Blog



First time here? Start exploring below.

ReadyContacts is a B2B Marketing Database 2.0 solutions company Learn more.
Qualify target accounts. Speak to role-based decision makers. Start with our FREE TRIAL.
Subscribe to our RSS Feed & follow us on Twitter.


Amazons Mechanical Turk - Hiring Intelligence?

Having checked out Mike Damphousse’s blog post Ethics & Wonder/Amazon’s Mechanical Turk/Kiva on his experience getting a data related task done through the new online marketplace, I had a chance to check it out myself and ponder what this means for us as a B2B Marketing Database Management company. Amazon’s Mechanical Turk has been making buzz waves among marketers among other professionals who have been exploring to see just how they can leverage this to get some of the more routine tasks done. Also getting them done at a very competitive cost by putting them on the listing for providers to bid.

Essentially an Odesk or Elance of task related work, Mechanical Turk is a market place for both providers and requesters of very process oriented tasks like researching, identifying, labeling, transcribing and data related tasks which need some amount of human intelligence as opposed to tasks that can be automated. With several willing providers who can work from the comfort of their homes a lot of these jobs can get done for a fiercely competitve price which even outsourcing companies wouldnt be able to match given their operating costs. Will this mean all marketing data related work is going to head towards this?

I’m not sure it will. A good part of what we do at ReadyContacts does involve that “human touch” simply because there is a difference in quality between automating that process and using human intelligence to get the results. For example we could simply build B2B decision maker lists by pulling records from an internal database and delivering them at a competitive price but we don’t. Our team calls into every target account and locates decision makers by asking others in the company for their specific job role rather than their job titles. Eventually a sales person will need to go through this step to identify if they are really talking to the right decision maker. Calling in to build a list is a lot more expensive but in the bigger picture, we deliver something that will save our customers time and help them get to their decision makers quicker. 

Similarly there are de-duplication applications available that will help you clean your CRM data and remove redundancy and help you lower costs as against hiring someone to do this. Will it work the same? Not entirely. For example, it would help remove two accounts titled “Citi Group” or two leads named “John Smith” but it wont know “Peoplesoft” is a part of “Oracle” or know if a certain company has been taken over, merged or a lead is no longer working in the same company. The same goes for email verification and appending. There is some human intelligence required in most of these tasks to do them better. Does this mean there wont be room for ReadyContacts and other B2B data companies?

Well Odesk and Elance have been doing millions of dollars worth of software development projects for the masses with very capable developers as far as India, Israel, Ukraine and Singapore sucessfully delivering projects. So I’m sure this can be replicated with Mechanical Turk for other tasks. However Odesk and Elance haven’t put larger software companies like Infosys out of work, its only opened a door to the masses as an affordable option to getting development support and Mechanical Turk should have the same effect. We’ll have to wait and watch.

Any readers with experiences having used Mechanical Turk for any of their projects?

 

 

  • I'll be a bit more specific as to what our project of 1000 tasks was, and I think it was a perfect fit for mturk. We had a list of companies provided by a client that they want us to go target for appointment setting. We then bounced that against a list of contacts we own using Soundex and some other techniques such as email matching to match up contacts with target companies. The issue we have is that Soundex only goes so far. Some companies just don't make the right match. So we posted possible matches and asked for people to look at the match and tell us if the companies were the same or not. Example:

    Sage and Sage Software Match ( ) Not a Match ( )

    In this case, human decisions round out the work that was started by algorithms. Good fit for this application.
blog comments powered by Disqus
  • Recent Posts
 
 
 
 
 
 

Monthly Archives

Popular Tags

Categories
B2B Lead (24)
CRM Data (26)
D&B (2)
ebook (1)
Hoovers (7)
InfoUSA (1)
Interview (2)
Jigsaw (7)
Lead Lists (29)
LinkedIn (7)
OneSource (2)
Spoke (2)
SugarCRM (1)
Twitter (4)
Web leads (8)
Zoominfo (4)