The key to decode Consumer Experience and Revenue Maximisation – An AI Perspective
Hyper-Personalisation has indeed become the buzz word today. With the constant and infinite availability of data, there is a strong focus on leveraging the benefits of Hyper-Personalisation to give even more finely customized offerings to the clients, thus improving upon the sales conversions. But then, were we not personalizing enough already; showing our prospects things depending upon what they searched for, or talked about at a forum and all this was being done across channels courtesy digital marketing. With the advent of the display networks, the mode to showcase the offerings became further explicit and a level of integration across channels could be achieved. For example, the product one checked on Amazon could now be shown on one another related blog or product site as well.
This was the seller side of the personalisation story, but to make sure that the seller does well, the customer’s requirements need to be understood thoroughly as well. To put down some facts on paper, only about 8% of the people feel that if a message or communication from the firm addresses them by the name, would they be okay to respond. In the same study, a whopping 56% of the respondents said that they feel that the basic, personalisation at offer are totally common place and it is not something that impresses them anymore. (Bannister, 2018)
A similar study done by Segment in 2017 said that more than 75% of the shoppers are not okay with the level of personalisation, and feel that even the bigger brands are failing miserably at creating personalisation. (Segment, 2017)
A research done in a related domain by Infosys revealed that about 41% of the consumers felt that there is a huge scope for personalisation in the shopping experiences being offered currently. (Infosys, 2018)
Indeed, these data points show how unsatisfied prospects currently are with the given state of personalisation at offer. For instance, whenever we surf the internet, we know that the product we just searched for, will be shown to us everywhere without fail. Knowing this fact not just kills the intended benefits of personalisation but also in multiple cases results in the customer giving a negative feedback to the ad which thus, skews the campaign performance metrics.
The reason is that the assumptions and categorisations for what to show to whom are still largely dependent on the cookies being left at the websites and there is only so much one can gather from that. Also, the purchase channels are not just restricted to online. The customer now goes online, checks for the features and then goes offline to the brick and mortar stores to compare. Only then is an informed decision made. Given the ability to collect data only from online sources, the offline channels that offer several critical insights considering the interactions could get missed out, and hence the personalisation insights thus received get significantly limited. For instance, Varun wanted to purchase a sports shoe and so he typed out on Google to search for what he wanted. This was his first “sports shoe” search ever in recent times. Before that he had searched for sports t-shirts, head gears, books, party wear and amongst shoes he had searched for formal shoes which do not have a round top.
He had on an earlier instance, looked for Nike Sports shoes in blue colour without round top but that had been about 5 years ago when he was in Lucknow, and so the data from that search term was long forgotten.
Now, coming back to the present time when Varun searched for a sports shoe again- Personalisation will help the Amazons and Flipkarts of the world to show him sports shoes of varied ranges; fine tuning them further depending upon the cost, brand or other filters that he used. Hyper-personalisation will however show him sports shoes, mostly blue colour or variants thereof with a tapering end and with Nike’s shoes being shown on priority. Which means, even after all this time, his data was still in the system and his preferences were thus acted upon.
Hence, while the ad of a simple sports shoe might have been ignored by him or might have prompted him to search further, the ad for a tapered Nike blue colour sports shoe jogged his memories making him go further – maybe to buy; at the very least, let him know that his preferred option is now available. So, as a response to the stimulus, he clicks on the ad, browses a bit and then closes.
Moving to the next step, Varun thought that since the shoe is now available, why can’t he go and try it out at a local store first. So, he goes online again and searches for Nike stores nearby. He finds that the MG Road, Bangalore store is closest to his place and thinks that maybe he should postpone it till the weekend after the pay day which is another 10 days from now.
On the 7th day from the day he first searched, Varun gets a message telling him that Nike’s MG Road outlet is hosting a Weekend Sale and there is an exclusive offer on certain Nike models (one of which is his preferred model too). Now, Varun has an incentive to purchase the shoe this very weekend, and so he goes and checks out the size because, maybe the shoe will never be this inexpensive again.
The facts mentioned in the paragraph immediately above are what Hyper-personalisation can achieve but Personalisation cannot. This is because Hyper-personalisation involves a concept called CDP (something that Personalisation does not) – a branch of intelligence which integrates data across multiple channels to provide critical analytics.
Also, the problem that we had mentioned earlier about how the customer data from offline point of sales fail to get integrated with the data collected online, it can be solved by a CDP as well because it facilitates an omni channel integration. (Omni Channel integration means that the multiple channels that the seller/firm is using, the data from all of those channels will be collected, stored and analysed at one place).
