What Is Marketing Segmentation? (And How To Do It Right)

Leaky funnels.

A constant struggle for every marketer.

You spend a ton of resources getting visitors, getting their email address and marketing to them. And yet many drop off.

The problem is most marketers send broadcast the same messages to everyone.

They don’t try to tweak or personalize messages for different groups of people.

… but you never behave like that in-person, in real life. It’s not human to broadcast like a robot.

A good sales person will ask questions to find out what matters most to who they’re talking to, then tailor what they say to those answers.

The same is true for your marketing messages, even if they’re automated and sent to thousands of people. Find out what matters most, then tailor what you say.

This is where segmentation can save the day!

What is marketing segmentation?

Segmentation is more than just market analysis. The real advantage of Segmentation is about to dividing up people to receive different, more relevant messages - not just the same "send-to-all" spam.

Marketers use segmentation for email marketing, ad targeting, personalizing websites, managing live chat and more.

Sales use segmentation for prospecting, qualification and bucketing leads in their pipeline.

Customer success and support use segmentation for tiering customers, prioritising tickets and managing upsell, cross-sell, and renewals.

Segmentation is important to personalization too.

Whether it’s email list segmentation or creating ad audiences - segmentation is central to solving the leaky funnel. The advantage of segmentation is you can be more relevant to more more people, so you can convert and close more deals.

By sending each person a more relevant message, you can speak to their specific wants, needs, concerns, and desires.

... instead of the generic, watered-down “what everyone wants” broadcasts. Blegh.

Definition: Marketing segmentation is the process of dividing up a set of customer (or potential customers) into groups, called segments, based on different characteristics

In this segmentation strategy guide, you’ll learn our complete segmentation model at Hull:

  • Best practices for choosing what segments to create
  • What size of segments you shouldn’t build
  • What types of data and segmentation variables are useful (or not)
  • The advanced segmentation analysis techniques to building conversion-maximizing segments
  • The seven techniques for syncing segments across all your sales and marketing tools
  • Tips and tactics for researching effective messages for each segment

Pssst! Grab a one-page cheatsheet of all our segmentation secrets.

Grab your segmentation cheatsheet

First, describe your ideal customer profile.

And this can’t be any description…

Remember, we’re trying to leverage customer data to scale and automate our messaging.

Our ideal customer profile must be able to help us clearly define what data we need to target our ICP. Follow our guide on creating an ideal customer profile process for data-driven customer operations, making sure your ideal customers have these three elements:

  1. Use nouns and verbs. Avoid adjectives. Nouns (things like names, names, job titles) and verbs (actions like clicks, replies, opens, purchases) make it much easier to tie to data later.
  2. Be specific. Dig into your sales and support teams to find the details which mark out great fit customers.
  3. Be quantifiable. Wherever you can, using quantifiable measures. This makes it much easier to classify and group contacts and accounts later.

The reason we want nouns, verbs, specificity and quantifiable ideal customer profiles is we can tie them to data!

Your marketing is only as good as your data.

… and your personalization is only as good as your data. Personalization is powered by your data, which is defined by your ideal customer profiles.

Without effective data, it’s hard to do the analysis and rule-building to create and maintain your segments.

Once you’ve got your ideal customer profile, you must source the data on each lead and customer. This is called customer profiling.

For instance, at Hull we’re interested in each leads firmographics and technographics - data about companies and the technologies they use.

One more set of data you need to consider - your customer lifecycle. What are the stages before becoming a customer (and the stages afterwards too), and what defines the movement between stages.

You need to source and shop for all this data.

Where you don’t have data, you should buy it. “Data enrichment” services like Clearbit and Datanyze aggregate and deliver business intelligence data like firmographics, technographics and other information that’s central to your lead qualification, sales conversations, segmentation and more.

Without this enriched data, you’re baking inefficiency into your sales operation. You’re forcing sales to do manual work or “discovery calls” to get the basic questions answered. Slooow, boring and a poor use of everyone’s time.

By buying in data, your sales teams can spend more time with better fit leads, saying the right things to close them.

Like Lengow, who fixed their 50% drop off in responses to demo requests with automated data enrichment and deep lead qualification.

From your list of data you need, figure out what you can source yourself. Buy the remaining data.

Customer segmentation analysis = what segments should you build?

