Each project arrives with its purpose, milestones and first tasks written out. Here is how the first three read.
Alternative data research charter and scope
Without a written scope, every vendor cold call becomes a trial and the research calendar fills with data nobody can use. A one-page charter stating which strategies and horizons you serve, which data categories are in scope, who signs off on spend and which categories are off limits for privacy or compliance reasons gives you a reason to say no quickly and a standard to say yes against.
First tasks
- List the strategies and holding periods the data you buy must serve
- Mark each data category as in scope, out of scope or compliance first
- Write the spend limits and the person who approves each level
- Review the charter with research and compliance once a year @recurring(yearly)
Vendor and data set intake register
Pitches arrive by email, at conferences and through brokers' data marketplaces, and the same panel can be offered by three resellers under three names. One register holding every data set you have heard of, with its source, category, status and the date you last spoke, stops duplicate trials and lets anyone on the team see what was already rejected and why.
First tasks
- Search your inbox for data vendor pitches from the last twelve months
- Create the intake register with vendor, source, category, status and owner
- Enter every pitch, including declined ones with the reason
- Link resold data sets to the vendor that collects the raw data
Standard alternative data evaluation scorecard
Comparing a receipts panel with a shipping data feed is hard unless both are judged on the same questions. A scorecard covering coverage, history length, point-in-time quality, latency, legal risk, signal evidence, cost and exclusivity, each scored one to five with a written reason, makes evaluations comparable across researchers and across years.
First tasks
- Draft the eight scoring criteria with a one-line definition each
- Describe what a score of one, three and five means for each criterion
- Rescore two past evaluations to test the scale
- Agree the weights with your research lead