Measure an AI giveaway from each user's renewal date
A promotion closing is not a free subscription expiring. Perplexity’s Airtel offer needs activation cohorts before its paid conversion can be judged.
January 16 ended new redemptions of Perplexity's Airtel promotion. It did not switch every participant from free to paid on January 17.
Perplexity's offer terms give eligible users twelve months of Pro from activation, with additional timing rules for existing subscribers. The Airtel announcement introduced the offer in July 2025. Someone who activated near the end therefore has a different renewal date from an early participant.
TechCrunch reports that downloads fell after redemptions closed, while usage remained above the pre-offer level. Its August 18 report is a useful prompt to investigate the promotion's economics.
But the clock for that investigation has to start at the individual user's first paid decision. I would reject a conversion chart that starts every participant's post-subsidy period on the campaign's closing date.
Three clocks, three different questions
There is a campaign clock: when the offer was available. There is an entitlement clock: when a user could actually use the free product. There is a billing clock: when that user faced a charge, renewed, cancelled or failed payment.
A download can occur on one clock without advancing the others. Eligibility is also not redemption, and redemption is not sustained use. Blending those events into one funnel makes it difficult to locate what the promotion achieved.
As an illustration, I would build two hypothetical cohorts, one activating in July and another in January. In the following August, the first cohort could contain people who had faced renewal, while the second could still be enjoying its original free entitlement.
Calling both groups retained after subsidy would mislabel the second group. Calling everyone who had not paid a failed conversion would make the opposite mistake. The product has not yet put every participant through the same decision.
My dashboard would therefore index time relative to entitlement expiry. Day zero means the same commercial event for every eligible user in that view. A separate calendar view would still help the operating team forecast support and billing volume.
Paid conversion needs a denominator you can defend
I would start with users who actually reached their first paid renewal opportunity, then distinguish a successful charge, a refund, a cancellation and continued paid use. Automatic renewal deserves its own interpretation; an initial charge alone is weaker evidence of durable willingness to pay than subsequent retained use.
Revenue also needs its channel boundary attached. Mobile-store receipts cannot automatically stand in for web subscriptions or the whole business. Gross customer payments, platform deductions and net receipts answer different questions. Mixing them into one headline ratio creates precision without comparability.
None of this proves the promotion worked. A generous offer can attract low-intent users, and a surviving free audience can still be expensive to serve. Nor does a cohort chart establish what would have happened without the offer. A credible causal comparison needs a design that addresses who received the promotion and why.
What the cohort chart does provide is a fairer operational question: after a particular group loses its free entitlement, what happens to use, payment and contribution? That answer can inform the next offer's duration, targeting or onboarding.
Before judging another AI giveaway this week, ask for its activation dates and entitlement rules. The campaign calendar is usually easier to find. It is also easier to mistake for the customer's decision.
An AI giveaway is tested at each cohort's renewal, not at the promotion's closing date.
