B2B Web Personalization Maturity Curve

Every B2B marketer in the world dreams of a state where every single touchpoint with a prospect or customer of your business is personalized.

But, what does it actually mean to be personalized?

My take on being personalized: content is relevant to the viewer. With that you, as the marketer, strives to provide value to that viewer about your business.

Value could take the form of education, a different perspective, or simply a spark of joy/emotion. You can measure the value you have delivered by means of engagement and action, such as a click and scroll.

In this post, let’s talk about the common phases or approaches that B2B companies are taking on the topic of web personalization.

B2B Web Personalization

If you have read my previous blog post, What is B2B Web Personalization, you may know that B2B web personalization centers on 3 elements:

  • Identification – the attributes to be used to identify that person
  • Content – the content that may be relevant to the person, including location of that content being shown
  • Algorithm – the logic behind the relationship between identification and content for the personalization

If you want to get an overview of B2B web personalization, read that blog post!

B2B Web Personalization Maturity

If that is B2B web personalization at a high-level, I would describe the B2B web personalization maturity journey as occurring in three phases.

1. Rules-Based Personalization with Reverse-IP Look-up

Personalization in this form is what most A/B testing platforms – Optimizely, Google Optimize and VWO – focus on.

These platforms act as a service layer between the three elements.

They take an identification attribute and read it via integration with a reverse-IP provider (e.g, Clearbit, Demandbase, 6sense) to surface the content.

The logic between the identification and content is rule-based and it’s often a 1-to-1 relationship.

For example, if site visitor is from the financial industry, show this image.

This type of personalization is table stakes at most mature marketing organizations, particular in B2B SaaS.

A/B testing (or experimental culture more generally) is something that every organization is driving for, especially where top-of-funnel volume and/or customer LTVs are high.

2. Dynamic or AI-Based Personalization

The core difference between the first and second phase is the personalization algorithm.

Instead of a if-this-then-that static rule, in the second phase there is a dynamic or AI-based algorithm (i.e, the personalization algorithm) that determines the content to optimize conversion based on the identification attributes.

With that, we see an increase in the number of inputs required in this second phase:

  1. Available identification attributes
    • [Individual] Behavioral – the activities that browser has done on your business web properties (e.g, page-view, IP-location)
    • [Account] Firmographic – the account information of that IP address (e.g, company industry, company revenue band)
  2. Available content

Both of these inputs are essential to determine which content is the best based on a single or multiple of identification attributes.

Having said that, the collection or aggregation of these two inputs could still be a rather manual effort into the personalization platform.

3. Full Scale Personalization – Customer 360 view

The last phase is built on top of phase 2, further leveraging the AI-based personalization algorithm with data.

The core differentiator in phase 3 is that the number and accuracy of inputs have exponentially increased, because of additional 1st party data (e.g, form fill provided by customers).

The available identification attributes now becomes:

  • [Individual] Behavioral – the activities that the browser has done on your business web properties plus activities within your products or free trial environments
  • NEW [Individual] Demographic – the static attributes of that individual once known (e.g, name, title)
  • [Account] Firmographic – the account information of that IP address plus what the individual provided.

With the massive amount of data becomes available, the business requires additional technology to organize the data.

Using the technology to blend and group the data attributes will be able to form identification segments.

If I use Amazon as an example: You (e.g., a 50 year old female in NYC) browsed product X, you are likely to put into a segment of “Interested in X”. And you can be in multiple segments at any given time.

On the other hand, the content algorithm organizes the available content and how they relate to other content.

Continuing the Amazon example, you will be shown product X in location B because you are in the segment of “Interested X”.

With the content algorithm, you might have a % of chances that you will be interested in product W and Y, in a similar lookalike audience.

To tie it all together, the personalization algorithm may also show product W and Y in location A and C.

Summary

Web personalization is a journey on which most B2B SaaS companies are currently embarked.

Although companies may be at a different points of the journey, the challenges and milestones are very similar depending on the volume of web traffic and relationships between website/product and go-to-market motions.

If you are interested to know more about my thoughts, add me on Linkedin!

What is Web Analytics? (Part 2 of 3)

This is a continued post of my thoughts on the topic of Digital Marketing Analytics, specializing web analytics.

What is Web Analytics?

Web Analytics is the collection and analysis of data from a web property, let it be a single website or a collection of (sub-) domains.

The five steps of the analytical process is very similar to digital marketing overall.

