Ten times I was handed a funnel and asked to make it pay.

Situation

PokerBaazi was buying users at volume, but the cohort curves flattened far too early. D15 retention sat at 34%, and a large share of first-time depositors never came back for a second session in the same week. With a first-time-deposit CAC around ₹9,000, every cohort that decayed early wiped out the margin the acquisition team had just earned. Retention was being treated as a CRM calendar rather than a revenue line.

Task

Lift early-life retention and turn repeat play into a predictable share of monthly net gaming revenue, without adding media spend. The commercial targets were a 60-day payback window and positive ROI by D180.

Result

• 34 → 42%D15 retention after the onboarding rebuild
• 25–30%of monthly net gaming revenue attributable to CRM and reactivation
• ₹9K → ₹6Kcost per acquired paying customer, helped by better downstream retention signal
• 3.4XROI by D180, with payback inside 60 days

Action

• Found the real drop-off before touching messagingRebuilt the cohort view in Branch and Tableau across install → registration → KYC → first deposit → first real-money table → D3 / D7 / D15. The largest single leak was not deposit; it was registered users who never reached a table. That reframed the problem from a CRM problem into an onboarding problem.
• Rebuilt onboarding with Product and DesignCompressed the KYC steps, added a guided first-game flow, built a practice-to-real-money bridge so new users learned the interface before risking a deposit, and attached a starter reward code to the first session rather than the install.
• Moved CRM from broadcast to cohortSegmented the base in CleverTap on deposit recency, frequency and value alongside skill tier and game-format preference. Built day 0 / 1 / 3 / 7 activation journeys, deposit-trigger flows, and tournament reminders keyed to the formats a player had actually played, across push, SMS, email and WhatsApp.
• Built a reactivation ladderDefined lapsed segments at 30, 60 and 90 days with different offers and tone at each. Win-back reward codes, personalised tournament invites, and frequency-capping plus suppression rules so the reactivation push did not fatigue the active base.
• Ran retention as an experiment programme80–120 experiments a month across offer construct, send timing, channel mix, creative and copy, holding a 25–30% win rate. Losers were retired weekly rather than left running.
• Fed retention back into acquisitionPushed D60 retention by source back into the acquisition side, so affiliate payouts, DSP optimisation and creative selection were judged on retained users rather than installs. Retention stopped being a downstream department and became an acquisition input.

Situation

Monthly media was at ₹60L and the business wanted a step change in scale ahead of IPL. Two constraints made that hard. Real-money gaming sits under advertising restrictions on the major platforms, and Apple's ATT framework had gutted the deterministic iOS signal that install campaigns had been optimising against. Scaling naively would have meant buying installs that never deposited.

Task

Scale monthly spend past ₹4.5Cr within eight months and absorb a ₹10Cr peak month, while reducing cost per acquired paying customer and protecting the quality of incoming depositors.

Result

• 7.5Xspend growth, ₹60L to ₹4.5Cr a month in eight months
• ₹10Crpeak month absorbed across IPL on Hotstar and Shark Tank sponsorships
• −33%cost per acquired paying customer, ₹9K to ₹6K
• 3.4XROI by D180 at 60-day payback

Action

• Changed what the algorithms were bidding towardMoved every channel off install optimisation and onto first-time deposit and deposit value. Meta ran event and value-based optimisation, Google App campaigns were split by action type across install, in-app action and target ROAS, and Apple Search Ads covered brand defence plus category and competitor terms where intent was already formed.
• Rebuilt the signal layer with EngineeringDesigned the SKAdNetwork conversion-value schema so iOS postbacks carried deposit tiers rather than a binary install. Wired Meta Conversions API and Branch server-side postbacks, and pushed offline deposit value back into the platforms so bidding saw revenue, not events.
• Built a creative engine, not a creative queueRan influencer and pro-player UGC, in-app feature promotion, reward-code hooks and reactivation angles as parallel concept lines. 80–120 tests a month with a scheduled refresh cadence, because in this category creative fatigue, not audience size, is what caps spend.
• Planned the peak months as operations, not campaignsFor IPL on Hotstar and the Shark Tank sponsorship, sequenced brand and performance so the sponsorship spike landed on a prepared funnel: pre-built store listings, retargeting pools armed before the airing window, and server capacity coordinated with Engineering for the traffic surge.
• Worked the store, not just the adsRan ASO on listing copy, screenshots and creative sets, and secured editorial placements with the Apple and Google partner teams, which lifted organic conversion on the same paid traffic.
• Held the line with daily guardrailsRan a team of four against daily CAC and payback thresholds with pre-agreed reallocation rules, so budget moved between channels on evidence within the day instead of waiting for a weekly review.

