IMS acquires SDI: the data firm nesting continues

The news that IMS Health is acquiring SDI (see IMS press release here) is a jolt to those of us who have worked in the pharmaceutical data industry for ten years or more.

Both companies declined to discuss the acquisition beyond a joint statement released last Friday afternoon. Unnamed company sources told Ed Silverman, who first reported the deal on his Pharmalot blog, that as much as 15% of SDI's staff could lose their jobs as a result of the acquisition.

Gary Gatyas, an IMS Health spokesperson, declined to speculate on timing for the completion of the deal, citing pre-merger notification requirements with respect to the Hart-Scott-Rodino Antitrust Improvements Act of 1976.

SDI itself was made larger by the acquisition of Verispan in 2008. Verispan was in fact a conglomerate formed by the 2002 combination of multiple data vendors like Scott-Levin, SMG, and Kelly Waldron. This reminds me of the Russian nested doll pictured here, the matryoshka.

Why have there been so many combinations in the data, and informatics consulting business? Driven in part by profitability issues, intense competition, adn the mergers of the pharmaceutical manufacturers themselves,

If this deal goes through, With this latest IMS acquisition there will come a short term hassle for major pharmceutical clients in 2011 as the landscape gets sorted out. Then longer term, there will be less competition for information sources, and perhaps slow down the rate of innovation.

Campaign Management Playbooks

It's a great deal of effort to design a relationship marketing campaign, from the first brainstorming of business goals, to the user experience designs, through to the final creation of communication pieces and the media planning. It's critical not to let that careful thought go to waste as you are about to head into market.

The campaign management playbook is where all of the operational specifications sit in one well-structured document. The playbook includes the segmentation specifications, and how segments are computed at registration time, or from the database. It maps out the communication plan by segment, and the business rules for determining under what conditions each communication is sent. Also included are the fulfillment requirements for each communication, and each testing plan.

The playbook should not be massive or burdensome; 10 to 25 pages, including figures and tables. This is the necessary script that all vendor partners must follow, so put time into a clear design.

Having this playbook is critical to insuring that when executed in market, your relationship marketing program brings the consumers or healthcare professionals have the experience that was designed.

Is this topic a bit dry? Sorry about that; but the intention with playbooks is to avoid the excitement of in-market mistakes later.

RM differences for consumers and professionals

I have seen this multiple times in each direction: well meaning companies and their I.T. organizations wanting to adapt their consumer relationship marketing systems to handle professionals, or inversely, attempting to modify their professional RM infrastructure to handle consumers.

After all, the "C" in CRM stands for "customer" and these are just two types of customers, right? The both have names, addresses, emails, true?

Yes, but that is just scratching the surface. These two segments are not exactly the same. Software and database vendors who specialize in one or the other are likely to advocate this supposed equivalence when they are trying to sell themselves from one customer vertical to another. Be cautious about this.

There are different business processes, data availability, and campaign rules that apply to each of consumers and professionals. Here is a sample:

For professionals, there are sets of affiliations that we care about, such as group practice memberships, hospital affiliations, medical schools. Multiple addresses result. Multiple specialties are important to note. Usually, professionals are called on by a sales force, whose activities on each call, details and samples, must in turn be noted as transactions that are separate from the 'transactions' of prescriber-level Rx data. There is the firm distinction between marketing and medical contact management that must be preserved.

Consumers have special attributes as well: census, demographic, and psychographic data are particularly important. Also, due to HIPAA compliance, for most companies there is not usually an option to measure direct behaviors, so surrogates from web analytics are even more important. Consumers also change residence and email addresses quite frequently, making data cleansing especially vital. Consumers are the ones in conversion and adherence programs with financial components like coupons, vouchers, or copay cards, depending on the healthcare category.

These differences can demand distinct data structures, and different campaign business rule designs. Does a company need to pay twice for RM infrastructure for each customer vertical? Many companies in fact do. Even those that have tried to adapt one system to the other, the lesson learned is: this is not a quick migration, it takes a great deal of planning, implementation time, and specialization.


For these reasons I tend to advocate using "C" in CRM is "consumer" and the "P" in PRM is professional.

Hacker's dictionary of gone by 1990s; still useful for CRM

When I am crunching (HIPAA compliant) data, or searching though files, deep into a project, I sometimes recall the bizarre "nerd" language of my grad school days at the MIT Lab for Computer Science. Most of these terms can be found at the Hacker's Dictionary Website

These terms are still coming in handy as one analyzes large CRM or PRM data sets and merges many disjointed files:

My favorite, with interpretations.

*Hack: To program for an extended period of time, and really love the creation.

*Munge: To analyze large data sets, and "work your magic" to get insights.

PERL, AWK, and SED: great Unix command line programs I wish I still had, for munging large text files.

*Snarf: To borrow from a colleague a file or document (electronic or paper)

*Grok: To pore deeply over an analysis, a code listing, a book, or a report, and really understand it

Try these terms yourself the next time you are grokking a few thousand rows of anonymized survey responses, or munging a ten-fold larger website click stream data set.