Question:Would you buy a 50.000$ car, if you can’t trust the quality of the petrol you fill it up with?
Would you implement a high-end marketing automation tool if you can’t trust the data you put into it?
Consider these facts:
- 88% of all data integration projects run over budget or fails due to poor quality data
- 75% of global businesses recognizes that data quality issues cause them to directly lose money through missed business opportunities
- The cost of poor data may represent 10-25% of total revenues.
While you have no control over the quality of Petrol, you CAN control you data and increase the data quality and build trust.
Implementing Marketing Automation
When implementing a Marketing Automation solution, like ExactTarget, Marketo, Eloqua or HubSpot the main objectives are often to provide each lead/customer with an individual/custom interaction process, sometimes called a: customer journey.
1-to-1 communication will provide the lead/client with the impression of having communication which is Relevant, Conversational and Engaging.
But if your Marketing Automation solution shall replace the attentive sales rep. the data being available for the automated work-flows to work, needs to be trustworthy.
Separating Trustworthy data from Poor Data?
Recently I attended a salesforce conference and the Managing Director from T-Systems (a German global IT services and consulting company headquartered in Frankfurt) was claiming in front of a large German crowd that the T in T-systems was for Trust, he received a big friendly laugh. But T-Systems are trustworthy and why, because history have shown that they are reliable, the cover a complete range of services and expertise and the service delivery is consistent over the years, and their clients know that, so they return.
Trust is something you build, not something you buy, so you need a blueprint of your data, and you need to monitor and show how this blueprint improves…you need metrics.
In an ideal world your data should be Reliable, Complete and Consistent, but reality often shows that it is not.
On Average: Only 33% of lead data is target-able due to lack of combinational segmentation data.
Identify your Data Quality Metrics
What to measure varies from business to business, but have you thought about this:
- How well your records are populated on critical segmentation data, geo-location, job-role, functional area, industry, and company size (revenue or employees)?
- And whether the values of these fields are consistent (standardized and normalized).
- Completeness of contact data fields like name, address, email, phone etc.
- Completeness of historical behaviour like previous product interest, last purchase etc.
- How many of your record are truly mail-able, phone-able, email-able?
- How old your records are?
- How big a portion of your lead data is actively being used in campaigning, and why?
- The Volume of new records being added by source/origin on a daily/weekly/monthly basis
- The number of potential duplicate records? By Owner/ by source?
In other words poor data can be expressed in various ways, but it all comes down to whether the data suits and supports the processes which ‘consumes’ the data. Think of it as a way to label your data records, in terms of describing what it can be used for (target-group) and what it is missing to become a target.
Categorize your data based on what it can be used for
Not all your data is complete, and not all sources provides reliable data, but you only have limited resources to improve the data quality and thus the effectiveness of your campaigns, so being able to choose wisely which records to use for which campaigns is essential to communicate in an engaging way.
Example: Typically when data is being migrated from various systems into one common database like salesforce, the data served specific objectives in each of the original data sources, but now when being brought together the data needs to be consolidated and not just serve one purpose but many purposes at a time. Identifying these purposes and measuring to which extend each records fulfil the purpose will help you understand your data better and where to invest your data cleaning efforts.
Let’s assume that you bring together 1) a (e)mailing list from a news letter which you send out regularly, 2) a list of clients which have made purchases with you in the last 3 years, and 3) a list of people who have registered themselves on our website (for a download, a pricing enquiry, or a webinar).
The challenge here is that your (e)mail list only contains the email address, and not much details on the person, your web-leads will most likely contain incomplete or inaccurate data since users often are hesitant to give away too much information on a web-site, and only your client data is complete, but since the client data mainly has been used for invoicing purposes you may have the address of the HQ, but not the address of the location where the person works.
So although you will need to match and merge the duplicate entries in order to link the data from the different sources together, you first need to consider standardizing and harmonizing the data to improve the matching rate. Then once you have the target-groups identified you can nurture each target-group (1-7) individually and make sure you spend your effort on those records which represents most business value (up-sell/cross-sell/multi-touch leads).
Get Rid of Garbage
Garbage in – Garbage out
What you may even realize that a large portion of your data is (if not garbage) then at least it is a shape where it isn’t useful or valuable. It is important to identify this part of your dataset and either scrap it or put it aside for when you have the time to improve the data quality on these through enrichment or other application of data cleaning treatments.
Including this type of data in your campaigns, will not only lower the response rate on your campaigns, but you risk including leads which definitely are not a target, and you risk ruining the relationship by e.g. sending promotion sign-up offers to existing clients.
Building Trust through Transparency?
For building trust in data you need transparency, and for transparency you need metrics as discusses above, and most important you need a good track history of good metrics.
Once you have your metrics identified and you start collecting them through dashboards or other reports: share them with your users!
“It is not a shame to show that the situation isn’t perfect; it is a shame not to show what you do about it.”
So like salesforce who publicly shows the system status for all their servers (http://www.trust.salesforce.com/), apply a concept of building trust through transparency.
Think of the salesforce serves as representing different subsets of your data, and run regular scans of your data to reclassify your records, so that you can show progress over time, and if you want to go deeper making drill-downs: For each subset run consistency check on critical fields like, country, state, postcode, city, phone, email, website etc, and do not forget the pick-list fields, as salesforce allows pick-list fields to be populated with values different from the available list of values when data is imported through the API or using other type of integration tools.
Salesforce has provided a basic Data Quality Analysis Dashboard which may be a good start or at least a source of inspiration: https://appexchange.salesforce.com/listingDetail?listingId=a0N300000016cshEAA
Or get a bit more out of it from this article: Making the Salesforce.com Data Quality Analysis Dashboards Work for You
Be pragmatic, it will never be perfect.
Consider the quick wins, where you get most value for the effort you put in. Working with high volumes is always time consuming and costly, so take time to invest your time on the data which represents the most value/opportunity for you.
So let your analysis and your metrics focus on separating good data from poor data, and based on this you can start identifying and prioritizing the treatments you want to apply to make the poor data good and the good data better.
DataTrim improve the reliability completeness and consistency by applying a set of data cleaning treatments which is called The Data Laundry.
The Data Laundry solutions and services adds experience based data cleaning processes to lead management, marketing automation, customer support and account management processes in salesforce and created direct impact on the day-to-day usage and productivity in a simple-to-use, collaborative and cost effective way.