Situation

Growth was concentrated on Meta and Google. As spend scaled, auction pressure pushed CAC up and the marginal user got worse. In real-money gaming that concentration is also a business-continuity risk, since a single policy review can pause the majority of a company's acquisition in a day.

Task

Build acquisition channels that added genuinely incremental paying customers at comparable or better payback, and reduce the share of volume dependent on two platforms.

Result

• 20%of total first-time deposits from four channels that did not exist before
• 4 channelsprogrammatic, CPI affiliates, brand partnerships, community and offline
• 12third-party community managers running Telegram and Discord
• Lowermarginal CAC on the core channels once volume pressure eased

Action

• Programmatic seeded with first-party eventsBrought Moloco and Criteo on with deposit events rather than installs as the optimisation target, and ran them against D60 retention and ROAS goals. Validated incrementality with geo holdouts before scaling budget, so the channel had to prove it was adding users rather than re-buying ones Meta would have found anyway.
• Rewrote the affiliate contract around retentionStructured AdCounty and Intellact on a 25% D60 retention KPI instead of a flat CPI. That single change aligned the network's incentive with the business. Backed it with Branch anti-fraud, click-injection filters, publisher-level blocklists and tiered payouts, and reviewed publisher cohorts weekly rather than paying on installs and auditing later.
• Went where the audience already had a wallet openBuilt brand partnerships with CRED, Swiggy, Paytm, Zomato and Blinkit using reward drops and coupon inventory. Each partnership got dedicated deep links and its own attribution path so the contribution was measurable rather than assumed, and offers were structured to reach people at a moment of transactional intent.
• Treated community as a channel with ownersStood up Telegram and Discord communities run by 12 third-party community managers, with per-community referral codes, leaderboards and event calendars. Extended the same motion offline into corporates, coworking spaces and hostels where the poker-playing audience clusters.
• Built the measurement to tell incremental from cannibalisedChannel-level cohort dashboards, MMP deduplication rules and periodic incrementality tests, so a new channel could be defended on contribution rather than on last-click credit.

Situation

Baazi Games had a large, engaged player base and a recognisable brand, but nothing to sell them. BaaziStore existed as an idea: poker kits, apparel and accessories, with no storefront, no product feed, no fulfilment process and no measurement. Everything from catalogue structure to shipping thresholds had to be decided before a single rupee of media went out.

Task

Launch the Shopify D2C channel and take it to meaningful monthly GMV within six months at a blended ROAS that made the channel worth funding rather than subsidising.

Result

• ₹55–60Lmonthly GMV within six months from a standing start
• ~4Xblended ROAS, with 8.4X on apparel and 4.5–5X on poker kits
• ₹15Lmonthly media supported profitably at launch scale
• +47%average order value, ₹1,500 to ₹2,200
• 1.4 → 2.1%storefront conversion rate

Action

• Built the commercial spine firstSet up the Shopify storefront, then designed catalogue architecture across a 20+ product range and 50+ SKUs, with pricing and margin tiers defined upfront. Knowing contribution margin per SKU on day one is what later made margin-aware bidding possible.
• Treated the feed as the campaignOwned Google Merchant Center end to end: feed optimisation, title and attribute structuring for how people actually search, custom labels encoding margin tier and seasonality, automated feed rules, and a standing process for disapproval resolution so inventory never silently dropped out of the auction.
• Structured campaigns around margin, not just categorySegmented Shopping and Performance Max by product category and margin tier, using custom-label segmentation and bidding strategy to push spend toward high-contribution SKUs instead of high-revenue ones. Meta ran catalogue prospecting and dynamic retargeting alongside.
• Used the player base as a cold-start advantageSeeded prospecting audiences and lookalikes from PokerBaazi first-party data, which meant the store launched into a warm audience rather than paying full price to discover one.
• Ran CRO on the pages that decide purchasesReworked PDP layout, imagery, copy, sizing guidance, trust signals and shipping clarity, and simplified the checkout flow. Storefront conversion moved from 1.4% to 2.1%, which is the same as a 50% media efficiency gain at no extra spend.
• Bought the order value upIntroduced bundling, accessory cross-sells, upsell journeys and a free-shipping threshold set just above the natural basket, lifting AOV from ₹1,500 to ₹2,200 and giving the campaigns more headroom on CAC.
• Closed the loop with operationsCoordinated fulfilment SLAs, returns policy and inventory status back into bidding, so campaigns were not scaling SKUs that could not ship.

Situation

BaaziStore was scaling on paid acquisition, and almost every order was a first order. Paid CAC was climbing with volume, while the categories being sold — poker kits, apparel, accessories — have obvious replenishment, upgrade and gifting cycles that nothing in the business was set up to catch. Owned revenue was effectively zero.

Task

Turn repeat purchase into a system rather than a coincidence: raise lifetime value per customer, build an owned revenue line, and reduce the store's dependence on paid media for growth.

Result

• 16–20%annual repeat purchase rate, sustained rather than spiked
• +47%average order value, ₹1,500 to ₹2,200
• Ownedrevenue line established through lifecycle flows, easing paid CAC pressure
• Betteracquisition targeting, once first-product LTV data fed back into bidding

Action

• Segmented the base on behaviour, not demographicsBuilt RFM segmentation in Klaviyo and defined working segments: new, active repeat, at-risk, lapsed and VIP. Every flow and campaign was addressed to a segment, which is what stopped the programme becoming a weekly email to everybody.
• Built the flows that run without anyone touching themWelcome, browse abandonment, multi-step cart recovery across email plus SMS and WhatsApp, post-purchase care and education, replenishment and cross-sell timed to each category's natural cycle, winback, reactivation, and VIP early access. Automated flows carried the base load so campaigns could be reserved for genuine moments.
• Built the calendar around when the audience actually buysTied campaign timing to poker season, major tournaments, IPL and the festive and gifting window, rather than a generic retail calendar. Gifting in particular turned out to be a meaningful second-purchase trigger for kits.
• Grew the owned list deliberatelyOn-site capture, opt-in at order confirmation, and recruitment from the Telegram and Discord communities, so the lifecycle programme had a growing base to work with rather than only the customers paid media happened to deliver.
• Measured second-order behaviour, not opensTracked repeat rate, time to second order and customer value by first-product cohort. That analysis showed which entry products produced the most valuable long-term customers, and the finding went straight back into Shopping custom labels and bid strategy on the acquisition side.

Situation

GemPundit sold certified gemstones across India and the US at roughly ₹2Cr in monthly GMV on ₹20L of media. It is a difficult commerce category: high average order value, a long deliberation window, heavy trust sensitivity, and buyers arriving from two completely different mindsets — some shopping for a specific certified stone, others led there by astrological guidance with no idea what they need. Most purchases involved a human conversation somewhere in the middle.

Task

Improve return on ad spend materially while continuing to grow volume, across two markets with different currencies, shipping expectations and cultural framing of the product.

Result

• 8X → 11XROAS on roughly ₹20L monthly media
• ₹2Crmonthly GMV maintained through the restructure
• Highercontribution per order after margin-weighted bidding
• Two marketsIndia and US run on separate structures and messaging

Action

• Split the demand by mindset, not by keyword volumeSeparated certified-stone buyers who search by gem, carat and origin from astrology-led visitors who search by problem or planetary remedy. Each got its own campaign structure, landing experience and creative register. Search-term mining and aggressive negative sculpting kept the two from bleeding into each other, and branded was split out so it stopped flattering the blended number.
• Rebuilt the account around price band and marginRestructured Shopping and Performance Max by gemstone category and price band, with margin-aware custom labels so bidding chased contribution rather than headline revenue. India and US ran as separate structures with their own currency, shipping promise, payment expectations and messaging.
• Made trust the conversion assetBuilt landing experiences around lab certification, origin documentation and expert consultation rather than discounting. In a category where the fear is authenticity, proof outperforms price, and the pages were designed to answer that fear before asking for the order.
• Designed for assisted conversion explicitlyMapped the real journey as enquiry → consultation → recommendation → purchase, and built the funnel to book a consultation as the primary action. Remarketing was sequenced by where someone sat in that journey, and WhatsApp plus email follow-up recovered consultations that had happened but not converted.
• Fed order value back into biddingImplemented value-based bidding with margin-adjusted conversion values and pushed confirmed order value back through GA4 and offline conversions, so the platforms optimised toward the orders that actually made money.
• Tested proof, not just creativeRan testimonial and UGC formats, certification-led creative, and PDP redesigns with price-band anchoring, evaluating winners on revenue per session rather than click-through.

