Case Study
The Followers Were Bought. So Was the Audience Meta Kept Finding.
Why an astrology practice with a large following couldn't buy a decent lead — and the funnel we built once the signal was clean.
- Brand
- Astrology practice (name withheld under NDA)
- Role
- Marketing Consultant (external)
- Period
- November 2023 – April 2024
- Market
- India
Jump to:The Short AnswerThe SituationReading the ProfileCleaning the SignalThe Trust Ladder FunnelThe Result, Stated HonestlyHow to Apply ThisEvidence & Limits
The short answer
The brief was simple: the leads were poor. Everything came from boosting posts, nothing came organically, and roughly 1 in 50 leads turned into a paying client.
Charting the funnel, the profile didn't add up. Subscriber-to-view ratio was badly out of line. Some videos had views and no comments or shares at all; others had nothing. The content hadn't changed, but the numbers were erratic in a way content alone doesn't explain.
We asked directly, and the client told us straight: engagement had been purchased — followers and views, bought to build the profile up. That candour is the only reason this was fixable.
It also explained the lead quality, which nobody had connected. Meta reads signals from your profile, your ad account and your past campaigns. Train it on purchased engagement and it will faithfully go and find more people who behave like the ones you bought.
We stopped the purchasing, rebuilt organic on what was demonstrably working in the category, and archived the posts that were dragging the signal down. Reach recovered in about two weeks. Then we built the funnel — and conversion went from roughly 1 in 50 to about 10 in 50.
Charting the funnel, the profile didn't add up. Subscriber-to-view ratio was badly out of line. Some videos had views and no comments or shares at all; others had nothing. The content hadn't changed, but the numbers were erratic in a way content alone doesn't explain.
We asked directly, and the client told us straight: engagement had been purchased — followers and views, bought to build the profile up. That candour is the only reason this was fixable.
It also explained the lead quality, which nobody had connected. Meta reads signals from your profile, your ad account and your past campaigns. Train it on purchased engagement and it will faithfully go and find more people who behave like the ones you bought.
We stopped the purchasing, rebuilt organic on what was demonstrably working in the category, and archived the posts that were dragging the signal down. Reach recovered in about two weeks. Then we built the funnel — and conversion went from roughly 1 in 50 to about 10 in 50.
The transferable principle
Buying engagement doesn't just fail to help. It teaches the ad platform to find you more people who behave like the ones you bought.
The situation
The client is an independent astrologer in India selling consultations and personalised reports, and isn't named here. The engagement ran from November 2023 to April 2024.
Acquisition was entirely boosted posts on Instagram and Facebook. No structured campaigns, no funnel, and effectively no organic lead flow. The ask was narrow and reasonable: better quality leads.
We started where you'd expect — by looking at what the practitioners who've built real brands in this space actually do. Astro Arun Pandit, Sundeep Kochar, Sohini Sastri and others in that tier converge on a common structure: sell a low-ticket personalised product — a kundali, health, marriage or finance report at ₹99 to ₹999 — and treat that purchase as the lead. The buyer has demonstrated intent with money, and premium services get offered to that list afterwards over WhatsApp and email.
That model was clearly right. But before we could build it, the profile told us something else was wrong.
Acquisition was entirely boosted posts on Instagram and Facebook. No structured campaigns, no funnel, and effectively no organic lead flow. The ask was narrow and reasonable: better quality leads.
We started where you'd expect — by looking at what the practitioners who've built real brands in this space actually do. Astro Arun Pandit, Sundeep Kochar, Sohini Sastri and others in that tier converge on a common structure: sell a low-ticket personalised product — a kundali, health, marriage or finance report at ₹99 to ₹999 — and treat that purchase as the lead. The buyer has demonstrated intent with money, and premium services get offered to that list afterwards over WhatsApp and email.
That model was clearly right. But before we could build it, the profile told us something else was wrong.
Reading the profile
Three things stood out, and none of them were about content quality.
Subscriber-to-view ratio was implausible for the size of the following. Some videos had views but no comments and no shares — not few, none. Others had effectively no distribution at all. Meanwhile the client had been producing broadly the same kind of content throughout, so genuine performance shouldn't have swung that violently.
That pattern is a signature. Purchased views arrive without the behaviour that accompanies real attention — nobody who was paid to watch leaves a comment or sends the video to a friend. The result is a profile that looks strong in aggregate and falls apart the moment you look at the ratios between metrics rather than the metrics themselves.
I want to be careful about how this reads, because it is genuinely common. Buying followers and views is widespread in the Indian creator economy and it is usually not done cynically — it is done because early distribution is brutally hard and a large number feels like it should help. The important part of this account isn't that a client had done it. It's that when we asked, they told us the truth immediately and stopped. Most of the value I added here was only available because of that.
