HiveMinds was spending two hours a week just finding out what happened last week
HiveMinds handles paid media for Jio Financial Services across personal loans, credit cards, insurance, and more, on Meta and Google. The account team is small: three or four people across analysts, account owners, and delivery unit heads. The budget isn't. It runs every day across two platforms, and the account had grown past the point where anyone could keep it all in their head.
Before Segwise, working out which creatives were doing well meant pulling data out of Meta and Google separately, deciding which metrics mattered for that week's review, and comparing ads by hand. That took two hours, at minimum, every week. It caused three problems, and if you run a busy account you'll probably recognize all of them.
The review was the job. Two hours a week went into figuring out what had already happened, not deciding what to do next.
Refreshes lagged the data. A set of creatives could show falling click-through and rising acquisition costs, but by the time the team had pulled it together and turned it into feedback, the next refresh was already in production. The weak ads kept spending in the meantime.
The deeper work got skipped. Competitor reviews, post-campaign teardowns, checking whether reward-led messaging beat product-led messaging: all of it waited for spare time. On a lending account, where messaging and compliance both carry real weight, that's the analysis you can least afford to drop. It was the first thing to go.
The team now starts with one ranked view and compares ads by the message they carry
Segwise pulls Meta and Google into one dashboard and auto-tags every creative, so the team opens each review with a ranked list of what's working instead of building one from scratch. They don't stop at click-through rate. They follow each creative down to cost per acquisition (what it costs to win one new customer) and conversions, because an ad that gets attention but doesn't convert isn't a winner.
The tags also let them compare ads by communication type: offer-led, product-led, and brand-led. That matters because the strongest type changes with the product (a loan and a credit card solve different needs), the platform (people arrive on Meta and Google with different intent), and the stage of the funnel (ads aimed at people close to signing up need to be direct, while ads for people just discovering the brand can take longer to make their case). The question in the room is now which message is working and what to make more of, not which single ad looks better.
For questions that used to need a custom report, they ask AI Chat. One example: they asked which communication type was performing best for personal loans and what the strongest ads had in common. Segwise compared performance by message type and pointed out the elements that kept showing up among the winners.
On competitors, Segwise ranks rival ads from the last 90 days by impressions, so the team looks at what competitors are actually putting money behind, not everything they've ever posted. That showed them where the category was crowded (loan amounts, monthly installment messaging, and quick-approval hooks are everywhere) and pushed them toward more specific, benefit-led angles instead of one more ad that looks like everyone else's.
Because Jio Financial Services is a regulated lender, every generated asset has to follow brand and compliance rules before anyone reviews it: correct logo use, approved colors, the right tone, and mandatory disclaimers where they apply. A high-performing ad simply can't run if it doesn't clear compliance. Segwise generates creative from the brand's own winning patterns while staying inside those rules, which cuts down on rework after the fact.
Since going live, the team has generated 400 to 500 static creatives and 20 to 25 videos in Segwise, and 50 to 60 of them went live as generated. The team built another 150 to 200 creatives themselves from what Segwise showed them. Together, that's 40 to 50% of everything shipped for the account.
A cleaner credit card ad brought acquisition costs down 25% on Meta and 28% on Google
The clearest example is a credit card campaign on Meta and Google. Instead of judging ads on click-through alone, the team traced performance through the funnel to find which message themes drove both engagement and sign-ups.
That led to a simpler, benefit-first ad: a reward on every spend ("Har spend pe reward"), no annual fee ("Har saal ₹0 fee"), the lifetime-free card itself with JioPoints on activation and spend, and a plain "Apply Now" button. No clutter.
The refreshed creative cut cost per acquisition 25% on Meta and 28% on Google. Click-through rate (the share of people who saw the ad and tapped it) landed at 2.5% on Meta and 3 to 4% on Google. Click-to-install, the share of those clicks that went on to install the app, came in at 20 to 25% on Meta and 15 to 20% on Google. Both were ahead of where the campaign had been running before the refresh.
“We stopped trying to fit three benefits into one ad. Once we picked the one that actually mattered and said it clearly, the numbers moved.”
The bigger change is that the deep analysis happens every week now
The time saved and the lower acquisition cost are the numbers people notice. The quieter change sits under both of them: the analysis that used to be optional is now routine. Competitor tracking and communication-type comparisons aren't side projects anymore. They're part of how every brief and refresh gets built.
Next, HiveMinds wants to score creative before it ships
HiveMinds' next step is moving Segwise earlier in the process. Instead of learning after launch whether an ad worked, the plan is to score each new creative against what has already worked for the account before it goes live. The team also wants more video generation and a shorter path from a finding to a live campaign. The main goal, though, is catching weak creative before it spends anything.
“We used to spend our creative reviews working out what happened. Now, thanks to Segwise's Creative Strategy Agent, we spend them deciding what to make next. Close to half of what we ship for the account is either generated in Segwise or built from what it shows us.”