Tuesday, February 18, 2020

Sick of SIC Codes?

Classifying your customers or prospects based on industry segments should be much easier than it actually is. It’s actually surprisingly hard. So hard, that Gartner stopped any market sizing and forecasting efforts based on verticals back in 2017. Chances are that if you measure your company performance based on verticals, there are only very few people in your company who can provide such data. And chances are that data is highly inaccurate.

Charles Darwin would struggle too...
There have been two main efforts to standardize the taxonomy of industry verticals. The SIC is the older, international standard albeit originally developed by the US government. NAICS is much a more recent taxonomy from the US Census Bureau. SIC, which was last updated in 1987 is so old and outdated that it should no longer be used but it lives on as many of the companies with headquarters outside the US don’t have an NAICS code and only are classified with a SIC. Also, SIC is in particular used in the UK and so if your EMEA headquarters are in Britain, chances are they will be using SIC codes just to keep their US marketing colleagues out of their turf.

The NAICS has been most recently updated in 2017 and is much more up-to-date than SIC but don’t hold your breath, it’s not that much better. In fact, while I called out SIC in my [hopefully] catchy title of this blog post, the challenges are the same with NAICS codes.

Both are using multi-digit code to specify the industry. For example, NAICS is typically used with 2 to 6 digits. The more digits you use, the more precisely you are diving into the taxonomy. You can select to filter companies based on their, say, 3-digit NAICS and you will get a list of companies with a specific level of granularity. If you want to be more or less granular, you add or remove a digit. That is immensely useful.

The problem with the standardized industry code is not so much the quality of the standardized taxonomy, although, try to classify technology companies using SIC - most common technology categories hadn’t been invented yet in 1987! The problem is the classification ambiguity.

At first, it sounds simple. After all, Boeing builds planes, Bayer makes drugs, and the Department of Energy is part of the government. But when you start looking closer, you realize that sometimes, Boeing is classified as a manufacturing company, Bayer as a laboratory equipment manufacturer, and the DoE under power & utilities. And those classifications are correct. In fact, Boeing might be also classified as a transportation company or aerospace & defense company and none of those classifications are wrong.

Part of the challenge comes from the source of the SIC and NAICS codes. There is no central register of SIC and NAICS codes. That means that every organization or agency that uses them is in charge of their own assignment. Companies might also self-select their NAICS code when filing various government reports (i.e. census, taxes, etc) but that often leads to different NAICS code selection for different purposes.

Another source of NAICS codes are the data companies such as Dun & Bradstreet that assign NAICS codes on their own. That is another source of ambiguity, especially when you start combining lists from different providers. 

On top of that, companies evolve. As they add new product lines or services to complement their products, the original NAICS code might no longer fit. Sometimes, they keep the old code but often, they end up with a highly ambiguous code such as “Business Services” which might be technically accurate but doesn’t really tell anyone what they do.

So, how do you solve the NAICS and SIC code puzzle? How can we get accurate and useful classification for your particular organization? The short answer is that there is no magical shortcut. If the NAICS codes that come with your data list don’t work, you’ll have to assign them yourself. Yes, this is a manual process and it’s impossible to do for your database of 100,000 suspects. But it might be possible for your database of 500 or even 5,000 customers and that’s where you should start.

To do that, you want to first build your own taxonomy. Because you’ll have to manually classify your customers, you can’t have a taxonomy with hundreds or even dozens of possible entries. There are over 1,000 6-digit NAICS codes, 709 5-digit NAICS codes, and 311 4-digit NAICS codes, Even if you limit yourself to just 3 digits, there are still 99 NAICS codes to assign. That’s too many! There are only 20 sectors using just 2 digits which is a good, humanly consumable number but you’ll find out that that won’t give you nearly enough granularity.

What you need is your own taxonomy - a taxonomy that is based on NAICS but stays very high level in some of the branches and perhaps ignores some completely. Maybe you don’t do any business in the agricultural sector at all? Or perhaps the public administration sector is not relevant? Leave them out.

At the same time, you need to go deeper than 3 digits in some of the more relevant areas. Manufacturing is a good example. If Medical Equipment Manufacturing is a relevant sector for you, you have to go to 4-digit NAICS code (3391). Obviously, you can’t add every 4-digit code, but you can pick the ones relevant to your business.

That means that you end up with a taxonomy that uses the correct NAICS codes but goes deeper in some areas while it remains shallow in others. It all depends on what’s relevant to your business. Now, that taxonomy should have no more than 20-25 entries. Anything more becomes humanly impossible to keep track of unless you have dedicated people who do nothing but this. If you are tracking more than 25 verticals, you need to rethink your vertical strategy - so why work with a taxonomy that contains hundreds of them?

Next, you need to establish some rules on how you will classify some of the ambiguous cases like the ones I mentioned above (Boeing, Bayer, etc). You have to decide how you will classify those companies based on what serves your business. Here are a few more examples:
  • LSG Sky Chefs is a company that provides food catering services to airlines. You decide whether they are a Food Services (722) company or an Air Transportation company (481). Think about the impact on your business. Would you find them at a gastronomical conference or at an aviation expo? Would they be handled better by the sales rep who’s focused on aviation or by a rep specialized on the food industry?
  • Kiwa is a company providing inspection services across verticals and you need to decide whether you classify them as Pharmaceutical and Medicine Manufacturing (3254), Oil and Gas Extraction (2111), or perhaps you create a category called Testing and Inspection Services that doesn’t have a NAICS code. What you probably don’t want is to assign them the code for All Other Professional, Scientific, and Technical Services (541990) which is where they usually are classified by default. 

 As you can see, the ambiguity can be pretty high and overwhelming, but the good news is that you can get started and create (and document) your rules as you come across the ambiguous cases.

Now, it’s time to think about the business process. Obviously, it is unreasonable to go through thousands of accounts and classify them all at once. That’s a lot of work. Instead, you should think long term. What’s the process that touches all customers in the course of a year? Perhaps it’s the contract renewal process. If that’s your best bet, then the team handling the renewals will be the team that assigns them their vertical the next time their contract is up.  

Another possibility might be to let customers select their vertical from your custom taxonomy - perhaps as part of a customer survey or when they register for a webinar on your website. Or, you can make the field mandatory for inside sales to fill out when they accept the opportunity. You will find ways to systemically apply your classification to make it feasible and consistent.

A vertical strategy is extremely important for most software companies. As you mature from selling features to selling solutions, you will want to start infusing more and more vertical language into your messaging. But none of that is very useful if you can’t target your prospects and customers based on their vertical. And that’s where the NAICS and SIC codes are very useful, but only if you make them work for you.

Happy marketing!