Now you’ve got the customer data, how do you slice and segment your email list, your ad targeting, your leads and everyone? One multi-channel, multi-segment targeting strategy.

Remember, your personalization is only as good as your data. Reliable, up-to-date and accurate data. Since we use data to define our segments in our tools, your segmentation strategy needs to be created in terms of data.

And not just any kind of data or segmentation variable. There’s a trick to it...

Think in “dimensions”

If you’re familiar with Google Analytics, you’ll be familiar with dimensions already.

Dimensions are attributes of your data. For example, the dimension City indicates the city, for example, "Paris" or "New York", from which a session originates. The dimension Page indicates the URL of a page that is viewed.

Instead of thinking of random segmentation variables (which leads to a chaotic set of segments), think of similar data types we can segment people and messaging by.

Data often lies in dimensions. Just like with dimensions in geometry (a line to a square to a cube etc.), dimensions give us a way of describing a set of attributes.

In B2B marketing, sets of common dimensions might include:

  • Job title
  • Country
  • Industry
  • Number of employees
  • Revenue

Dimensions can be used to describe both individual contacts and company accounts.

You can use dimensions on top of each other to build ever more refined, more targeted segments. Take a look at these three segmentation examples with more dimensions added.

  • Job title (CEOs)
  • Job title + city (CEOs in New York)
  • Job title + city + industry (Healthcare CEOs in New York).

Building segments according to dimension-based rules like this (instead of random sets of segmentation variables) helps keep segments manageable, consistent and keeps naming conventions clear. More on that later…

Your segments should be big enough to be useful

The idea with segmentation is to group common contacts and companies together so you can send a more relevant message to them.

But your segments should still be worth the extra effort to create custom messages for them.

The rule of thumb: Don't build segments smaller than 500 people. The extra effort isn't worth it.

Let's run the numbers quickly - if we have an email list’s goal is only to generate clicks. An average click rate of 1%, where 1 in 100 who we send the email to click. Say we can improve that to 5% within a segmented email (This is not atypical. Once you know who you’re writing to, you can tailor your message to be far more personal. I averaged about 10% click rate on hundreds of thousands of “personal emails” in my last role).

Imagine we have a list of 2000 people, and we could create two targeted segments of 500 people each, with a generic email to everyone else. If we can drive 5% click rate with the two targeted lists, we can drive better results.

Before, we’d expect only 20 clicks (1% of 2000). Now, we’d expect 25 clicks from each of the targeted segments (5% of 500 = 25, for both lists) for 50 clicks (25 + 25) in total, and 10 clicks (1% of of 1000) from the remainder. 25 + 25 + 10 = 60 clicks in total, instead of 20. That’s triple the result for triple the work (3x the number of segments and messages).

Any worse than this and it doesn't add up to a meaningful result. You'd be better off and generate more clicks by creating a new, separate campaign to send to the whole list again.

A segment smaller than 500 probably isn’t worth building and maintaining. You’ll see similar problems for targeting ads, website personalization, live chat and so on.

If your messages are triggered by some automated workflow (not sent all at once), the same rule applies. Is the segment that will trigger the workflow and receive the message over a period (like a few months) bigger than 500 people?

Similar to making segments that are big enough, your segments have got to make sense as a way of grouping people together.

Your dimensions must have groupable data!

A segment of CEOs makes sense as a segment. A segment of people called David does not, because you can’t group people into a manageable number of segments (think how many possible first names there are!) You’ll end up with so many segments to build and create messages for.

Don't forget you can use dynamic content to personalize the content within a message. This enables you to still tailor your message without having to invest in creating and managing separate segments and separate messages.

Remember, whenever you’re working on personalization - segmentation first, dynamic content second. Decide who you’re targeting and talking to before deciding what to say.

Where you’re trying to refine your messages and send them to a more precise audience, it’s better to build bigger segments to group people together by something bigger (like lifecycle stage) and then use dynamic content with templates to personalize the content within it according to {firstName}, {companyName} and your other dimensions.

Lots of conversions already? Choose dimensions with the biggest “spread”

When you’re starting out and are choosing which dimensions to segment by, it can be easier to focus on what’s going to most easily create a more effective message.

Perhaps it’s by pricing. Or customer persona. Or job title.

But, as you grow and have more “real data” and real conversions, you segmentation strategy should mature too. As should your segmentation analysis!

