How would you evaluate a dataset we're going to purchase that has competitor data on rides (in the U.S)?

The first question that I would think of when evaluating a dataset that contains data from a competitor that is available to purchase is whether the data is usable.

This is a two-fold question, the first authenticating the source of the data and how it was collected or obtained, and the second is whether is ok for the firm to actually use that data for research or competitive advantage.

To move the question forward, I'm going to assume that this data is cleared for internal use by our legal teams.

A clarification question I have about this exercise then is, what are we looking to evaluate - whether the firm should purchase the data, or how can the firm use the data if it was purchased.

Evaluate whether we should purchase the data

Assuming this data is for sale, it is unlikely the entire dataset is available to be reviewed, so I'd request a snippet of the data - an entire cross-section.

Along with an estimate of the number of unique records.

Using this snippet, one can infer the model/design of the dataset.

We should check if the most important pieces of data that I would want to see in a rides dataset are available.

Data such as geolocation, timestamps, rides start and ride end, ride cost, unique users etc.

I'd also from the granularity of the data if it can be logically aggregated to a usable level, for example, data pre-aggregated to a customer level would not help us understand anything about trip length.

Next, I'd ask for certain data description functions such as data formats for the columns, missing data count, statistical information about numeric columns to evaluate data completeness.

Outside of metadata checks, I'd be interested in the recency of the data and the demographic it represents. Is it in a place we already do business, want to do business, or somewhere we have and foresee no presence? ie. evaluate the business sense for the company to acquire the dataset.

It would be nice to have documentation describing the data and its usage!

In case the dataset is purchased, we would want to extract as much useful information from it as possible.

On the full dataset, I'd rerun the data descriptions tests, followed by trying to fit the data to a design that is already used by Uber.

If that is possible/easy to do, one can run all the metrics that UBER runs on its data sets to evaluate its business.

In case that is not possible, one might need to write our own logic to create comparable metrics. 

We can run metrics like DAU, Trip times, average customer LTV, average driver LTV, wait times etc. If UBER has a presence in the same city, it might be possible to evaluate its performance in context with the competitor and identify key areas of improvement along with identifying the areas where UBER is already doing very well in. It would help guide the next experiments, products, marketing campaigns, and other areas of interest for the firm.