From 15 subscribers to 240 — with a 98.67% lead quality rate.
A high-protein Indian fusion meal delivery startup in the Toronto GTA had a great product and no engine to sell it. We built the engine — Meta Ads, ManyChat, Zoho CRM, Conversion API — and ran it for ten months. By month six, the founder asked us to pause the campaign because the kitchen couldn't keep up.
A good product, and no system to sell it.
Three Indian co-founders running a high-protein meal delivery service across the Greater Toronto Area. Subscription model. Multiple plans — starter, weekly, monthly, muscle gain, fat loss, veg, non-veg, breakfast-lunch-dinner combinations.
They had 15–20 unreliable subscriptions at best. They'd spent 1–3k CAD on event sponsorships targeting the Indian community in Toronto and converted exactly zero. Not one subscriber.
Their Instagram had under 300 followers. No human face on the brand — just photos of tiffins. No one watching could tell what the service was, how it worked, what plans existed, or how to subscribe. The offer wasn't clear. The product wasn't clear. The system didn't exist.
The core issue wasn't the food. It was that no system existed to turn attention into subscriptions.
Three root causes — before a single ad ran.
We don't run ads on a broken funnel. Before we touched Meta, we mapped exactly why a subscriber wasn't being made. Three things were missing.
This wasn't a marketing problem. It was a system problem.
Two free moves before we spent a dollar.
We gave Protein Pals two pieces of feedback before formal onboarding — and both started working immediately.
These two moves — before any ad spend — gave Protein Pals their first taste of traction. Inbound DMs picked up. Subscriber count crept past the early-twenties. That was the signal we needed: the foundation was strong enough to put weight on.
What we actually built.
End of May, we directed Protein Pals' first ad shoot. Founders on camera, lifestyle, packaging, real kitchen footage. By the end of June, the backend was live and the ads were running.
The creative strategy
We started with engagement ads paired with ManyChat to validate the offer. Message conversations were coming in at 0.5 CAD each. The automation pooled audiences into categories, qualified interest, and built custom audiences we could re-target later.
When the lead-gen campaigns went live, the leads were pre-warmed. Quality was high from day one. We tested four angles:
Each angle attracted a slightly different segment of the ICP — all valid, all converting. Founder-led ads ran at 3–4 CAD per lead. Then we layered UGC.
UGC dropped CPL from 3–4 CAD to 2–3 CAD with close to zero junk. Real people talking about real food performed better than founder-led ads because it felt like a recommendation, not a pitch.
Then Meta started rejecting our ads.
We aren't shy about sharing great results. We also aren't shy about sharing the hard parts. This was hard.
Starting in August, ad after ad got flagged. Meta's reasons were generic and contradictory — "forms collecting data" cited as a violation, even though Meta's own lead-gen forms exist to collect data. We escalated to Meta support. We challenged the reasoning. We got nowhere. CPL spiked from 2–3 CAD to 4–5, then 8–9 in September.
We iterated relentlessly. Four creatives, four forms, four campaigns — minimal budget, single goal: find what Meta wouldn't reject, then scale it. Protein Pals stayed with us through this period. They believed we'd figure it out. That trust is the foundation of how we work — partners, not vendors.
The comeback
Late September, Saad personally sat through multiple late nights — shortlisting 8–10 creatives from a pool of 20–25, mapping them against 15+ audience segments, scripting a campaign structure designed to withstand Meta's review while delivering peak performance.
That campaign scaled to 70–80 leads per day. 700 leads in 10–12 days. UGC-led creative was now running at 1.13–1.50 CAD per lead — a 60% reduction from the founder-only ads. And then something happened that we consider the ultimate proof of a system working.
Translation: Stop the ads for now. There are already too many leads. We don't have enough people to handle them. The backlog isn't clearing. They paused advertising because demand exceeded operational capacity. They moved out of their small kitchen into a larger commercial facility. They needed more staff to handle calls and deliveries.
This is what a revenue system looks like when it works.
10 months. 1,942 leads.
At a 98.67% pre-qualified rate. The sales team stopped chasing and started picking. The founders ran out of capacity before they ran out of attention.
Cost per lead — the journey
Revenue (conservative)
At a conservative average lifetime value of 100–200 CAD per converted subscriber:
- 242 customers × 100 CAD = 24,200 CAD
- 242 customers × 200 CAD = 48,400 CAD
- On 6–7k CAD total ad spend = 4–8× return on ad spend, minimum.
With 420 leads still in pipeline at the time of writing and a 20–30% retention rate on subscriptions, the actual revenue impact is significantly higher than the floor above.
60–70% of converters were vegetarian.
Protein Pals didn't expect this. Neither did we. But the CRM — which captured dietary preference on every single lead — made it undeniable.
The majority of their paying subscriber base preferred veg meals. This insight reshaped the entire strategy. We adjusted ad creatives to lead with vegetarian dishes. We briefed UGC creators to feature veg meals prominently. We restructured targeting to lean into this audience segment.
The point: a lead generation system isn't just about generating leads. It's about generating intelligence. The CRM data didn't just help us optimise ads — it helped Protein Pals understand their own market better than they did before.
The loop, end-to-end.
Four components, one feedback loop. Each is a service we run — but only as part of one revenue system, never separately.
- A prospect sees an ad on Instagram or Facebook.
- They engage — tapping "Menu" (PDF menu with dish photos) or "Trial" (free 2-day trial) in ManyChat.
- The automation qualifies them by capturing name, zip code, city, dietary preference, profession, and health goal.
- Their data flows into Zoho CRM, categorised and tagged.
- The founders call the lead, have the conversation, and update the CRM with the outcome — qualified, contacted, closed, junk.
- That outcome flows back to Meta via the Conversion API.
- Meta's algorithm uses that feedback to find more leads like the converters — and fewer like the junk.
Every cycle makes the system smarter. Every lead gives it more data. Every conversion teaches it what a good customer looks like. That's why the CPL dropped from 3–4 CAD to 1.13 CAD over time — the system was learning.
This is the difference between running ads and building a revenue system.
What Protein Pals didn't expect.
When Rohan and Mayank came to us, they expected ads. What they got was a system.
They expected leads. What they got was market intelligence — the discovery that 70% of their customers were vegetarian reshaped their entire menu strategy. They expected a vendor. What they got was a partner who sat up late rebuilding campaigns when Meta kept rejecting them, who told them to stop running ads when demand exceeded capacity, and who treated their business like our own.
That's the Daee difference. We don't run ads. We build systems that generate revenue.