Case studies
Real problems. Systems that fixed them.
Five engagements across fintech, broking, healthcare, beauty and manufacturing. For each one: what was breaking, what we built, and what changed afterwards. We don’t publish percentages we can’t stand behind.
● Client names are kept private. The problems and the work are real.The dashboard said installs were rising. Real users weren’t.
A consumer investing app (mutual funds and digital gold) for first-time investors in Bharat, Android only, running paid acquisition on Meta, Google and a CPI network.
1,00,000+installs, twice the original 50K goal
14%of “installs” Google was bidding on actually opened the app, caught and fixed
<1%gap between Google Ads and the MMP after the fix
!The problem
- Google Ads was optimising for Play Store downloads. Only about 1 in 7 of those turned into a first app open.
- The MMP recorded Meta correctly, but most Google installs showed up as “Organic”, so nobody knew what Google was really delivering.
- Paid installs registered at about 5%, against about 34% for non-paid users. Onboarding wasn’t the problem. Traffic quality was.
- Generic “start investing” ads did nothing for people who think investing is a scam.
⚙What we built
- Attribution repair: traced the MMP’s Android postbacks and imported the MMP-verified first_open event into Google Ads.
- Changed the bidding goal from downloads to verified first opens, so the algorithm pays for people who actually open the app.
- Segment-led creative in Hinglish for skeptics, salary day, travel goals and “your food order vs a ₹100 SIP”, with 3 variants per campaign, checked against compliance rules.
- Three-channel mix: Meta, Google, and a click network on a flat CPI, plus one weekly report across all channels.
✓What changed
- The app passed 1 lakh installs, well beyond the 50K Play Store badge target.
- Google Ads and the MMP now agree to within a few installs (2,458 vs 2,477), so spend decisions rest on numbers both sides trust.
- The team now reviews channels on registrations and intent, not install counts.
If the ad platform is optimising for the wrong event, it gets very good at buying the wrong users.
StackGoogle AdsMeta AdsApptrove MMPPlay ConsoleWeekly cross-channel report
Four channels, one CRM, and an MIS built by hand every morning.
A SEBI-registered stock broker with a PMS desk, running WhatsApp, email, web push and RCS from one customer data platform, with a content team trying to keep up with search.
!The problem
- Campaigns ran across four channels, but performance lived in separate exports. The daily MIS was stitched together by hand.
- Content production was the bottleneck. Every article was researched, written, formatted and uploaded manually.
- Investors asked the same questions over and over, and support answered each one by hand.
- The PMS desk sells to HNIs, where untargeted ads don’t work, and it had no named-account plan.
⚙What we built
- Automated MIS pipeline in n8n that pulls CRM and ad-platform data into one scheduled report.
- Content pipeline: topic → AI research → AI draft in the brand’s style → WordPress draft for editor review. No copy-pasting.
- WhatsApp AI finance assistant that answers common investor questions, with a human handoff and no investment advice.
- ABM execution plan for the PMS desk: tiered ICP, buying-committee map, trigger-based account scoring and multi-touch sequences.
✓What changed
- The MIS arrives on its own. The team spends that time acting on the numbers.
- Articles go from topic to review-ready WordPress draft automatically, so editors edit instead of starting from a blank page.
- Routine investor questions are answered instantly on WhatsApp. Complex ones reach a human.
- The PMS desk works from a defined account list and playbook instead of broad ads.
Most “marketing problems” in BFSI are really plumbing problems. Fix how data moves and the team gets its week back.
StackWebEngagen8nClaude / PerplexityWordPressWhatsApp APIMeta & Google Ads
A great product with no way to reach the clinics that needed it.
Dental and medical equipment brands growing in India, plus a dermatology device launch, where sales depend on dealers, distributors and doctors’ trust.
!The problem
- Sales were concentrated in a few metro cities. Clinics everywhere else had no local dealer to buy from.
- No system for finding, qualifying or onboarding dealers. Growth depended on whoever happened to call.
- New product launches relied on word of mouth, and doctors had no reason to trust an unfamiliar brand.
⚙What we built
- Dealer acquisition engine: dealer ICP by city tier and specialty, targeted outreach, and a clear onboarding path.
- Launch playbook for new devices: positioning, clinician-facing content, demo campaigns and dealer enablement kits.
- Doctor & hospital digital presence, so the clinicians using the product could be found and recommended.
✓What changed
- Brands went from a few cities to an India-wide dealership network.
- Launches follow a repeatable playbook instead of starting from scratch each time.
- Dealers get a defined offer and onboarding, which makes the network easier to grow and keep.
In healthcare, distribution is the marketing. Make the dealer successful and the brand follows.
StackICP & territory mappingOutbound sequencesCRM pipelineLaunch campaigns
A platform migration that could have erased years of search traffic.
A premium hair and beauty salon in Abu Dhabi moving its website from WordPress to Shopify while competing for high-intent local searches.
!The problem
- Shopify changes every URL. A service page like /services/nail-services/ becomes a completely different path.
- Without a plan, every ranking page, Google Business link and shared link would break overnight.
- Services were bundled together, so no single page could rank for “nails in Abu Dhabi” or “keratin treatment Abu Dhabi”.
- Few local backlinks meant little authority against established salons.
⚙What we built
- Full URL map: every old WordPress URL matched to its new Shopify path, with redirects and updated internal links.
- Service landing pages, each written for local search intent, with booking calls-to-action.
- Local SEO content answering what clients actually search before booking.
- Backlink strategy built on UAE directories, local partners and lifestyle publications.
✓What changed
- The migration shipped with old service URLs mapped to their new pages instead of returning 404s.
- Each service now has its own page that can rank and convert.
- A local authority plan is running instead of waiting for links to appear.
A redesign is the riskiest week in a website’s life for SEO. Plan the redirects before you plan the colours.
StackShopifyWordPress (legacy)Redirect mappingGoogle Business ProfileSearch Console
Buyers search by city and use case. The website had one page.
A manufacturer and installer of invisible grills and balcony safety systems serving homeowners across several Indian cities.
!The problem
- Demand is hyper-local and specific: “invisible grill for balcony in Pune”, “child safety grill Bangalore”.
- One generic product page couldn’t rank for hundreds of city × product searches.
- Writing every page by hand would have taken months and produced thin, near-duplicate content.
- Social posting was irregular, so the brand dropped out of sight between enquiries.
⚙What we built
- SEO audit covering technical fixes, keyword clusters, and city and use-case demand.
- Programmatic SEO engine: a data sheet of cities, products and use cases feeds an n8n workflow that writes distinct pages and publishes them through the WordPress REST API.
- 30-day Instagram & Facebook calendar, then ongoing content and social management.
✓What changed
- New city and product pages are generated from a sheet instead of written one at a time.
- Each page targets a real search with local details, not a copy of the same text.
- The brand posts on a consistent schedule, so it stays visible between enquiries.
When demand is local, the website has to be local too, and at scale that takes a system, not a copywriter.
Stackn8nWordPress REST APIGoogle SheetsAI contentMeta Business Suite
The common thread
Every engagement follows the same four steps.
- Find the real leakUsually it isn’t the ad or the copy. It’s a tracking event, a broken handoff or a manual step.
- Fix the data firstAttribution, CRM fields and URLs have to be right before any campaign can be judged.
- Automate the repeatableReports, content, follow-ups and pages get built by workflows, not by hand.
- Hand over a systemDocumented, owned by your team, and running in your stack, not ours.
Your turn
Recognise one of these problems?
Tell us what’s breaking in your marketing ops. We’ll reply with where we’d look first. It’s free, and there’s no obligation.
- Read by a person, not a bot
- A specific first fix, not a generic proposal
- Your data stays yours