Dragos Petria — 21 August 2019
9 min read

Ocean.io’s beef with industry codes

Ocean.io’s beef with industry codes

Reading through most of our website or digital communication will leave you with the impression that we’ve got a problem with industry codes. And that's not… not true. In our quest to provide the most comprehensive B2B data for sales and marketing organizations, we’ve dug up our proverbial hatchet and waged war on industry codes. This post is our attempt at explaining just what exactly our problem is.

The increasing importance (and difficulty) of customer segmentation

Customer segmentation isn’t a new concept. It emerged in the 1920s, as a solution to an increasingly complex and interconnected business environment.

Segmentation is the practice of dividing a (potential) customer base into groups of individuals or organizations that are similar in specific ways relevant to sales and marketing. Efficient segmentation relies on identifying key differentiators in demographic, geographic, psychographic, behavioral, and firmographic information.

At its core, it acts in two ways:

1. It makes sure that a company is trying to reach and sell its products or services to the segment of the market most likely to benefit from this activity;

2. It streamlines a business’ spending, making sure it doesn’t waste its limited budget on reaching people unlikely to convert to paying customers.

That being said, while B2C companies are having a fairly easy time using demographic, psychographic, and behavioral segmentation methods to reach the right customers, B2B companies have historically had fewer and less nuanced ways to segment prospects. In the 1980s, firmographics emerged in the B2B world as a response to the more sophisticated B2C segmentation.

Under the umbrella term “firmographics”, we have a number of different variables, most commonly company size, location, and industry.

The introduction of widespread internet has completely changed the business world and created massive upheavals in the way salespeople and marketers conduct business. It’s hard to overstate the constant overflow of advertising and information coming to our screens every day. Between ad blindness and an inbox full of spam emails, it’s never been as important to properly segment your outreach to ensure that your value proposition reaches a person that can see the value part of it.

Increasing complexity

Starting a business is as easy as it’s ever been. From digital registration of sole proprietorship companies to investors wanting to capitalize on the “next big thing” and the growth of the digital economy, there are opportunities everywhere. With this abundance of companies and a service economy that is becoming increasingly complex and data-driven, we’re witnessing a new level of sophistication when it comes to companies’ value propositions and it doesn’t seem to be slowing down.

The digital boom of the late 90s created (and largely destroyed) the first wave of big e-business. Since then, there have been successive layers of complexity, building on that groundwork.

As an example, let's take the advertising industry in a broad sense:

  • It moved from offline to online with the popularity of search engines, social media, and the increased access to the internet;
  • It then diversified into different types of online advertising: PPC (pay per click) vs PPM (pay per thousand impressions);
  • First-hand ad engines such as Google and Facebook took the market to a mature point;
  • Other advertising companies such as Taboola, Outbrain, and Disqus claimed their share of digital advertising through their own networks for websites and publications.

In terms of format, those adverts evolved from static text and banners to dynamic and automated text and banners, videos of varying formats, and influencer posting on social channels, product tagging, etc.

The targeting capabilities changed from simple demographic targeting to much more sophisticated approaches. From simply retargeting all of the users that have visited your website to targeting some of them across platforms and devices, Facebook and Google have gathered psychographic and behavioral data on their users that lets advertisers target based on generic interests shown through searching, browsing, and interacting with different websites. They can also trigger these ads when and where viewers are most likely to convert.

It has reached a point where ad platforms such as Facebook or LinkedIn are able to deconstruct a group of customers into their defining characteristics, and suggest what they call “lookalike audiences”: using their customer base, they identify other users with similar characteristics (to varying degrees), which is sometimes rather alarming but extremely effective.

A quick glance at other “industries” will quickly tell us that this level of increased complexity and integration of technology is true across the board: the financial sector, with the development of blockchain technology and cryptocurrencies; e-commerce; the Internet of Things; web development agencies; search engine optimization; the affiliate industry as a whole; or CRM systems.

Going back to the three key variables for firmographic segmentation, we can start seeing some issues:

  • Company size is becoming harder to determine. Different businesses need to use different metrics for sizing: number of employees, web traffic, revenue, etc;
  • Location is becoming less relevant, with the decreasing costs of shipping and logistics to the fact that digital businesses are often not limited by physical location, but by language and cultural barriers instead;
  • Industry is probably the biggest issue. Traditionally, this has been determined by secondary data sources, such as government agencies or data vendors. The most universal way to determine industry has been through industry codes. Every business needs to specify its primary activity in the form of a code corresponding to the type of activity it intends to perform.

Issues with industry code classification

Building on the complex global economy that we now operate in, there are several issues that render using industry codes as a proxy for customer segmentation ineffective at best and outright misleading at worst. This is the problem that firmographic segmentation was intended to solve – effective targeting and streamlining your spending on the right customers.

Industry codes are not accurate anymore

As we’ve mentioned before, the digital revolution is dramatically changing businesses. New niches come and go every day and new technologies mean businesses have different value propositions in the same “industry”.

A good example is the cryptocurrency exchange company, Crypto Facilities: it is listed under the NAICS code 523, which is defined as “Securities, Commodity Contracts, and Other Financial Investments and Related Activities”. This boxes Crypto Facilities together with Credit Suisse, Fidelity Investments, etc. While they do have a degree of commonality in that they operate within finance, their services are drastically different.

Simply put, the business world is so complex now that, despite a government's best efforts, it’s impossible to keep track of and generate new industry codes for all of the new markets being created.

Codes are not actualized

The actualization issue is two-fold. We live in an entrepreneurial culture where change is ubiquitous – start-ups are taught to ideate, test, and pivot as often as it is needed.