Subscriber-to-view ratio was implausible for the size of the following. Some videos had views but no comments and no shares — not few, none. Others had effectively no distribution at all. Meanwhile the client had been producing broadly the same kind of content throughout, so genuine performance shouldn't have swung that violently.
That pattern is a signature. Purchased views arrive without the behaviour that accompanies real attention — nobody who was paid to watch leaves a comment or sends the video to a friend. The result is a profile that looks strong in aggregate and falls apart the moment you look at the ratios between metrics rather than the metrics themselves.
I want to be careful about how this reads, because it is genuinely common. Buying followers and views is widespread in the Indian creator economy and it is usually not done cynically — it is done because early distribution is brutally hard and a large number feels like it should help. The important part of this account isn't that a client had done it. It's that when we asked, they told us the truth immediately and stopped. Most of the value I added here was only available because of that.
Cleaning the signal
The problem was now bigger than the brief. It wasn't only that the leads were poor — it was that every acquisition surface was being optimised against a corrupted picture of who this audience was.
Three changes:
Stop the purchasing entirely. Every additional bought view was another training example telling Meta what a typical viewer of this account looks like.
Rebuild on what was demonstrably working in the category. We gave the client scripts and formats to start with, but the durable part was that they took over the research themselves — studying what was performing for comparable practitioners and producing to that pattern. That mattered more than our scripts did, because it made the capability theirs.
Archive the posts dragging the signal down. Videos with no genuine likes, comments or shares were removed from the profile so that the average the algorithm reads reflected real audience behaviour rather than the purchased history.
Reach recovered in roughly two weeks. That was faster than I expected, and it says something useful: these systems are reading recent behaviour heavily, so a corrupted signal is not a permanent sentence. It is a debt you can pay off quickly if you stop adding to it.
Three changes:
Stop the purchasing entirely. Every additional bought view was another training example telling Meta what a typical viewer of this account looks like.
Rebuild on what was demonstrably working in the category. We gave the client scripts and formats to start with, but the durable part was that they took over the research themselves — studying what was performing for comparable practitioners and producing to that pattern. That mattered more than our scripts did, because it made the capability theirs.
Archive the posts dragging the signal down. Videos with no genuine likes, comments or shares were removed from the profile so that the average the algorithm reads reflected real audience behaviour rather than the purchased history.
Reach recovered in roughly two weeks. That was faster than I expected, and it says something useful: these systems are reading recent behaviour heavily, so a corrupted signal is not a permanent sentence. It is a debt you can pay off quickly if you stop adding to it.
The Trust Ladder Funnel
With a clean signal, we built the acquisition structure — the one I now use across clients and call the Trust Ladder Funnel.
The principle underneath it is sell to qualify. Conventional funnels qualify a lead and then try to sell to them. This inverts that: you sell something small first, and the purchase itself is the qualification. Someone who has paid ₹499 has told you something no form fill can.
Three steps, each asking for slightly more than the last.
Step one, the ads. A short video giving a quick, direct answer to a real problem and ending on the offer — cheap attention that carries click-through. Alongside it a longer video: the personalised report explained properly across career, finance and relationships, plus who the astrologer is and who they've helped. That one isn't buying clicks, it's buying qualified ones. Splitting the two jobs across two creatives is the part most people skip, and it's why click-through and conversion rate can be managed independently instead of traded off.
Step two, the video sales page. The product in depth, objections handled before they're raised, and the ₹499 purchase sitting on the page. The price is deliberate: high enough to prove intent, low enough not to require a decision.
Step three, WhatsApp. A live channel catching whatever doubt survived the page — which on a spiritual purchase is rarely nothing.
Premium services are then offered to people who have already bought once — which is a fundamentally different conversation from a cold pitch.
The principle underneath it is sell to qualify. Conventional funnels qualify a lead and then try to sell to them. This inverts that: you sell something small first, and the purchase itself is the qualification. Someone who has paid ₹499 has told you something no form fill can.
Three steps, each asking for slightly more than the last.
Step one, the ads. A short video giving a quick, direct answer to a real problem and ending on the offer — cheap attention that carries click-through. Alongside it a longer video: the personalised report explained properly across career, finance and relationships, plus who the astrologer is and who they've helped. That one isn't buying clicks, it's buying qualified ones. Splitting the two jobs across two creatives is the part most people skip, and it's why click-through and conversion rate can be managed independently instead of traded off.
Step two, the video sales page. The product in depth, objections handled before they're raised, and the ₹499 purchase sitting on the page. The price is deliberate: high enough to prove intent, low enough not to require a decision.
