efficiency lever in month three.

What is the payback period at incumbent economics?

Roughly eleven to twelve months at incumbent revenue per client, stretching to about fourteen if the guided F&O contraction lands. The category does publish revenue per active client, indirectly: Groww's FY25 revenue was ₹3,901.7 crore against 1.33 crore NSE active clients in August 2026, or roughly ₹2,933 (about $33) of revenue per active client per year. Read that as a ceiling rather than a forecast — it is a blended incumbent figure covering cross-sell, brand-driven organic accounts and a base built over years, none of which a challenger's first cohort has.

That is media cost only. It excludes creative production, compliance ops and the fact that a first trade is not yet an active client under NSE's trailing-12-month definition.

Now restate it against the contraction this post opened with. Inc42 reports F&O historically supplied 70–90% of discount-broker revenue; F&O notional volumes fell 29% in FY25, and Angel One has guided to a further 20–25% loss on that line. Apply a 20–25% decline to a book that is 70–90% F&O and revenue per active client falls to roughly ₹2,340–₹2,470, stretching payback to about 13.5 to 14 months on the same media cost. At the pessimistic end of the input ranges — ₹4,911 per first trade — it does not pay back inside two years at all.

So the honest summary is arithmetic rather than opinion: the unit economics work, thinly, at incumbent revenue per client and the midpoint of two constructed inputs, and they stop working if either input moves against you or the F&O contraction continues at the guided rate. Any plan here should carry a cross-sell line — mutual funds, MTF, insurance distribution — whose job is to lift revenue per client above ₹2,933, because the base case has no margin for error in it.

The KPI ladder, in the order these numbers become trustworthy:

Note the last row. NSE's active-client definition is your north star and it matures slowly, so any CAC number you quote before month four is an estimate wearing a suit.

One India timing note: AppsFlyer's festive analysis (October–December 2024, 20.5 million installs, $419 million in UA spend) found the festive window is a nine-week sequence rather than a Diwali spike, and that iOS fraud rates hit 60% post-Diwali, up 176%. Launch into it and your fraud controls need to be live before you spend.