While a company might change direction and fundamentally change what it is doing by finding new and exciting applications for its offering, they will rarely go back and change their industry classifications accordingly.

The second part of this issue is that when a government does pick up on new markets and decides to create an industry code, this is generally done because a specific niche has reached a certain size. All companies incorporated before the industry code was created will have outdated classifications, of the “other” variety.

Industry codes are not universal

Industry code classifications have been developed by each country independently throughout the last century. A direct consequence of this is that countries will have different numbers of industry codes, as well as different classifications for the same industry branches as they are relevant for the country. There is no way to universally translate one country’s industry code to others.

Another good example is Microsoft. It’s a giant, well-established global corporation so we’d expect boxing it in an industry code in different countries to be quite simple, right? Well, obviously not, since I’m typing this out...

Looking at Microsoft’s NAICS classification in the US, they’re defined as “All Other Support Services” – very descriptive, isn’t it?

On the other hand, when we look at the Danish registration for Microsoft, it’s NACE code, 461800, roughly translates to “Agency trading with specialized product range”.

Once again, we’re talking about Microsoft. We all know (most of) what they do. If someone asks you what Microsoft is selling, you’d say “software” in a heartbeat, and maybe dive a bit into what kind of software: operating systems, productivity apps, etc.

If industry codes can’t get Microsoft right, how can we expect them to be reliable when we’re talking about the myriad of markets that are developing every single year?

Not all countries have centralized directories

Even if industry codes would be enough to properly segment companies for acquisition efforts, there’s one last bottleneck.

This last bottleneck is that not all countries have a well-structured, centralized and updated company directory that we can rely on in terms of data gathering. In these countries, you usually have third-party solutions for firmographic data but relying on an unstructured source of incomplete data is a risk for companies that take their data seriously, as we do. You simply can’t rely on bad data. In the digital world, you are only as good as your data.

Move over industry codes, big data is here

Roughly a year ago, the team behind Ocean.io started building a platform to help salespeople and marketeers navigate the increasingly murky waters of company data in order to empower their customer acquisition strategies.

It didn’t take long for us to understand that, to empower our users to drill down into our huge database of companies and reach a segment of companies that would meet their criteria well enough for our platform to be relevant, we couldn’t rely on industry codes as a common denominator.

As a solution, we’ve started building a search engine for B2B data, a platform where you can use your own input to find companies that fit your segmentation criteria, no matter which market you operate in, and identify ways to reach this segment of companies, no matter what kind of acquisition organization you’re operating with.

A lot of thinking and back-and-forth later, we’ve decided that a company’s own website is a great starting point for several reasons.

First, there are very few companies nowadays that don’t have a website meant to attract customers and partners alike. Websites are largely the most accurate middle point between a company’s internal organization and the business world at large. Companies have a need to properly explain what they do, who they are, what they’re selling – it’s in their best interest to carefully craft their message to the world and to update it as often as needed.

As such, there is a lot of valuable information on each company website. You can understand a lot about a company by reading the homepage, solutions, products, pricing, about, and contact pages. If you can understand the way a company presents its value proposition, who they are as an organization, and who their products or solutions are built for, you have a good starting point for profiling and boxing the company in a segment.

Pretty clever, right?

Clever or not, it’s not a trivial task. Luckily, we have a big team of very talented data scientists and engineers that love a good challenge.

Using big data that we’ve been gathering and structuring for years and a lot of thinking, endless computing, ideating, tweaking, machine learning modelling and testing, we’ve built our proprietary technology that lets us identify the similarities between companies by using a lot of different data points.

Once we built this engine, we developed our Beta platform on top of the search engine and added plenty of ways to filter and segment to make sure that any company, in any vertical, can use Ocean.io to identify the perfect segment(s) for their inbound or outbound efforts.

In a nutshell, any user of our platform can tell Ocean “this is my ideal customer” and, in a matter of minutes, can reach a segment of companies that (almost) perfectly fit their value proposition, regardless of country or language barriers, then identify contact information and export these to their CRM system or a .csv file, and start selling or marketing to this group of companies.

You can start a search using the company website for an existing or ideal customer of yours, and we generate a list of similar companies in order of how similar they are. You can then filter and segment using keywords, technologies that they use, social media presence, web traffic, as well as location and language. We give you plenty of information to go on, from employees and their contact information for your sales organization to their brand names and websites for your marketing organization to build LinkedIn and Google Ads campaigns.

You can also start a search with keywords and technographics, if the market that your ideal customers operate in is irrelevant.

You couldn’t do that with standard industry codes. They’re too broad, too inconsistent, and too unreliable.

We’re never going to be satisfied with poor quality data, or even fairly good data. We want the best data possible, at all times – that’s at the heart of our platform and our company. And that is the source of our beef with industry codes.

The road ahead

Our job is not done and we’re aware of it. We’re working hard to build on top of our current algorithms. We’re gathering more data and are constantly finding ways to improve our machine learning algorithms to keep up with the increasing difficulty of effective B2B customer acquisition through both sales and marketing.

We have a pipeline full of new features and filters that we are developing and implementing together with our customers and partners. Since our official platform launch in late 2018, we’ve been taking in all of the feedback that our users share with us to make sure that they get the most bang out of their buck when using Ocean.io

We’re also planning to develop some complementary solutions to give our users the ability to use their own customer data alongside with Ocean.io data to push the boundaries of customer acquisition in the digital world.

Our biggest strength is our data, and we have a commitment to improving it and finding new ways to generate value.

Stay tuned!

In the mood for sharing?