Renewing or Cancelling a Data Subscription
The recurring decision of whether a data vendor is still earning its invoice, and the lightweight process research teams use to make that call before an auto-renewal locks them in for another year.
Prerequisites: A Scorecard for Ranking Candidate Datasets
Data subscriptions are easy to start and easy to forget. A vendor feed gets onboarded for one project, the project ends, but the invoice keeps arriving because nobody owns the decision to cancel it. Left unmanaged, a research desk can end up paying for a dozen feeds nobody actively uses, while genuinely useful ones sit unrenegotiated at outdated prices. The renewal decision needs an owner and a recurring checkpoint, not a default to "keep paying."
The practical process is simple: before each subscription's renewal date, someone checks how many live strategies or models actually pull from that feed, whether any of them still rely on it in production, and whether a cheaper or better alternative has appeared since the last renewal. A feed that no live strategy has touched in six months is a strong cancellation candidate regardless of how useful it once seemed. A feed that several strategies depend on is worth renewing even if it's expensive — but that's also the moment to push back on price, since vendors expect renegotiation at renewal time far more than mid-contract.
Worked example
A firm pays $40,000 a year for an alternative-data feed onboarded eighteen months ago for a single research project. That project was shelved, and a query of production strategy code shows zero references to the feed's tables. At renewal, the desk cancels it, saving $40,000 annually with no impact on any live strategy — the kind of cost identified only because someone actually checked usage before the auto-renewal fired.
Treat every data subscription's renewal date as a forced decision point, not a formality: check actual usage by live strategies, compare against current alternatives, and use the leverage of "we might not renew" to negotiate price — the default of silently renewing is how research budgets quietly bloat with unused feeds.
Further reading
- Kolanovic & Krishnamachari, Big Data and AI Strategies (data vendor sections)