Before I took over full ownership of the renewals revenue number, no one on the marketing side ran a rigorous analysis of the funnel behind it. Success was measured purely on the rate at which eligible campaigns renewed. New-product sales were growing fast enough that renewals kept climbing on their own, so the number was treated as a given — gravy on top of the sales total rather than a stream that needed forecasting. The CFO maintained a model of renewal contribution to the P&L, but there was a lack of cross-departmental communication.
When new sales slowed to the point where they were no longer replacing un-renewed campaigns, we hit a leaky-bucket problem: total renewal revenue started declining even though the renewal rate itself hadn't moved. I was asked to explain the gap between a healthy renewal rate and a shrinking number, and to quantify what we should expect in the next calendar year.
I started by analyzing what actually predicted whether a campaign would renew. The strongest factor turned out to be how many times a campaign had already renewed, with a secondary effect from contract term (Adwerx's brand products auto-renew monthly, quarterly, annually, or bi-annually until canceled). I used historical renewal rates for each of those segments as the foundation for the forecasting model.
From there, the model accounted for the full complexity of the business: how many campaigns sat in the renewal base, when each was next due, which renewal-count segment it belonged to, and when in the calendar year that revenue would land. For non-annual campaigns, I calculated how many additional times each would renew within the forecast period and adjusted the rate accordingly at each occurrence. Because the model was built ahead of the fiscal year, I also had to project campaigns renewing before the model's Jan. 1 start date into their post-renewal segment, and layer in a seasonally adjusted forecast of new auto-renewing sales between then and the forecast period, using our existing sales forecast as an input.
Summing all of it produced a monthly renewal forecast for the year, on top of which I built a daily pacing model — accounting for business days, holidays, and days-per-month variance — to track performance against target in real time.