How a Global EdTech Fixed a Sub-1 ROAS by Rebuilding the Funnel

Before EngagementSub-1 ROAS (0.5x)

After EngagementUp to 3.0x ROAS

Operator Review

Co-founder leading organic search engineering, local presence strategy, and client growth architectures.

The Short Version

A global EdTech platform was selling executive business education and tech upskilling courses across five wildly different regions on essentially one playbook — and the economics were underwater. In some geographies, return on ad spend had fallen below 1, meaning every marketing dollar was coming back as less than a dollar. The instinct was to fix the ad accounts. The real problem was a funnel that leaked from the first click to the moment of purchase, run identically across markets that needed completely different approaches. By rebuilding strategy, creative, and the funnel geo by geo and course by course, the 8-month systems rebuild delivered a 60–70% reduction in CPL, a 50–60% reduction in cost per demo booking, and ROAS improvements ranging from 80% to 300% depending on market and course — turning a loss-making engine profitable.

The Hero: A platform with a great product and broken economics

Our client was a multi-geo EdTech platform selling elite executive business education and senior software engineering masterclasses across the US, India, UK, MEA, and SEA — five markets with almost nothing in common when it comes to how people buy education. Price sensitivity, payment behaviour, trust signals, decision speed, even what “good” looks like in a creative all change dramatically between Mumbai, London, Dubai, and Austin. (Brand anonymized for confidentiality.)

And like the best companies that come to us, they were already excellent at the part that's hardest to teach: once a prospective student actually joined a demo and completed it, the product sold itself. Demo completion rates were strong. The teaching was good, the value was real, and people who experienced it were convinced. The problem had nothing to do with the product. It had everything to do with the system that was supposed to get qualified students into those demos — and convert them afterwards.

That's what made the numbers so painful. They had a product that worked, and an acquisition engine that was quietly losing money in market after market.

The Situation: The engine was burning more than it earned

On the surface, campaigns were running across every geo. Budgets were deployed, leads were flowing, demos were happening. But almost every metric that determines whether a paid engine is viable was in the red:

• Cost per lead (CPL) was high across the board — they were overpaying for volume.
• Cost per demo booking (CPDB) was high — even the leads they got were expensive to move one step down the funnel.
• Lead quality was poor — wrong intent, wrong geo, wrong fit.
• Call connect rates were poor — the team couldn't reliably reach the leads they'd paid for.
• Demo joining rates were poor — a booked demo too often meant a no-show.
• Demo completion was strong — the one bright spot, and proof the product and pitch worked.
• Demo-completion-to-conversion was poor — even students who saw the full value weren't buying at the rate they should.
• And so lead-to-conversion was poor overall — every weak stage compounded into a leaky whole.
• The bottom line: ROAS was below 1 in several markets — as low as 0.5 in some geos. Fifty cents back on the dollar.

When ROAS sits below 1, you don't have a marketing inefficiency. You have a business-model emergency — you are, mechanically, paying to lose money, and scaling makes it worse.

The Complication: One global playbook running five different markets — through a leaking funnel

Here's where the obvious diagnosis fails. Looking at a sub-1 ROAS, most teams reach for the ad accounts: pause the losers, tweak bids, refresh creative. That treats the symptom and misses the disease. When we went deep, the real challenge had two interlocking layers — neither visible in a campaign dashboard.

Layer one: an undifferentiated strategy across radically different geos and courses. Budget was being distributed across channels and campaigns without a real, market-level or course-level logic. A student weighing a course in the US has a different price expectation, payment method, objection set, and trust trigger than one in MEA, SEA, India, or the UK. Each course had its own economics too — different price point, different margin, different buyer. Yet they were largely being marketed with the same channel mix, the same budget assumptions, the same creative, and the same messaging everywhere. Strong channels in one geo were starved while weak channels in another were overfunded, because there was no per-geo, per-course plan telling the money where to go. This is why ROAS could be acceptable in one market and 0.5 in another — the same playbook that roughly fit one market actively destroyed value in another.

Layer two: a funnel that leaked at every stage between the click and the purchase.

• The landing pages were generic — one template trying to sell multiple courses across five regions, so high-intent visitors didn't see themselves and dropped.
• The forms were friction-heavy, shedding qualified prospects at the very moment of highest intent.
• There was no lead qualification framework — every lead was treated identically, so the team chased unfit leads as hard as genuine ones, which is exactly why connect rates and demo-joining were so low.
• Creative wasn't tested or matched to market — it was produced and shipped, with no framework to learn what actually worked, where.
• The narrative, communication, touchpoints, and follow-up weren't engineered to carry a student from interest to a joined demo to a confident purchase — so even strong demo completions leaked at the conversion step.

