WHO — Q.2 of 6 questions to crack your personalisation conundrum

We are having a warm summer here in London and across Europe. So before you head off to your well deserved summer breaks, here is the second intellectual milestone in my own journey to share a few tuppence from my near 10 year journey from the world of digital commerce and personalisation (whoever used this word in 2005 is an undiscovered genius)

In my previous post, we discussed how we can frame a personalisation strategy starting with the all important question — WHY. Well articulated business objectives help brands and campaign experts work towards common goals and foster collaboration across digital and marketing stakeholders. So let’s move to the next step in the Personalisation journey — defining relevant audiences, or answering WHO?.

Many personalisation campaigns today fail despite having clearly defined goals and amazing content, mainly because of lack of relevant audiences. This is clearly ironical. Seth Godin — author of the brilliant marketing books (including my favourite Small is the new big) sums it up pretty well in quote mentioned above. Seth is talking about marketing initiatives and campaigns that miss the mark on the WHO? front. A real-life example of your audience not listening is spraying flyers in the town centre assuming everyone is out shopping for your product. There may be a few who pick the flyer and some who will read the content out of courtesy. It is difficult to know anything about these lovely people willing to answer to this spray technique.

Taking this a step further in digital — we are talking about equivalent marketing techniques such as more than 75% of emails that don’t even catch attention of recipients or home page display banners for ALL visitors irrespective of their interest, affinity, purpose or interest. Everyone is considered a prospect. Marketers today have tons of behavioural data about customers and online visitors, ability to understand segments from fashionable platforms like DMPs (data management platforms) and social tools like Facebook Custom Audiences. Here is a nice article on the perils of Spray and Pray marketing that’s worth a quick browse.

To answer the Who? of my personalisation campaigns — I break the audience definition challenge in 3 parts:

  1. Selection criteria: Work out a simple set of rules to qualify audiences in and out;
  2. Audience on-boarding: Ensure I can onboard or even find the audiences we need
  3. Opportunity sizing: Work out whether benefits of executing campaign for identified number of potential customers outweigh costs of creating and executing the campaign
  1. Selection criteria

Rule no.1 is to derive your criteria from your campaign goal

What audiences give us the best chance to meet our marketing goals. Recently I received an email from an airline with a ‘personalised message’ to complete a flight booking I had abandoned a month ago. If the campaign was looking to drive prospect to the purchase funnel it got the simple ‘recency’ criterion wrong — in the process leaving an impression of spamming with me. Of course a recency criteria of one month could work for a different campaign by the same airline. For example to generate consideration for the airline for my next personal or business trip.

For a campaign to up-sell EarPods, chances of engagement are much higher if you target customers browsing your digital estate on iPhone rather than a Samsung phone.

Defining audiences carefully during the planning phase can also help avoiding wastage of marketing money on audience that are not relevant for your campaign. A simple example is an acquisition campaign for new customers where you can reduce CPA by identifying and excluding your existing customers.

Rule no. 2 is to use the customer data and behavioural data to create target audiences

Digital visitors provide us wealth of data that can be converted into information to decide what audiences may work the best for a marketing campaign. This information can be categorised as User-Identity Information and Visitor Digital Behavioural Information.

1.1 User-Identity information: Digital Marketing defines a customer as visitor to your digital estate — who you identify as a unique individual consistently across any channel. Typically these are customers who have provided you some PII information in past. For your existing customers, a wealth of information is available in CRM databases which can be used to personalise digital experiences. Typical examples are Customer Life Time Value (CLTV), Customers who haven’t been online for a while (for an SMS or Push notifications based campaign), Gender, Demographics, Purchase patterns to name a few.

Privacy is an opportunity: Have a good open discussion with your Regulatory and Privacy teams to crack what customer information can be used for marketing purposes (both by internal and external marketing platforms). Balance this against how much you really need to know to personalise your campaigns — there might be no need to present a 1:1 offer in a display advert, instead it might be enough to draw attention with a personalised creative message.

1.2 Digital behaviour: An online visitor is anyone interacting with your brand digitally. It could be someone completely new (often referred to as Prospect) or your existing customer. In this scenario, data collected by digital channels cannot be mapped to any CRM information and therefore is anonymised. This behavioural information which is typically persisted across sessions using cookies and device-ids, can be great source of insights to build powerful audiences. One of the most common examples is Basket Abandoners — you need not know the individual who dropped-off to retarget when she returns to your site. Another example is Recommendation tools that use algorithms using which products you show interest in, compare this pattern with other similar visitors and recommend what you may like to buy.

Depending upon business objective of marketing campaign, you may decide to use either of these 2 types of information or combination to create target audiences.

Tip here is to make use of your analytics data to understand behaviour patterns of various segments of your digital visitors and apply findings to the audience that you want to target. I came across a case-study a few months back where analytics data led marketing teams to it find that more than 1/3rd of their PPC brand search terms are being used by their existing customers to login. Using DMP and Google RLSA, they are now saving millions of dollars by identifying existing customers on-site and excluding them from campaigns exclusively meant for acquisition.

In summary:

  • Understand who’d be right audiences for your campaign, given the objective of the campaign
  • Make use of customer data available in CRM and digital behaviour data to while defining criteria for relevant audiences

2. On-boarding audiences for the campaigns meant for a set of your existing customers

When you are executing a campaign which is meant to target a specific set of customers from your existing base, either for segment-based or 121 personalisation — a very important step is to onboard customer offline data to digital channels. Many online personalisation tools as well as DMPs offer this capability now-a-days , but underlying process comprises of following building-blocks irrespective of technology you choose.

2.1 Customer-identification: how customers will be identified in digital channels, using non-PII identifier which remains consistent for a customer across every touch-point. This is important for any multi-channel organisation so that a customer who has done a transaction in Telesales should not be spammed through emails and push notifications to to complete the abandoned basket for the same journey, only because offline channels use a different identifier and digital channels use another — for the same customer. Here remember Privacy and Regulatory processes!

2.2 Customer-data extraction: Preferably an automated process which extracts customer-ids that qualify for selection criteria of target audience and other related attributes that is required to build experience (e.g. % discount particular customer is eligible for at Basket step), from offline data systems like CRM

2.3 Customer-identifier mapping between offline and online channels: In scenarios where customer identifier in offline and digital channels is not the same, mapping process takes extracted data in step above and converts it into a digital-identifiable format.

2.4 Customer-data on-boarding: This final step involves making extracted data available to digital tools so that a customer can be identified and presented with an experience driven from using corresponding offline data (or combination of digital behaviour data — both owned and paid media)

Often overlooked during campaign-planning, not having clear plan for on-boarding may delay campaign execution as above mentioned steps require involvement of various teams, especially ones who own first-party offline data.

In summary:

  • Define the process of audience on-boarding, if needed and align with responsible teams for data-extraction and on-boarding to digital channels
  • Automate the processes to avoid any manual intervention and errors\
  • Ensure you have a nod from Privacy & Regulatory teams and clarity around what data you can share and what you cannot

3. Opportunity sizing

A big advantage of defining relevant audience for a campaign is to estimate benefit from the campaign. Using digital data and statistical modelling, you can figure out the size of potential audience and use it to determine whether it makes sense to execute a campaign. This is a simple ROI question. After all, every campaign takes effort and time to create which is cost to business. It may not be beneficial to run a campaign promoting a 5G ready phone in a town where 3G network penetration is below 1%. Of course, you need to take number (1% of population of a big city could still be more than entire population of mid-sized city) and potential average order value (think of big enterprise orders where a single deal is worth few hundred thousand dollars) into consideration while doing these calculations.

Note that not every campaign has to be sales oriented. A service-campaign to inform a couple of households in a remote area about an upcoming electric-outage, using SMS doesn’t need to do any opportunity sizing. Here business objective is to have highly-informed and satisfied customers. So go back to your business objective and brainstorm whether you need to size opportunity.

In summary:

  • Align with your business objective whether it’s required to size the opportunity
  • If yes, make use of data to find out potential benefit and do cost-benefit analysis to assess whether its right to execute your campaign

Takeaway: Identifying audiences for your campaign that are relevant to campaign-objective, have clearly defined process to make those available for 121 personalisation and large enough in number to benefit the business — is important part of every digital marketing campaign. Answering the WHO will help you achieve much better conversion rates by targeting only relevant audiences and save crucial marketing money by not targeting unintended audiences.

In next post, I’ll share how data can be used to identify most relevant locations where personalised experience should be shown — answering WHERE?

Previous posts in this series:

  1. 6 questions to crack your personalisation conundrum
  2. WHY — Q.1 of 6 questions to crack your personalisation conundrum