Customer Login  |  

The Human in the Loop

by Kevin Dodds

After half a decade of ever-increasing crowdsourcing volume, IEI has decided to put crowd-centered technology at the center of everything we do. This doesn’t mean that we now only use external crowd workers ‑ on the contrary, our in-house teams continue to grow. It does mean that we are centralizing our work around the WorkFusion platform to make it simpler, faster, more accurate, and, in the end, less expensive for our clients. At the same time, we are preparing projects for work by qualified outside resources when needed.

How is this different from the way we’ve been using crowdsourcing technology?

  • Integration: Steps that come before and after our human workforce’s contribution are tightly integrated into the workflow. There’s less direct data entry and no cumbersome and risky “hand-offs” of datasets. Each task is part of one cohesive, automated process.
  • Scaling: All teams can be augmented by trained external resources if a project gets behind schedule or needs to ramp up to handle higher volume. Extra labor is thus “on tap” and can be quickly deployed to get work back on track.
  • Transparency: Customers can have complete real-time transparency into workflow, worker productivity data, and critical success metrics. In addition, they can also do quality control tasks on the platform themselves, if desired.

With regard to integration, we have several data processes that begin with the extraction of data from given sources. These may be repositories of structured data files such as PDFs, digital image files or Word docs, or websites where new content is retrievable via change monitoring and harvesting. In both cases, we’ve used unconnected, distinct processes up front to retrieve data, only to then load that data into the editorial systems we use to for data disambiguation, research, editing, and validation. Now we no longer need to separate these processes or spend time coordinating exports and imports. This saves time and eliminates the risk of potential hand-off errors.

Post-processing steps like data validation and quality assurance (e.g., confirming the deliverability of an email or street address) can also be part of the platform. During QA, reviewers can easily search for workers by their production volume, throughput times, results of past QA checks, and many other parameters before doing their reviews. This improves QA throughput and accuracy in two ways. First, better targeting of QA resources reduces error rates. Second, centralized storage allows easy review of source metadata such as URLs, file names, telephone survey recordings, etc.

In terms of scale, having clients with high-volume reoccurring work over the years has helped us assemble a solid group of highly qualified individuals from all over the world who can execute large data campaigns on demand. We are adding new qualifications to this labor pool for handling a range of new project types beyond our core search and retrieval work:

  • Outbound telephone research
  • Image tag curation
  • Data acquisition in multiple foreign languages
  • Data extraction and normalization from invoice image files
  • Data capture of SEC filing data

Finally, as we expand the variety of services we offer and our workflows become more complex, it’s increasingly important for our customers to see what’s going on “under the hood” while we run their data campaigns. By embracing the WorkFusion platform, we accomplish this significant goal and can allow our hands-on customers to do deep forensic analysis of their data while work is underway.

If you would like to learn more about our crowd-centered approach to data management and watch video clips of how some of our processes work, just drop me a line.

posted by Shyamali Ghosh on November 10, 2015

{ 0 comments… add one now }

Leave a Comment