The best segmentation strategy will slice up your user base by the biggest spreads in conversions. One segmentation strategy might have a bigger difference in conversion rate over another. This indicates one segmentation dimension is more influential in the conversion than another.

Job titles may be more important than cities.

Industry may be more important than number of employees.

Language may be more important than country.

Who knows? You need to dive into your customer data to find what matters most.

It’s not about what has the highest conversion as a group. The key idea here is to find where you’ve a bigger variation between different groups you can target.

That means the current messaging is resonating a lot more with one group over another group.

These are the dimensions that will create the best segments. It means you can tweak your messaging to make your best performing groups convert even better and create a whole new message to help everyone else catch up (or exclude them all together).

Hull - Spread in conversions

This takes a little bit of data and segmentation analysis to work out and compare different options. What we want to do is to compare how wide the conversion rates are amongst different dimensions.

  1. Pull a list of contacts for a period before and after the conversion you want to segment and improve. Make sure you include all the segmentation dimensions in the data (e.g. Trial accounts to paid accounts. Pull the complete list of trial accounts in a period).
  2. Calculate the conversion rate of each value within a segmentation dimension (e.g. New York: 15%, London: 22%, Paris: 19%; CEOs: 8%, Sales: 32%, Marketing: 18%)
  3. Calculate the standard deviation of the conversion rates by segmentation dimension (e.g. City: 3%, Job Role: 11%)
  4. Evaluate which segmentation dimension has the highest standard deviation in conversion rate (City: 3% < 11%: Job Role). This is your primary segmentation dimension.

If you have a significant number of conversions already and an established segmentation strategy you’re looking to improve, grab your template, personalized to your organization here.

In this segmentation example, job role is the most significant cause of variation compared to country and industry.

Dimension Standard Deviation (in Conversion Rate)
Job Role 22%
Country 13%
Industry 15%

From the data, it’s clear we can gain the most uplift with a segmentation strategy based on job role, since it has the widest spread in conversions. We can make the best performing job roles convert even better, and adapt our messaging for other job roles to help them catch up.

Create naming conventions for your segments

Once you’ve identified the segments to build, you need to create names for your segments too.

The neat freak in every team will love creating consistent names in their segmentation model. But there’s another good reason to care about naming conventions.

You want your segmentation names to be clear. Anyone should be able to read them and quickly understand who it’s targeting. We want to be able to share segments across different tools and teams, and for it to make sense in each tool and team.

It should be brain-dumb obvious:

  • What lifecycle stage is it for?
  • How is it being grouped together?
  • Will there be similar segments?

Of course, this is a matter of team preference.

If you’re looking for an idea, my preference works like this. Since English reads from left to right, I like to include the high level information (like the lifecycle stage) on the left, with the segmentation dimensions in order of priority. This makes it easily readable. I prefer to keep verbs and actions on the right too as they tend to be variations of “core” segments.

Here's a set of common B2B segmentation examples following this naming convention:

  • {lifecycleStage}: {primaryDimension} - {secondaryDimension}
  • Lead: {jobRole} - {country}
  • Lead: Sales - US
  • Lead: Sales - France
  • Lead: Marketing - US
  • Lead: Marketing - France
  • Customer: {accountPlan} - {jobRole} - {activityLevel}
  • Customer: Enterprise - Sales - Fading Away
  • Customer: Starter - Marketing - Active

Once you know what segments you want to build and how you'll name them, it's time to build!

How do you build your segments?

Building effective segments is hard. Keeping segments up-to-date can be even harder.

Once you know who you want to target, actually targeting them can be hard to do “out-of-the-box” with your customer segmentation tools and data.

The data you need may be scattered and siloed across many different tools and databases, because your data is tracked, stored and used in different places.

You need to get all your tools in sync with each other. The best teams build an effective customer data management strategy to break down the silos of data, and make sure each tool within each team has complete, up-to-date, reliable data to build the segments you need to target people.

Most companies use these seven customer data integration methods to keep profiles and segments in sync between all their tools:

  1. Basic one-click integrations
  2. Importing and exporting regularly (often via some spreadsheet)
  3. Workflow tools (like Zapier)
  4. All-in-one tools
  5. Custom coded integrations
  6. Data warehousing (and custom code or manual imports to get data into your tools)
  7. Customer data platforms like Hull.

The method doesn’t matter to your segmentation strategy. But what really does...