Step three, WhatsApp. A live channel catching whatever doubt survived the page — which on a spiritual purchase is rarely nothing.
Premium services are then offered to people who have already bought once — which is a fundamentally different conversation from a cold pitch.
Boosted postsObjective: Engagement
Boosted postOn a corrupted signal
Free enquiry
Pitch consultation
Roughly 1 in 50 became paying clients — the lead cost nothing to give, and proved nothing.
Trust Ladder FunnelObjective: Direct sales
Short video adQuick answer — buys reach
Long video adDepth and credibility — buys qualified clicks
Video sales pageObjections handled
Convinced
Buys report₹499 — intent proven
Residual doubt
WhatsApp query
Answered, then buys
About 10 in 50 became paying clients — and every lead had already paid once.
The result, stated honestly
Lead-to-client conversion moved from roughly 1 in 50 to about 10 in 50 — 2% to 20%.
Now the caveat, because it matters and it is the argument rather than a hole in it. Those two populations are not the same. An old lead was a free enquiry from a boosted post. A new lead is someone who has already paid ₹499. Part of that tenfold gap is the funnel converting better, and part of it is that we changed what counts as a lead.
That is the entire thesis. Paying for qualification is precisely what makes the downstream number possible. Anyone comparing a free-lead conversion rate against a paid-lead conversion rate and presenting it as a clean tenfold improvement is either not thinking carefully or hoping you won't. The honest version is more useful: this structure produces fewer leads, each of which is worth roughly ten times more to your sales process.
Which is also why the low-ticket product should never be evaluated on its own margin. At ₹499 it barely matters as revenue. Its job is filtering, and it does that better than any lead magnet, because the filter is a payment.
Now the caveat, because it matters and it is the argument rather than a hole in it. Those two populations are not the same. An old lead was a free enquiry from a boosted post. A new lead is someone who has already paid ₹499. Part of that tenfold gap is the funnel converting better, and part of it is that we changed what counts as a lead.
That is the entire thesis. Paying for qualification is precisely what makes the downstream number possible. Anyone comparing a free-lead conversion rate against a paid-lead conversion rate and presenting it as a clean tenfold improvement is either not thinking carefully or hoping you won't. The honest version is more useful: this structure produces fewer leads, each of which is worth roughly ten times more to your sales process.
Which is also why the low-ticket product should never be evaluated on its own margin. At ₹499 it barely matters as revenue. Its job is filtering, and it does that better than any lead magnet, because the filter is a payment.
How to apply this
What transfers out of this account into any similar one:
- Read the ratios between metrics, not the metrics. Views with no comments or shares, or a follower count that doesn't match reach, is a signature. Aggregate numbers hide it; ratios expose it.
- Ask the client directly, without making it an accusation. Bought engagement is common and rarely cynical. You need the truth more than you need to be right about it, and you will only get it if answering honestly is survivable.
- Understand that a corrupted profile signal is an advertising problem. The platform learns who your audience is from your organic behaviour. Fake it, and it will optimise faithfully toward more of the same.
- Stop, rebuild, and archive. Stopping alone leaves the history in place. Removing the posts that misrepresent your audience is what lets the recent signal dominate — and recovery can take weeks rather than months.
- Hand the content research back to the client. Scripts you write are a dependency. A client who has learned to study what works in their own category is an asset that outlives the engagement.
- Sell something small to qualify, rather than qualifying and then selling. A payment is the only intent signal that costs the user something to give.
- Split the top of the funnel across two creatives. One short ad for reach and click-through, one long one for qualification. Asking a single asset to do both is why those two metrics usually trade off.
- Judge the low-ticket product on the leads it filters, not the margin it makes.
Evidence and limits
This is the least verifiable case study on this site, and I'd rather say so than dress it up. The client is under NDA — not named and not linked. There is no live URL you can inspect and no third party you can ask.
The figures are ratios rather than absolutes: roughly 1 in 50 before, about 10 in 50 after, and reach recovering in around two weeks. Spend, lead volume and revenue aren't disclosable. As set out above, the before and after populations are defined differently, and that difference is part of the result rather than something the numbers correct for.
What you can check independently is the model itself — the practitioners we studied are public, and their low-ticket report offers are visible on their own channels.
The figures are ratios rather than absolutes: roughly 1 in 50 before, about 10 in 50 after, and reach recovering in around two weeks. Spend, lead volume and revenue aren't disclosable. As set out above, the before and after populations are defined differently, and that difference is part of the result rather than something the numbers correct for.
What you can check independently is the model itself — the practitioners we studied are public, and their low-ticket report offers are visible on their own channels.
What can be independently checked:
