Simple Digital Analytics Hacks for Audience-centric Marketing

D.W.

12 min readJan 2, 2019

…analytics itself is never a tool, but a mindset.

Overview #TLDR: Here we hack our digital analytics tools on Audience level to explore growth potential of digital business. In this article, we have a same-old same-old intro about businesses, a boring bonus technical lesson, a few hopefully-useful examples (featuring GoBear, Linkedin Ad), and some of my thoughts sharing at the end to wrap things all up.

As always, happy reading. Enjoy.

“Always make the audience suffer as much as possible.” — Alfred Hitchcock

Don’t get it wrong, Alfred is more likely to mean communicating with audience through strong emotions [1], as this is what makes audience ticks. At least movies work that way, still, digital is just another form of media, right?

Jokes aside, the growth on digital marketing in last decade is increasingly emphasizing the importance of audiences with more and more targeting capacity available among paid, earned, and owned traffics. This is good for the industry as this enables marketers to delivery more targeted messages to their audiences. Yet, Digital Analytics on this spectrum has been often overlooked (guess maybe attention was most spent on automation & Machine Learning. Like myself…), and digital analytics remains being used only as a quantitative reflection of behavior rather than a qualitative tool for strategic decision (at least in my hometown, it is). Under the growing shadow of other buzzword-ish techniques on customers segmentation or etc., i think i would take the chance to show the true color of this grounded digital tradition after spending two years in a trending omni-channel business, digital finance.

But before we talk about the nerdy bits, let’s talk about Business.

The Role of Audience in Business Today

(for the record, as of Jan 2019)

To drive business we talk about strategies, to define strategy we talk about markets, or simply a collection of types of audience. Not only “Audiences” is the most commonly known & understood noun among strategy planning, it’s also frequently being used as the foundation for building up the whole strategic concept, as well as visualizing whom & how the business is interacting to / with. Here’s the list of tools and processes that utilize this concept:

the Reachable targets or fans in Facebook, Googl GDN, and Linkedin Ad.
the Readers Profiles we want to communicate via our content strategy
the Relationship Group we build relation for CRM strategy
the Segments we communicate within Bank, Marketplace & eCommerce
the Accounts as in Account-based Marketing popular among Finance firms
the Personas and Profiles in Design Thinking
the Users as in Users Stories within Agile Development

People is the always key to every business (be it we sell stuff to them as customers; or we sell them as stuff, well, like Facebook.). To understanding our audience, the old-school way would be segmentation & modeling technique through CRM with transactional data, or otherwise the trendy way would be clustering, classification and prospective modeling using Machine Learning or even Deep Learning. Nevertheless, the things we want to learn from our audience is always the same: what they are, what they have, what they like & love (or not), and what they do / would do.

So how Digital Analytics could facilitate an audience-centric analysis? The key is how we measure them.

Basically this is what our Analytics Tool sees without any twists. (credit: William Murphy from Dublin, Ireland)

The Scope of Measurement

Basically, we want to achieve the following from our choice of analytics solution (be it Google Analytics or Adobe Analytics or else):

with belief that, the behavior, or digital footprint, reflect the different aspect of the person himself, we thus monitor certain key actions or events that could hint his profile or persona.
the tracking should enable abstraction of a group or the segment of people, rather than individuals. (as we need to deduce insights of types of audience to promote strategies, not tactics)
given the abstraction of groups, we should be able to do segmentation of digital data, like funnel conversions, by groups, for audience analysis
the tracking have to be able to label a person, not a visits, a pageview or else. In other words, data should *persist* for persons regardless their other behavior on the site.

With this mutual understanding, now we can talk about how we could enable Audience information for our analytics tools.

*Bonus Lesson : Do You Know Visitor Level Tracking?

Feel free to skip this section if you know what persisting data means here, or if it’s too technical to you. Otherwise, stay a while and listen.

Normally when we use analytics tool with its out-of-the-box configuration, it provides insights mostly on visits level, as the tool itself monitor data throughout a visit. For example, “how long does my audience stay on my site per visit on average?” or “how many visits come from facebook.com?” . the former one averaging the summation of duration on site for all visits, and the latter one summing all visits that are sourced from facebook.com. Technically speaking, these type of monitoring means to store data until the visit/sessions is ended. (in Adobe Analytics we call this variables expired by visits, also an out-of-the-box setting). To facilitate Audience-centric analytics, we need to enable labeling data and events on a visitors level, or alternatively speaking, configuring the data to be expired by visitors OR even never expired. the configuration details could already be all over the internet so i don’t bother to repeat, here’s the portals for your convenience.

For Google Analytics, use Custom Variables, the following snippet will fire the variable, a dimension, that sticks to the visitor level.

_setCustomVar(index, name, value, 1)  #1-visitor level

For Adobe Analytics, configure Conversion Variables, or eVars as Never Expired under Expire After in Admin console, then fire the eVars as we normally would.

Generally speaking, with these, we can tag people once and for all based on certain events they’ve triggered. For live action, you may take a look on how I did a cohort analysis hack using this feature in the old days here.

Labeling — Give Quantitative Traffics Meanings

“If you want to create messages that resonate with your audience, you need to know what they care about.” — Nate Elliott

Digital analytics beginners usually focus a lot in quantifying traffics, and what we indeed interested is in the quality, the meaning, of them, with respect to our Audiences. To achieve this, we label our Visitors, the People that visits our digital channels, with Audience-level tags. And we have three ways to label our visitors:

Hack #0 — use out-of-box solution
Hack #1 — label by behaviors
Hack #2 —label by classification from upstream sources
Hack #3 — put everything together!

Hack #0 : Out-of-box Demographics & Interests

I know i’m sort-of cheated on this (😈), but i guess it’s still worth noting Google Analytics does provide an estimated demographics & interests assessment for your site traffics. Good for traffics segmentation and business planning, and you can find the details here on Google Support, otherwise the internet is flooded with this function for your reference too.

* No luck for Adobe users if they seek for default solution. Still at least they can enable it through data importation.

Hack #1 : Audience Profiling by Digital Footprint

Now let’s roll our sleeves and get our hands dirty.

To learn more about your Audience, it’s nothing better than studying what they’ve input. Why? Because for some business, customers input is essential and usually be true & real as it’s part of the application process, one typical example is Insurance Quotation flow. Here’s an example on GoBear quotation engine for travel insurances, with Family, # of children, and additional covers being our most interested items.

Let say we fire the Google Analytics Custom Variables once the visitor press “Show My Result”, what we might get? We could…

label him likely having a Family when he choose for my family.
label how many children he likely has when choosing 1+ children.
label him possibly a sport player/lover if he choose Skiing coverage.

What if we also exercise the same labeling practices for Loan & Mortgage quotation engine? 😈

The beauty of this is that not only it doesn’t rely on completion of the transaction (which normally requires for complete picture of “customer” under CRM point-of-view), it’s also absolutely anonymous *— we indeed do not know who the person really is, nor we need to, because we’re only interested in the subsequent segmentation or audience types that this visitor is representing, based on what their behavior (the inputs) might represents.

We will later discuss how to leverage this kind of labeling with a real business case. For now, let’s move on to the next level: what if we don’t have a journey that reflect essential audience information?

* Warning though, tracking personal identifiers like emails and names are highly, definitely, absolutely NOT recommended. You don’t want to fall into the privacy trap, do you?

Hack #2 : Abuse the Power of Social Media

Not all business has so much essential information through the journey, indeed, most of them do not. To work around this, we put on our white hats and start hacking for the personal data, which, is always rich on social media.

Facebook audience targeting has been famous for its richness & depth and commonly discussed already, so instead of using it as an example, let’s use a more unique network: Linkedin.

For starters, Linkedin Ad Targeting provides marketers with a wide range of options for audience targeting[2], including industry, experiences & skills, job titles & even job seniority (which is somehow a good indicator of HNWI, or High-net-worth Individuals) . The depth on career-related information expands a whole new level of spectrum for targeting, particularly benefit to advertisers who promote professional services, like eMBA and pension funds in B2C, or cloud services & company coaching in B2B area. Marketing strategies on Linkedin would definitely worth it’s own article, but for now, let’s get back to our focus on Audience Analytics — the audience selection is on Linkedin, how can this benefit to us?

We need not overthinking this, in fact, all we need to do is to bridge up this information through Campaign Tagging.

In case you do not know, campaign tagging is a way to label our upstream traffics driven from an external source (e.g. Linkedin Ad) by appending additional parameters at the end of the destination URL. Summoning our old friend Google URL Builder as an example, we could use its parameters, call utm parameters to identify the nature of the source traffics, including sources (url), medium, campaign, terms, & content (or even more if you adopt a more custom notation, like campaign tracking code cid in Adobe Analytics). For details, you may refer to Google Support for the usage of utm.