  1. Data collection – Collect web touch-points via Google Analytics and Google Tag Manager
  2. Data piping – Depends, to be explained later
  3. Data analysis and insights – Perform analysis to obtain insights
  4. Visualization – Surface insights via platform reports, Google Data Studio, and/or other visualization tools
  5. Optimization – Digest insights and adjust programs run by marketing to optimize for conversion or KPI

What is the purpose of Web Analytics?

It is important to understand the core purpose of a function before diving into the detailed implementation.

What are the questions or insights is the business trying to answer?

In simple terms, the purpose of web analytics is to understand:

  1. The customer journey prior to a conversion (which is typically an email submission in B2B) via organic channels (paid channel journeys should fall under advertising analytics)
  2. How web projects impact web outputs to inform investment decisions

For #1 customer journey, think about every single pageview and how they relate to a conversion.

What are the most viewed pages in last month? What are the top converting pages?

Keep in mind, however, the relationship between sessions and users.

This adds complexity to the types of questions / answers we can ask because, as you can imagine, B2B web visitors do not just come to your website one time and convert immediately. They come to your website multiple times in different length of sessions and over different time periods.

For #2 web projects, think about the impact of a new page or a re-design and how it affects the KPIs that you care about.

Is the new pricing page more engaging? Does long-form content keep visitors on the site longer?

These are the questions you can answer to evaluate the time and resources you invest in your team to justify and prioritize projects.

The Complexity of Questions and Answers

At first glance, web analytics is relatively straight forward to understand.

However, once you spend a bit of time diving into the detail, you’ll quickly realize some answers are not that straight forward.

Here are two buckets that typically add complexity:

1. Joining data sources

Most businesses optimize their website with anonymous data assuming everyone is equally important.

That may be true for B2C sites where everyone could have an equal probability to check-out.

But the reality is, B2B businesses weight accounts and buyers differently depending on how they segment customers (e.g., potential account size and individual buyers’ persona/title).

While it is possible to know the attributes of a visitor once someone submits their email on your site, this capability requires an integration between your CRM and analytics solution.

While it is possible to uncover which advertising campaigns drive better quality leads, that capability also requires an integration between advertising platforms, your CRM and analytics solution.

A lot of companies also want to connect web data with their product data, especially if the company has a trial environment for people to try the product.

2. The customer journey is complex

If your marketing team has done a good job, there are probably hundreds, if not, thousands of touch-points before a record of any kind is surface-able to your CRM.

In addition, from a web analytics point-of-view, the relationship between landing pages (from different source/medium), sessions and users can be complex depending what types of answering you are looking for.

For example, you are launching Campaign A and Campaign B at the same time, each has its own conversion asset.

You want to know: Does Campaign A perform better than Campaign B, in terms of absolute conversion volume and conversion rate?

Under normal circumstances, you would look at 1) how many conversion each campaign generated and 2) # conversion over pageview for conversion rate.

However, there are some edge cases that complicate this picture:

  1. A single person can actually convert in both Campaign A and Campaign B
  2. The person may not convert in the asset of Campaign A or B, but other evergreen CTAs such as Contact-Sales or Free Trial, despite Campaign A or B may bring the traffic to site.
  3. When it comes to calculating a conversion rate, the traffic (or advertising budget) may not be equal in the same time period, influencing the # pageviews denominator.
  4. The CTA of Campaign A and Campaign B are different and therefore have different levels of friction, for example, one is watching a video and the other is downloading a whitepaper.

While the business question is straight forward, these edge cases muddle the comparison and make it hard to do an apples-to-apples comparison.

This is especially so if the sample volume is low (i.e, hundreds not thousands of conversions), which is common in B2B settings.

To address the complexity above, you either will have to understand and accept the limitations of a basics analytics solution, or decide that you want more precision.

If you prefer the latter, it’s probably time to look for a more advanced analytics solution, which will typically involve putting multiple data sources into a data warehouse for easy ETL and aggregation.

Conclusion

Again, it is important to start with figuring out what questions you’re asking for the business.

Then from there, understand what your current systems are capable of answering.

Remember, it will take time for your teams to get to the same page. Be patient and start smalll tackling one piece of analysis at a time.

✌️

Digital Marketing Analytics – Introduction (Part 1 of 3)

This is Part 1 of 3 in a series of digital marketing analytics. This series is more of a brain-dump of my thoughts when thinking about the topic.

What is Digital Marketing Analytics?

The number 1 SERP from Google of “Digital Marketing Analytics” is from Hubspot: “the translation of customer behavior into actionable business data.

I agree with the definition at a high-level.

