"There is money to be made in e-discovery," said Jason Lichter, director of discovery services and litigation support at Pepper Hamilton. "But if you were to plot on a chart the per [gigabyte] processing charges vendors have charged over the last 10 years, it would be a precipitous fall."
Vendors used to charge upwards of $2,500 per gigabyte of information, but now charge almost nothing for processing, Lichter said. He said they look to recoup money through hourly charges at time of production.
That is where Michael Boland of Drinker Discovery Solutions, a recently formed subsidiary of Drinker Biddle & Reath that brought the e-discovery delivery model in-house, thinks a firm can do better. He said he hasn't seen a vendor that is cheaper than what his company is offering. And Drinker Discovery Solutions offers flat fees or retainers to give clients predictability in costs, Boland said.
But Boland is aware of the timeframe for making a profit on e-discovery work. First there is the technology side. He said the investment in a firm's technology needs to see a return on investment that outpaces the timeframe for when that technology will become obsolete. In the e-discovery space, that timeframe is three to five years, he said.
In terms of the entire e-discovery service delivery model, Boland said there is money to be made, "but on a diminishing scale as time goes on."
There may be between five to eight years for firms to make money off of this before it is just something they have to provide or know who to call to stay competitive, he said. The pricing models are "ridiculous," Boland said, adding they are "very commoditized." He said his firm doesn't want to get into beating down prices for managing a gigabyte of data from 4 cents to 2 cents.
"There's no place for it to go except away so you either have to change your pricing model or go to an alternative one," Boland said.
With little to be done on the collection and processing side in terms of price reduction, the Rand study cited predictive coding as the best way to reduce document review charges even further.
"Every document you don't review is money saved," Berrent said.
Predictive coding provides for a human to review a sample set of documents so that the computer program can "learn" what search terms to find and then take over the bulk of the review. Rand noted predictive coding has not seen widespread use for a number of reasons, including namely lack of judicial direction and outside counsel's concerns over testing new technology at the risk of their clients.