The context has thus been set; however, the question remains: What is Hyper-personalisation and what does it offer to the seller? For this, let us move to the next section.
Knowing Your Customer Better – The 360° View
Let us start this section with a question – why do we want to know our customers better? The answer can be given in multiple ways depending upon the product or service at offer, but the crux remains the same- to sell better, generate more revenue and maximise profits (in case of sales and marketing, we can call it maximizing campaign ROI).
But, given the current optimization options have started reaching saturation and Google’s analytics (the most advanced campaign manager and lead generation channel right now) becoming more of a black box over time, how do we optimize further and improve upon the results? Is there any specific value add that can be given in terms of products, services, offers, brand value, communication etc?
Hyper-personalisation is the value add that is the answer to all the concerns mentioned above. Hyper-personalisation and Personalisation both involve tailoring services/products to accommodate the preferences of specific individual types – mostly groups or segments. Hyper-personalisation, however, goes one notch above and tailors the services/products to meet the needs of one specific individual. Hence, if it is a funnel – Hyper-personalisation filters the output of Personalisation.
Customers today demand personalized treatments and it makes sense too considering the ample options at hand. A recent study found that 75% (McKinsey, 2017) of the customers expect a consistent experience across all the channels or touchpoints, and about 65% (McKinsey, 2017) opined that they would prefer sticking to a brand if the options offered are more to their specific liking. (Talking logic, the personalized Louis Vuitton bags sell super expensive but still they are sold because it gives the owner the pride that something was custom made for them. Now, filter it down to the increase in the market of customized gifts. The market exists and is thriving because to have a pen with your name inked on it looks good and grabs eyeballs.)
An extension of the aforesaid Customer Experience Impact Survey found that close to 86% (Gensys, 2018) of the people interviewed said that they are willing to pay more for a custom experience, but sadly just 1% of this sample audience felt that any of the existing brands meet their customization expectations.
With so much pressure to personalize offerings to match the expectations, why are the brands not working towards it anyway?
Well, they are, and that is where the concept of CRMs (Customer Relationship Management) and DMPs (Data Management Platforms) came from.
However, the major challenge that remains is that most of these data are stored in silos with little or no integration. For instance, the cookie stored on a Google search platform and the one stored on Amazon or on the brand’s website, even though they exist simultaneously, have little or no integration or connect.
But how can we develop the connect? Is dumping all the data in one place an answer (such a place is called a Data Management Platform or DMP as mentioned above)? Maybe, but then, such a deluge of anonymous data will only complicate things and create confusions given the multiple probable linkages that could be generated. So, what do we do then?
Let us start with understanding the distinction between an Anonymous and Identified individual/data point on such a data management platform.
An Identified individual or data point is known by a name, phone number, email address or a bank account (there can be other identifying parameters too but let us stick to the ones mentioned for an easier understanding). An Anonymous individual or data point may have all these too but gets stored in the system as a cookie or device ID. Hence, while all the relevant information is available, it cannot be linked back to a single identifiable point because the data was obtained through different channels.
The footprints that an individual leave online is usually across multiple channels. How does that get aggregated then?
A central database that lists all the identifiers associated with an individual seems to be a good idea. This central database has specific rows and columns defining all the identifiers such as a cookie ID for a particular website visited, an email address from the CRM, the IP address for the system, and so on and forth, thus providing a master sheet of sorts to know all the probable touch points the particular individual has interacted with over time. The central database then branches out to other smaller databases, which contain the actual data from the source systems for these identifiers such as the websites visited, advertising platforms interacted with, the web journey to the point of sale, etc.
The path defined in these databases help in creating a unified view for an individual, thus identifying the person from the group of the data. This unified view thus created is called the 360° View, the name derived from the fact that one has to access just one point to see the entire interaction history of the person concerned.
To give a better perspective, a 360° View offers the following (but not limited to) information of an individual:
- Activity Trail: A detailed trail of the individuals’ online or app activities, their purchase histories, the path they followed before finally making a purchase, etc.
- Interaction Trails: Data about the customers’ engagement with the respective brand, its competitors, the purchases across this brand, their loyalty with the brand concerned, the feedbacks given, etc.
- Campaign Effectiveness: Data about which campaigns performed better, which channel turned out to be the most productive in directing traffic or maximizing conversions, the responses made to personal campaigns such as emails or SMS, details of un-subscription or subscription histories, etc.
