A recent flurry of press attention to the suggestion that Secretary Clinton could use data from her upcoming book tour to aid her presidential campaign caught our attention.

Ultimately, Simon & Schuster said that Clinton would “unequivocally” not have access to data they’re collecting.

But if she does decide to run, that data could be extremely useful, if used in the right way.

It used to be that you could tell a lot about a person by what they kept on their bookshelves. Today, even with the proliferation of electronic publishing, we can still use what people choose to read in their spare time as a lens to predict their voting behavior. You still can’t judge an e-book by its cover, but you can learn a lot from the patterns of the people buying it.

As a general rule, organizations should be collecting data on every interaction they have with people, in every channel. Whether they are ready to put it to use today, or planning for a potential future.

When it comes to book purchases, retailers will never release the individual data they collect on purchases. (And this is probably a good thing.) However, it is certainly feasible to release sales data by small units of geography, like county or ZIP code – either publicly, or to publishers and authors. This data would be incredibly useful for directly targeting or modeling support at a geographic level, particularly in instances where we can identify specific books which closely track voter’s expressed preferences and likely behavior.

Specific titles can provide a lot of insight into expected behavior all across the political spectrum:

Examples from current best-seller lists: Thomas Piketty’s “Capital in the 21st Century” / Elizabeth Warren’s “A Fighting Chance” / Ben Carson’s “One Nation” / Tim Geithner’s “Stress Test” / Arianna Huffington’s “Thrive: The Third Metric” / Sheryl Sandberg’s “Lean In”.

On the individual level, we can predict with a fair degree of certainty how a buyer of any of these books might vote. But people buy books for lots of reasons, and we caution against making the mistake of directly imputing a political bent to a reader who might pick up a book simply for reasons other than to satiate their ideological views.

But, in the aggregate, we can use purchases of certain titles to contribute to predicting expected voting outcomes at the individual, group and geographic levels.

In the 2008 and 2012 general elections, campaigns had a very basic sense of this dynamic, as book releases by candidates like President Barack Obama’s best-selling “The Audacity of Hope” and Senator Rick Santorum’s “It Takes a Family” set the stage for political movements. But the e-book industry was then a mere fraction of what it is today.

With the 2014 midterms around the corner, and the 2016 presidential primary season to follow soon on its heels, to get a sense of where their prospective voters’ hearts are leaning, smart campaigns will be wise to consider their potential supporters’ bookshelves in addition to their voting behavior.

Michael Simon is President and co-founder of HaystaqDNA. In 2008, he ran the in-house Obama for America analytics department.