Build one set of segments, then share them everywhere.

Just like with your tracking plan - don’t create a mess of many different systems and segments that stay siloed within each tool (and therefore, within just one team). Fragmented data, including segments, divides teams and fuels politics.

Segments often get created within core marketing tools like marketing automation, email, ad platforms and analytics tools. Whatever method you use to sync data between these tools, it is vital that you’re able to share reliable, up-to-date segments and data between tools too.

Your ad targeting might be full of great, useful insights and ideas that are converting better.

Why keep that from your email marketers?

Or your sales development team?

Why miss out on opportunities you’ve already found to work?

In the same way you want to share one set of data around each customer’s profile, the goal is one set of segments shared everywhere. This puts your whole team on the same page, all able to leverage the same insights, and all getting uplift from your segmentation analysis and strategy.

How do you use those segments?

Use segments to send different messages to different groups of people.

This is the chance to tweak, tailor and cut your generic one-size-fits-all messaging and focus each message on the most relevant thing.

As Guillaume Cabane at Segment says,

I like think about shoe size in the US. The median male shoe size median is 11, right? That’s only 20% of the population. That means that if you had to find the “one size fits all” and you had that on your website, you would only show the shoe size 11 and all the others would leave.

You cannot drive uplift from segmentation without different messaging too.

Else you're dividing people for no reason. This is like asking everyone for their shoe size, then giving them all size 11 shoe. It does not make sense!

This is your chance to become much more personal, {firstName}.

Wait, what?

Personalization is NOT "Hi {firstName}!"

If I yell your name across a room, sure I might get your attention. But I haven't connected with you in any meaningful way...

What is the most meaningful way to segment your messages?

This is where your dimensions come in. Let your dimensions define your messaging.

If you're segmenting by job title, then your messaging should reflect the different roles and responsibilities.

If you're segmenting by industry, then your messaging should reflect the industry specific challenges, language and case studies.

If you're segmenting by city or country, then talk about what's most relevant to their location.

These all have something in common. They all tie people’s identities. Remember, the goal of segmentation is to send a more relevant message. Those should be messages which each individual will closely identify with.

Your segmentation dimensions will let you zero in on who they really are. This is how you can leapfrog other messages and competitors and craft a truly meaningful, personal message.

But how do you make many meaningful messages? (Try saying that ten times…)

As with any message, you want it to be grounded in what matters most to who is receiving it. Don't jump to conclusions. Don't stereotype. Don’t skip the research you’d do with any other message.

This. Is. Hard. Work.

Remember, you’ve got to create as many unique messages as segments you have.

This is why we said earlier that doesn’t make sense to create tiny segments (less than 500 people). You create too much work for yourself for too little gain.

But, if you do your segmentation right, you’ll have all the clues to start your research with.

You’re no longer writing to everyone and trying to make something that works for your entire list. You know exactly who you’re writing to - it’s much easier.

Go back to your segmentation dimensions. Use that to as the source of interviews, data analysis and research into what message matters most.

If you’ve done the segmentation analysis around the variation or “spread” of your average conversion rate, you’ll clearly see where different messages DO and DON’T resonate with different segments.

With each segment for each lifecycle stage (or goal), make a note of what their specific problems and pain points are, and the messaging that you want to test to resonate most with that segment.

In our own segmentation example, onboarding emails for Hull might vary by job title:

Job Title Problem Messaging
Marketers How could I use this? Use cases
Sales How can I set this up with my CRM? Salesforce setup guide
Engineers How do I set this up? API documentation

Let’s Recap! How to create a segmentation strategy.

Create an effective ideal customer profile, and source the data to build this. Use this data to define your segmentation dimensions (not random variables). Create a common set of segments based on the widest spread in conversion rate, then sync them across all your tools and channels with a clear naming convention.

Crucially, make sure you write unique messages to each segment - this is your opportunity to personalize your sales and marketing.

Want a handy one-page cheatsheet of all our segmentation secrets? Grab it here.

Grab your segmentation cheatsheet

Ed Fry

Prev 'Ed of Growth at Hull, working on all things content, acquisition & conversion. Conference speaker, flight hacker, prev. employee #1 at inbound.org (acq. HubSpot). Now at Behind The Growth

If you've questions or ideas, I'd love to geek out together on Twitter or LinkedIn. 👇