What is Data Normalization?
Data normalization is a process of making your data less redundant by grouping similar values into one common value. For example, a country field could have these possible options for the United States – U.S., USA, US, United States of America. A query would filter all of these data points as unique entries, even though their content is really the same. Normalizing this field will provide you with only one value of United States.
Why should you normalize your data?
Data must always be usable – regardless of whether you are working in marketing or sales. If you use data to either segment your email list, sales territory or report on profitability, messy data will hurt your effectiveness. Normalizing data ensures that you have actionable segments. You can derive reliable analytics based on the segment and have the ability to split your sales territory by various fields/values.
How to determine if a field can be normalized:
When doing data cleanup, you may wonder if I field should be normalized or not. StrategicDB has established the following rules that will enable you to decide if the field should be normalized or not:
- Is this a field that can help you make better decisions or is it used by anyone?
- Is complete for at least 50% of records in that entity? Or do you wish for it to be complete?
- Can the field be grouped into different groups?
- Do the values change over time or is it consistent?
If you answered yes to all of the questions, you should normalize that field. What happens if you only answered yes to some of the questions?
- If the field is not used, is barely complete but you can group it into different parts and it is consistent. Even though the field can be normalized, the effort of your normalization exercise will be greater than its results.
- If the field is used, is important for decision-making and is mostly complete and consistent but cannot be grouped into different groups, it is probably impossible for you to normalize. For example, an open-ended question to the advantages of your software will give you many different choices that you may not be able to normalize. However, for certain surveys this data can be still normalized by using positive responses and categorizing problem areas, etc…
- If the data changes with every year or a quarter you will be constantly normalizing the data as no commonality can be found. For example, names of social media and other innovative technology changes faster than you can keep up therefore normalizing this data will be an ongoing time-consuming exercise with little benefit in the next quarter or year.
What fields should you normalize?
When trying to decide what fields should be normalized, you first need to ask yourself the following question – what fields can be standardized? What fields will help me find insight, be used in segmentation and help with operations?
What fields are commonly normalized for company/account data type:
When normalizing a company or account entities, companies would at the very minimum normalize the following fields:
- Billing country or Mailing Country: When normalizing a country field, companies typically use the ISO standard defined country codes. It is not uncommon to run into issues when dealing with disputed territories. When normalizing disputed territories and newly created countries (which are not found under ISO), you can choose to put a placeholder for the country or add a new entry to your database. You may encounter countries that are split such as Yugoslavia. You would then be able to use City or Phone # Area code to fill in the appropriate country.
- State or province: Some companies choose to use two digits codes for state or province while, others prefer the full name. Regardless of your choice it should follow the same standardization.
- Industry: It is an important field for segmentation and in some companies is used for sales territory assignment. In North America, companies typically use the NAICS (North American Industry Classification System) or SIC (Standard Industry Classification) when normalizing their industry. Depending on your data needs you may want to stick to the two-digit NAICS code. Other companies may want to create their own custom industry list. For example, if you are selling coffee beans you may have the following classification: independent coffee shops, large chains, retail, wholesale and specialty stores.
- Annual Revenue Range and/or Employee Range: Most companies elect to segment there marketing and sales teams based on the size of the company. Some companies choose to normalize both annual revenue range and employee range to indicate the company size while others collect one or the other. It is important to normalize the fields based on your data. For example if 90% of your account records are coming from companies with under 50 employees, it will make more sense to break down the ranges to “Under 10”, “10 to 24” and “25 to 49” and “50+”. However, if your company sales to both SMBs and fortune 500 companies your ranges would be much and probably would start with “Under $1 Million” or “Under $5 Million”.
What fields are commonly normalized for lead/contact type?
When normalizing a contact or lead entities. Companies would normalize the following fields:
- Titles: normalizing titles is one of the most important fields for any marketer. By normalizing titles, marketers can target specific job functions and job levels. However, that would depend on your data and your ability to store the data. Depending on your storage capabilities and the type of data that you have. You may choose to normalize your titles, into two fields: job function and job level. For example, if you’re selling primarily to marketers and your data consists of 95% of marketers. You may only need one field with the following job normalized titles: “CMO”, “VP Of Marketing”, “Director of Marketing”, “Marketing Manager”, “Marketing Specialist”, “Marketing – Other” and a generic “Other” or “Non Marketing” title. However, you may also choose to break that data into a more granular level. Your job normalization field would then look like something like this: Job Function: “Content”, “Operations”, “Social Media”, “Analytics”, “Demand Gen”, “Web”, “Email”, etc… Job Level would be: “Vice President”, “Director”, “Manager”, “Specialist”, “Other”.
- For leads: when normalizing leads you would also have the same fields as you would for account normalization plus titles.
You’ve normalized Your Data. What’s next?
Once you have normalized your fields, you should review where your fields are populated from. There is no point of normalizing, if you will continue to produce data that will need further clean up. Therefore, upload process, forms and system’s programs should be adjusted to normalize the data as it is coming in.
After data is fully normalized and is coming in to the system in the format that you want you are ready to revisit all your dashboards, segments, modelling and sales territories. For marketers, normalized data means that they can now deploy emails specific job function and level, they can use new fields for their lead scoring and have more accurate reporting based on the new segments. For sales, normalized sales data means that sales territory can be assigned by industry, company size or/and geography. Management can have a more detailed view into their ideal target market and profitable segments and can make better decisions.
If you need help with normalization, feel free to contact us.