All this, but we talked of a central database which connects all the data at one point – isn’t that either hypothetical or what a data management platform does?
So, first – it is not hypothetical, and second – the concept we are talking about is more than a data management platform, it is indeed an extension of that however, and is called a CDP or Customer Data Platform
Customer Data Platforms – Enabling Hyper-Personalisation
Before starting with the concept of Customer Data Platforms, let us have a look at the two terms used interchangeably with CDP. The fact that this interchangeability is not justified will be established on the way.
The first such term is a CRM or Customer Relationship Management (platform). A CRM is aimed at implementation of customer centric processes, mostly spread across Sales, Marketing, Post Sales Relation Building and Revenue Monitoring. Tied up with other business analytics tools or data/technology platforms, CRM can be instrumental in web content management and data management too.
CDPs and CRMs are similar in the terms that both of them collect data and are capable of data segmentation and categorisation. However, CDPs are better prepared at handling faster data processing and data analytics derived from multiple sources – a feature CRMs can offer only after several complex integrations with third party providers. Moreover, CDPs come with the functionality of real time data processing- a feature not belonging to the CRM functionality kit.
Next in line comes DMP or Data Management Platforms. DMPs are the tools focusing mostly on Customer Segmentation, Targeting and Campaign Optimization. Their capability involves integration of first-and third-party data aiding in converting the anonymous data into an identified version of itself.
DMPs can work as complimentary to the CDPs. However, the former was built mainly as an advertising solution and hence they focus mostly on Device IDs and cookies as the identifiers. Also, the data memory is restricted only to one purchase cycle, once done the data is scraped off.
The usage of CRMs and DMPs has resulted in creation of several data silos. The fact that the quest for personalisation advancement has resulted in a data deluge complicates the problem further.
As a solution for this, CDPs were developed. An enterprise CDP is capable of aggregating all the available data from all conceivable sources, be it a CRM, a DMP, a loyalty management software or anything else. A CDP can unite all these siloed customer profiles into a single Personal Identification Number (PIN) (CDPI Insights, 2018). More so, it comes with the flexibility to collect raw data as well; an ability which allows the user to utilize even that data that wasn’t initially defined to be collected.
The illustration below will clear out the functionality of CDPs further:
AI in the Scheme of Things
Close to 50% of the interactions happening in the digital space today are a multi-channel (McKinsey, 2017). Hence, as is obvious, the data is spread across all these channels- not necessarily integrated with each other or even consolidated at a single place.
Irrespective of how big the brand is, the failure to create a universal experience across all the engagements, be it online or offline is becoming a significantly incurred cost. As per the CMO Council Research-2019, less than 7% (CMO Council, 2018) of brands have been able to deliver real-time data-driven engagements across the multiple interaction channels. Taking a specific case for instance – the case of paid marketing. Basis the data obtained from analytics, one can choose between the performing and non-performing keywords. However, knowing the extent of performance and the reason behind it, can help the firm to reduce its ad spend by a minimum of 20%, while improving the lead generation by about 30% and the lead quality by- well, that is something we have not being able to calculate but that there would be an improvement is a sure shot!
At Racetrack, we understand that all these things are easier said than done. With the infinite human capabilities, which still remain vastly unexplored, the significance of AI as a support function has grown in the industrial circles. What started as basic automation and grew into chatbots, has now evolved into a critical sales intelligence tool which has enabled monitoring of the traits and sentiments of consumers. The next AI capability that we are working on currently is a CDP model, so that when someone interacts with any of our frontend AI platforms (or Non-AI platforms too), the backend acts as a CDP where the said data can rest forever, uniquely linked to the individual consumer.
Achieving Hyper-personalisation now, seems just a few steps away!
Bannister, K. (2018, April 24). Beyond the Basics: Why Customers Are Demanding Next-level Personalisation. Pure360 BrightTalk.
CDPI Insights. (2018). Understanding CDP. CDPI.
CMO Council. (2018, June). The Need for Hyperpersonalisation. CMO Council Research.
Gensys. (2018). Customer Experience Impact Survey . Gensys.
Infosys. (2018). Rethinking Retail: Insights from Consumers and Retailers into an omni-channel Shopping Experience. Infosys.
McKinsey. (2017). Customer Experience: New Capabilities, New Audiences, New Opportunities. Mckinsey & Co.
Segment. (2017). The 2017 State of Personalisation Report. Segment.
This article has been co-authored with Racetrack.ai