Situation

Tetr sells a high-ticket, sales-assisted education product with an average contract value around ₹1.5Cr, a nine-stage buying journey and a multi-month consideration cycle, across six global markets and a portfolio spanning UG, PG and four executive programmes. When I joined there was no channel strategy, no attribution, no operating cadence and no team. Monthly media was ₹2L.

Task

Build the growth and demand generation function 0 to 1, deploy ₹23Cr of media for the 2026 cycle against a 70-close target, and keep customer acquisition cost defensible against the contract value.

Result

• 83 closesagainst a target of 70, 19% above plan
• ₹27.7LCAC against roughly ₹1.5Cr average contract value
• −38%blended CPL, ₹2,700 to ₹1,680, at matched ₹3Cr monthly spend with quality flat
• 1,500Xspend scale in four months, ₹2L to ₹3Cr a month
• ₹50Cracquisition budget now planned for the 2027 cycle against a 200-close target

Action

• Built measurement before buying mediaWorked with Engineering and Data to stand up GA4, a disciplined UTM taxonomy, Meta Conversions API, Google Enhanced Conversions and offline conversion pushback from LeadSquared. With a multi-month cycle, this is the only way closed revenue traces back to campaign, channel and market. Without it, scaling to ₹3Cr a month would have been spending blind for a quarter before learning anything.
• Designed channel architecture per market and per programmeBuilt campaign structure across six markets and six programmes, with Google and Meta carrying roughly 75% of spend and LinkedIn, Snapchat, Tyroo and programmatic covering the remainder. Each market got its own creative, proof points and budget logic rather than a translated version of the India plan.
• Optimised on downstream stages, not on leadsThe CPL reduction came from refusing to buy cheap leads. Budget was shifted using application-stage and consultation-stage signal rather than raw lead volume, which is why CPL fell 38% while lead-to-application quality held flat at matched spend.
• Ran the nine-stage funnel jointly with SalesOwned lead → qualification → application → consultation → interview → offer → enrolment as a single accountable funnel, with weekly review of stage conversion rates and pipeline velocity alongside Sales. Marketing was judged on closes, not on MQLs.
• Built the nurture layer for a long cycleConfigured LeadSquared for lead scoring, routing with response SLAs, and nurture journeys across email, WhatsApp and webinars, plus sales enablement material for the consultation stage. In a multi-month decision, the nurture programme is what keeps a lead alive long enough to convert.
• Hired and ran a team of sixBuilt the function across acquisition strategy, media buying, experimentation and reporting, and set the testing cadence across geography, cohort, audience, message, creative and landing page.

Situation

Techmagnate's client work was growing, but its own acquisition depended on referrals and outbound. Lead flow was inconsistent month to month, quality varied wildly, and the sales team had no predictable inbound to work. The agency was, in effect, the worst-served client on its own roster.

Task

Build a repeatable in-house acquisition engine that produced qualified pipeline with predictable volume and cost, and feed the sales team leads it actually wanted to call.

Result

• 500+marketing services leads a month
• ₹1.2Crin qualified sales pipeline per month
• −26%cost per lead
• +35%lead-to-opportunity conversion

Action

• Gave each service line its own front doorBuilt dedicated landing pages per service — SEO, paid media, ecommerce marketing, ASO — each carrying its own proof, case studies and commercial anchoring, instead of routing all demand into a generic contact page that converted badly and told sales nothing.
• Bought intent, distributed contentGoogle Ads covered high-intent service, competitor and geo terms. Meta handled retargeting and content distribution rather than cold prospecting, and an SEO and content programme carried the mid-funnel research terms where buyers compare approaches before they compare agencies.
• Made the form do qualification workReplaced the single contact form with multi-step forms capturing service need, budget band, industry and timeline, added calendar booking for direct consultation, and set up call tracking so phone enquiries were attributed rather than invisible.
• Closed the loop between sales and spendBuilt lead scoring and routing with a first-response SLA, and, importantly, made sales send disqualification reasons back. Those reasons became keyword negatives, audience exclusions and creative changes. That feedback loop is where most of the 26% CPL reduction and the 35% lift in lead-to-opportunity came from.
• Reallocated toward what converted downstreamAnalysed lead-to-opportunity progression by service line and geography and moved budget toward the combinations that produced opportunities, not the ones that produced the most leads. Reporting ran in GA4 and Looker Studio with Search Console and Ahrefs covering the organic side.

