Tuesday, December 9, 2008

The Contact Washing Machine


I think we can all agree that B2B marketing has shifted from a purely creative discipline to a much more operational, process-oriented, data-centric, analytical discipline. It's a long road though, as the data we get to work with is often... well... terrible.

Data comes in through so many sources, most of which are not controlled in any way - free-form text. Web forms, tradeshows, lists, internal systems, etc. We're then expected to use that data for analysis, segment targeting, lead scoring, etc.

The best thing that anyone in Demand Generation can do, to set up for long term success, is to build a "Contact Washing Machine" that standardizes and normalizes this data as it comes in. For Eloqua users, I'll talk about some of the pieces of a good Contact Washing Machine on this blog's sister blog - Eloqua Artisan (http://eloqua.blogspot.com/2008/12/whats-in-name-job-titles-and.html).

If not, the things I typically see addressed on all incoming data are:
  • Standardize Title to allow it to be used for Segmentation or Scoring
  • Standardize Country to a 3-letter or 2-letter code
  • Fix Zip codes from New England... they have a leading 0 and Excel drops that if the list has ever been in Excel
  • Map to sales (or field marketing) territory
  • Validate physical address (if you'll be using this)
  • Match company to existing company list (or use a standard code such as DUNS)
  • Map industry to SIC, NAICS or other industry code (if this is important for you)
  • Any other data standards or fields that are key to your business

Putting this in place, and having all incoming data flow through it, is a great way to avoid the cycle of data continually degrading over time until a major undertaking (at great cost) is done to cleanse it, whereupon it immediately begins degrading again. Time is too short for bad data...

BOOK
Many of the topics on this blog are discussed in more detail in my book Digital Body Language
SOFTWARE
In my day job, I am with Eloqua, the marketing automation software used by the worlds best marketers
EVENTS
Come talk with me or one of my colleagues at a live event, or join in on a webinar

7 comments:

Neil Sequeira said...

Spot on Steve! The primary cause for messy CRMs is the fact that data from all the different sources is dumped right in without going through any intermediate "Contact Washing Machine" before it gets to the CRM.

Neil
ReadyContacts

Steven Woods said...

Thanks Neil, it's definitely a key requirement for success. I would say that I've noticed a sea-change in awareness of how critical data is to success in marketing lately. 2009 will be a big year for data quality, I suspect.

mkamrk said...

Steve, we too hope and believe B2B marketing data management will evolve into a critical new category in 2009 and beyond. I think a lot has to do with marketers realizing that lead data management is as critical as CRM investments, infact more important as a CRM with bad data is just not valuable in anyway. At ReadyContacts (http://www.readycontacts.com), we have been laser focus on this niche and I'd love to borrow your "Contact Washing Machine" name to explain a lot of what we do :-)

Tim Wilson said...

You need to be careful here, Steve. Contact data management is a balancing act, and some elements of this post start to smell a bit like you've removed your customer-centricity hat while writing it. "Standardizing title" is a good example -- sure, marketers want the title standardized, but titles are all over the map across companies. So, "standardizing" means either "make a really long list so that everyone can find a close approximation of their title" or "make a short list and let the customer pick which one most closely approximates his actual title." BOTH introduce data quality issues: the first because many people will just give up finding the best match and pick any one that might be remotely reasonable, and the latter because you're expecting your lead to have a perspective that matches yours to do that mapping. If you also make this *required* information, then you would potentially really be deluding yourself if you do data profiling on the field: "This field is 95% populated and all of the values are standardized! Our data is 'clean.'" Likewise, I've *never* had a positive user experience on a site that requires me to "look up" my company -- DUNS does not have 100% coverage, it's got hierarchies built into it, and it's a lonnnnnnnnnng list to try to get through. "Physical address" is a good one -- if the contact is US-based, then there are standards like the USPS CASS standard and online tools to standardize and validate that address...but the contact has to have the opportunity to override what the tool spits back (either as a "corrected" address or as a "we can't validate that address" error). All of that being said, I 100% agree that as much cleansing at the point of entry should be included. But, it's also critical to realize this isn't black and white -- "getting the cleanest possible data" HAS to be weighed against "promoting a positive user experience."

Steven Woods said...

Tim,
Great points and I should actually have clarified some more options for those more difficult situations like title. Depending on situation, an option that can work well is (a) free form text on the web form for the optimal user experience, (b) secondary fields within the marketing platform for "normalized title level" and "normalized role", (c) the data normalization on incoming data touchpoints filters the free-form data into normalized fields, while leaving the free-form text in place, and (d) rules such as lead scoring, routing, and personalization can be built on the normalized fields.

Long winded, but hopefully that clarifies a bit more about the execution of the inline data management.

Is that more in line with what you have seen working Tim?

Tim Wilson said...

Hah! I wish I really had "the answer" on that one. I've seen the extremes on both ends (forcing the user to do much translation all the way to letting it be totally freeform), and my suspicion is both wind up being worse than a middle ground. What you're proposing...is pretty intriguing. It seems like a bit of cleverness and a tool that can do parsing and thesaurus-management (I don't know if that would be a data cleansing tool like Trillium...or more of a taxonomy/search tool like Endeca -- leaning towards the latter) would be what's really needed. BUT...interesting possibilities: minimal freeform-type text, combined with some more solid selectable fields, run that through a mapping engine, and then generate a welcome e-mail that tries to play back what you "know" about the user (and provides content that would be relevant to him/her) and asking them to confirm/tune their profile. Oh...to dream...!

Arunav said...

Thats really great. data cleansing and extracting meaningful information is one of the greatest challenge faced today. Check out the free tool Google refine from google.