To take it one step deeper, Digital Marketing Analytics is “driving data-driven decisions (with a focus on calculating return on investment) to provide insights on obtaining a maximal return on every dollar invested across all marketing channels or campaigns”.

If you have a good foundation of digital marketing analytics, there is always more your business can do.

An example of next level would be using the data to forecast and predict return with data modeling.

What is “return” in B2B digital marketing?

If getting return is one of the objectives of digital analytics, we need to define what is return.

Return can be calculated on a number of metrics in B2B – it could be revenue, pipeline and leads.

Why?

Given the whole point of B2B digital marketing is to fuel the business, the metrics on which you measure return should match the sales/marketing funnel progression that the business is built upon.

This is the most intuitive way for digital marketers and their work to form a direct relationship with revenue and other revenue-generating teams.

Once you have those output metrics established. You can join it against cost data to calculate cost per revenue, cost per pipeline and cost per lead. These become your efficiency metrics.

If efficiency is ‘good’ (i.e., falls within your threshold of what’s acceptable), it makes sense to continue to invest with positive returns.

What about other metrics?

Many other metrics closer to the top of funnel – for instance impressions, clicks, web visits, conversions – have a less direct relationship to revenue.

They are typically have a many-to-many relationship, which makes their impact very difficult to measure accurately.

However, there is a temptation to anchor on these as they are typically available out-of-the-box (via ad and campaign platforms) and require less computation.

Definitions are important

We touched on the definition of “return” above, but there are many more business metrics, in terms of inputs and outputs, that are critical to unpack the drivers of any business.

To start with Sales, the inputs are often headcount — assuming each headcount performs as expected to hit 100% attainment (the efficiency), the business will get predictable revenue (i.e, outputs).

Many may say that the unit economics of sales staffing is science, the management of sales selling is art.

For marketing, the marketing activities that drive the outputs are the art, and the science is what we are trying to unpack with digital analytics.

To set a baseline, let’s align on the following digital marketing metrics, which may give more context later:

  • Top of the Funnel (TOFU): A web visit
  • Middle of the Funnel (MOFU): A lead (or conversion event, e.g., email capture)
  • Bottom of the Funnel (BOFU): A stage X opportunity accepted by AE

The Ultimate Challenge

The real challenge of digital marketing analytics is to understand the relationships between pre-TOFU and MOFU in a cross-device, multi-touch environment.

From MOFU to BOFU, most companies can do it pretty well with a Marketing Automation Platform (MAP) (i.e, Marketo, Pardot, Hubspot..etc) and Customer Relationship Management software (CRM) (i.e, Salesforce)

To understand the relationships between pre-TOFU and MOFU by factual data analysis, you will need to collect the data.

And it is difficult, because of 2 reasons.

  1. While Lead>Account>Opportunity is a 1-to-1 relationship, impressions>web visits>leads are many-to-many relationships
  2. The sources of impressions and web visits are complex — it could be search, paid, direct, referral
    • Compliance regulation like GDPR, ad-blocking technologies, browsers removing cookies and privacy mode make it even more difficult to attribute the sources

It is worth noting that if the business has no data, there is no science to be done.

Now, we should be on the same page on business understanding in terms of digital analytics and the marketing funnel.

Let’s dive deeper into data/digital analytics with inputs and outputs.

Inputs and Outputs in Digital Analytics

The framework of inputs and outputs are important to structure what is need to produce certain outcome, that includes both technology and human effort.

In broad stroke, a digital analytics process cycle can be split into 5 steps:

  1. [Input] Data collection – the process to collect raw data
  2. [Input] Data piping – the process to manipulate raw data to a structured format
  3. [Output] Data analysis and insights – the process to extract insights from structured data
  4. [Output] Visualization – the process to surface structured data to be digested easily
  5. [Input & Output] Optimization – the process to translate the insights to execution literally

Depending on the maturity of the business, the implementation of the 5 steps could look very different.

For a startup, a general analytics tool like Google Analytics makes step 1 – 4 easy and step 5 becomes the iterative process that your team is executing.

If your business has grown out of Google Analytics from an analysis standpoint, you could consider expanding the analytical capability by extending the number of data sources available and performing the analysis out of a data warehouse directly.

Conclusion

Business stakeholders at each organization need to identify the requirements and subsequence actions that would be performed with the insights in order to make the investment of data analytics worthwhile.

As the business continues to grow, it may be necessary to specialize in web analytics and advertising analytics. I will share more details in part 2 and part 3.