Situation

A Series B HR Tech platform needed demand generation across India and MENA on ₹15L a month. The buying committee spanned HR, IT and Finance, each with different objections. Marketing and sales did not share a definition of a qualified lead, which meant volume could rise while sales-accepted quality stayed flat, and nobody could tell which was happening.

Task

Build a demand generation and pipeline engine producing consistent qualified volume across both markets, and align marketing output to the definitions sales would actually accept.

Result

• 190MQLs per month on ₹15L monthly spend
• 62SQLs per month, roughly a third of MQLs converting
• Two marketsIndia and MENA running on localised structures
• SharedMQL definition agreed with sales and enforced in the CRM

Action

• Mapped the committee before the channelsDefined ICP by employee size, industry and HRMS maturity, then mapped the buying committee and what each member needed to hear. HR cares about adoption, IT about integration and security, Finance about payback. That mapping drove creative, asset and landing-page decisions rather than sitting in a deck.
• Matched channel to buying stagePaid search covered category, competitor and problem-aware terms for demand capture. LinkedIn ran job-title and company-level targeting for demand creation and content syndication. ABM used account-targeted ads with personalised landing pages for priority accounts, and SEO and content carried problem-aware research traffic.
• Agreed the MQL definition with sales, then enforced itIntegrated Salesforce and HubSpot with lifecycle stages, a fit-plus-behaviour lead scoring model, routing rules with response SLAs, nurture automation and a handoff that passed context rather than just a contact record. A shared definition is what made the 190-to-62 conversion a number both teams trusted.
• Separated the content track from the demo trackGated assets, compliance and HRMS buyer guides, and webinars ran as one track with their own nurture. Demo requests ran as a separate, faster track. Mixing them is the standard way a B2B funnel starts reporting healthy MQL volume that sales quietly ignores.
• Localised MENA properlyBuilt regional landing pages, bilingual creative, messaging around regional labour and payroll compliance, and scheduling aligned to the local work week. MENA was run as its own market with its own economics, not as an India campaign with a wider geo setting.
• Optimised on MQL to SQL to opportunityAnalysed progression through those stages by campaign, keyword and audience, and shifted investment toward the higher-intent sources, which is what kept quality intact while volume grew on a fixed budget.

Situation

Stratbeans sold an enterprise learning management system into L&D, HR and IT buyers across India, the US and Australia. There was no inbound engine, a modest budget, and sales cycles of four to five months driven by committee decisions, security review and procurement. Demand arrived unpredictably and was worked inconsistently.

Task

Build the full-funnel B2B growth engine from scratch and take responsibility for converting the demand as well as creating it, across three markets with very different buying norms.

Result

• 50–70MQLs per month at ₹1,500–2,000 cost per MQL
• 10–12%MQL to SQL conversion
• ₹3.5–4.5Crin annual sales pipeline
• +15–20%lead-to-close conversion
• 4–5 → 3–4months enterprise sales cycle

Action

• Built organic around how LMS buyers actually searchTargeted comparison and evaluation terms, compliance training, SCORM and xAPI questions, and onboarding-training topics, with dedicated comparison and alternative pages for bottom-funnel queries. On a small budget, organic was the only channel that could carry volume sustainably.
• Used webinars as the qualification mechanismRan a webinar programme built on customer stories and practical L&D problems. Attendance and question behaviour told us more about buying intent than any form field, and the recordings became gated assets and nurture content afterwards.
• Kept paid narrow on purposeGoogle Ads ran only on category and competitor terms with tight negatives. With a limited budget against enterprise competitors, precision mattered more than reach.
• Wrote a qualification framework and used itQualified on budget, learner count, incumbent LMS and implementation timeline, with defined follow-up cadences. That framework is what pushed MQL-to-SQL to 10–12% and kept the pipeline number honest.
• Built the collateral that unblocks enterprise dealsROI calculators, security and compliance one-pagers, integration documentation and implementation timelines, given to prospects early rather than on request. Moving the technical and security validation forward in the process is the main reason the cycle compressed from four or five months to three or four.
• Fed objections back into contentRan pipeline reviews, kept an objection library from live deals, and turned recurring objections into pages, assets and nurture content, so the next buyer met the answer before they raised the question.