As long as people land onto your site with a tagged URL, additional info as denoted by your parameters would be captured and stored at Visits / Session level by-default. This is helpful for our Audience measure because we can in fact present the same destination URL with different tags for different audience groups targeted, so that their information & profiles will be anonymously captured once landed on our site. Don’t get it? Here’s an example on how it could work:

A simple view on Audience Communication plan WITH Tagging enabled.

Let’s assume we want to pilot a new eMBA campaign to HNWI, School Achiever and Mid-to-Senior people without a MBA. For each audience type we define corresponding campaign tagged URL by using different content parameters, then we publish the Linkedin Ad according to each targeting setting and start to let the traffics rolling in. After a while, you should be able to see this on Google Analytics, under Source/Medium > Ad Content report :

A dummy report with performance by Audience type

Despite we (our website) were incapable to identify tricky audience type like HNWI or School Achievers, Linkedin Ad targeting now not only send us the targeted traffics but also help us labeling them accordingly.

Even better is that we can convert our sessions level data into a Visitor level segment. The example below uses Google Analytics to help locating group of qualified professionals with busy schedule due to parenting (well guess we can promote the flexibility of eMBA to these audience. Never sacrifice family for career progression, right?). In simple wordings, this filter capture Users who have non-MBA label (from Linkedin campaign) AND Family label (from Facebook campaign).

If someone who visit us through Linkedin with non-MBA tag and also visit us from Facebook through Family…

See how we can combine the power from two social media, Linkedin & Facebook, so as to empower us for profiling our Audience? Good. Now, let’s move on and see what we could do we all these Audience data.

Hack #3 : Surfing with the Audience Insights

By enabling Audience Profiling through internal & external measurement hacks, we now have opened a whole new spectrum of Digital Analytics, allowing ourselves to finally view the data in Audience Perspective. Let me share with you two of my cases with you.

Case #1 Product-Market Fit Validation — Finding product-market fit is always difficult, especially for startups, which have 42% of odd to be failed in the first place as simply no market need[3]. I have once experienced launching a new digital product within a well-established finance institution with large population of existing customers. The product is hypothesized to be popular among young people yet the age distribution of existing customers is indeed (way) more mature. To minimize the spend, the launch focuses mainly on current customers, supported with the Audience analysis above. What we have found was that the statement itself is not 100% true:

a) Young people does convert better, yet only if they’re Single;

b) even for Young people, female converts less, which later found out to be resulted from lesser tolerance on pricing;

c) Mature people converts better than Single people if they’re married.

To grow our business, what we could do next was simple: acquisition campaign for Single & Young people (as we lacks this population), and enhance offering to Married people among our customers as main strategy.

Case #2 Scale Communication Tactics for Similar Audience — Two directions we could work on after we have locate these high value audience groups: either work it out internally, by journey optimization, or externally, through remarketing. For Google Marketing Platform users generally you can do so by Creating an Audience, and pass it either to Google Optimizer for journey twisting or Google Adwords for retargeting. Try this now & trust me you will definitely fall in love with Google tool afterwards if you didn’t.

For Adobe Experience Cloud users, Adobe Audience Manager with Adobe Experience & Adobe Campaign will be your alternative solutions for customizing internal and external experiences respectively.

Another similar hack for Facebook pixel on Audiences would be using Custom Conversions. Given that our URL landing would be tagged with different Audience Type (under utm_content, remember?), we can then easily segment visitors and retarget them on Facebook with tailor-made messages. I believe Linkedin Ad Targeting allow remarketing with similar method as well. Though i’m no expert in this area, i think i better leave this exploration to you. 😃

Last Thought

With the recent growing attention on Machine Learning & AI techniques, digital analytics seems to be playing a lower hand and often overlooked. Personally, after self-taught myself on a few Deep Learning topics, my belief is even strengthened that no matter how technology grows and expands, analytics itself is never a tool, but a mindset — a thought process that enable us to find answers from those aloof metrics, and bake them into warm & delicious stories that facilitate decision making. Always remember in business world we interact with People, not numbers, or at least until we turn them into dollar signs. 😈

Thank you for the read, my dear reader. I hope this article serve you well and enlighten you in some ways. Should you have any questions, feel free to leave a comment, or connect me on Linkedin.

As always, stay healthy, stay hungry. Farewell, and see you around soon.

Cheers,

D.W.