The Implication: What a sub-1 ROAS actually does to a company

It's worth sitting with how destructive this is, because “ROAS below 1” understates it. Every unqualified lead doesn't just waste ad spend — it consumes sales-team time to chase, inflates cost per demo booking, and clogs the pipeline that the team's strong closers need kept clean. A poor connect rate means the team works flat out and reaches no one. Poor demo joining means the platform's single most powerful asset — a demo that converts when completed — is starved of attendees. And a weak demo-to-conversion step means that even when a student sees the full value, the system fails to close them.

Stack those together across five geos and multiple courses and you get the worst possible position: a company paying, in some markets, two dollars to earn one — and unable to scale its way out, because scaling a sub-1 ROAS just loses money faster. Worse, the temptation under pressure is to spend more to hit revenue targets, which deepens the hole. They were at the exact inflection point where the next decision either fixes the economics or makes them structurally unrecoverable.

The Need-Payoff: What they actually needed

Reframed correctly, the need was not “cheaper leads” or “better ads.” It was: a market-specific and course-specific acquisition strategy, feeding a funnel engineered to carry a qualified student from click to a joined demo to a confident purchase — with a creative system that actually learns what works in each geo. Fix that, and CPL, CPDB, demo joining, conversion, and ROAS all move together, because they're all downstream of the same root cause.

The Resolution: Rebuild per geo and per course — then engineer the funnel and the creative

We didn't optimize campaigns and hope. We rebuilt the system in three coordinated moves:

1. A separate strategy and budget plan for every geo and every course. We did thorough research on each course's offering, pricing, funnel, and touchpoints, and on each geo's buyer — then built dedicated plans, not one global template:

• Separate budget planning per geo and course, sized to that market's real intent and economics rather than split by habit.
• Separate channel allocation per geo and course — funding the channels that genuinely produced qualified, convertible students in that specific market, and cutting the ones that didn't.

This directly attacked the “poor budget distribution” and geo-variable ROAS at the root: money started flowing to where it actually returned, market by market and course by course.

2. A rebuilt funnel from first click to purchase.

• Integrated a lead qualification framework so the team could instantly separate genuine prospects from noise — lifting lead quality and connect rates by letting the team spend energy where it counted.
• Redesigned the landing pages to be geo- and course-specific and intent-matched, so the right student saw the right proof and converted into a lead instead of bouncing.
• Redesigned the forms to strip friction at the highest-intent moment, recovering qualified prospects that were being lost at submission.
• Rebuilt the narrative, communication, touchpoints, and follow-up framework to escort students from inquiry → booked demo → joined demo → confident purchase, repairing both the demo-joining and the demo-to-conversion leaks.

3. A complete creative system — studied, tested, and measured. Rather than producing creatives and hoping, we built a disciplined creative engine. We studied Meta's Andromeda ad system deeply, ran thorough competitor research, and rebuilt our understanding of the product, its USP, and its key highlights from the ground up. From that, we crafted both human and AI-assisted creatives — and, critically, ran them through a complete creative-testing framework with defined success metrics for every test, so the system learned what worked in each geo and each course rather than guessing. Creative stopped being an output and became a feedback loop.

The Results: Reported growth and efficiency gains

With per-geo, per-course strategy live, the funnel rebuilt, and the creative engine learning, the metrics moved together — and the magnitude varied by market and course, exactly as you'd expect when the root fix is market-specific:

*Results reported over the 8-month systems rebuild. Client identity and projects fully anonymized for brand confidentiality.

The shape of these results is the proof. CPL and CPDB fell sharply while lead quality rose — costs and quality usually move in opposite directions, and seeing them improve together is the signature of fixing allocation, qualification, and creative rather than just buying cheaper clicks. Demo joining jumped ~70%, finally feeding the platform's strongest asset — a demo that converts when completed — the volume it had been starved of. And because the demo-to-conversion step was rebuilt too, that extra volume flowed through to a higher conversion rate, a larger average ticket, and — the number that actually matters — a ROAS lifted by 80% to 300%, turning loss-making markets profitable.

Why it worked: The funnel and the market fit were the problem, not the ad spend

The reflex when ROAS drops below 1 is to blame the platforms or to spend more. The breakthrough came from refusing both and asking the harder question: where is qualified, convertible intent leaking — and is the same playbook quietly destroying value in markets it was never built for? The answer wasn't in the ad accounts. It was in a global strategy applied to five local realities, and a funnel that couldn't carry a student from click to purchase. Fix those — geo by geo, course by course, stage by stage — and the “bad ROAS” problem dissolved, because it was never really an ads problem.

That's the core of how Fluxsy works: growth isn't won by spending more, but by rebuilding the system that turns spend into qualified, convertible demand — tailored to each market, measured at every stage, and tested until it compounds.

Byline & operator credentials

Written by Tushar Verma, Co-Founder & Sales/Operations Lead at Fluxsy. This engagement is completely anonymized for brand confidentiality.

Compounding Growth Insights

Growth & Retrieval Intelligence

We document strict performance blueprints, privacy-centric analytics, and pricing structures directly linked to conversions.Highly recommended read for this page: Signal & Inertia