Website Personalization Isn’t a Luxury. It’s a Necessity.
Recently, VWO product marketer Siddhartha Kathpalia chatted with Noble Studios about the Value of Conversion Rate Optimization. His insights into optimization strategy resonated with readers, so we invited him back to discuss an equally important topic: personalization, which Kathpalia likes to think of as “hyper segmentation.”
Enjoy our talk with him below and, if you’d like to learn more about personalization efforts and personalized content within your organization, contact us, and we’ll be in touch.
What is website personalization, and why does it matter?
Website personalization presents real time information that matters to an individual and not generic information that may be irrelevant. Thus, website personalization is hyper segmentation.
It matters because not all information is relevant content to every website visitor.
How does website personalization work?
How does it compare to and contrast with typical A/B testing technically and methodologically?
Website personalization works based on hyper segmentation, where every individual forms a segment of their own. A/B testing and Website Personalization are very different. A/B testing helps validate what works best, while website personalization is presenting relevant information. A/B testing helps identify what is relevant information, thus laying the foundation for a personalized web experience.
What are the benefits of personalization for site visitors?
What’s in it for the users visiting your site?
Personalization and unique experiences help visitors bypass irrelevant information, thus avoiding a lot of fluff that comes their way every now and then.
Why does personalization work?
What factors are at play, and why does a personalized experience lead to better website performance?
Personalization may or may not work. If decisions are made based on “gut” feelings or beliefs, personalization may even be detrimental. Infrastructure is required to achieve personalization which is a cost. Incorrect infrastructure or personalization done wrong can be a very costly affair.
If done well, visitor experience and conversion rates go up, with budget spent at the right place leading to high ROI and increased conversions.
Are there some types of website personalization that are more useful than others?
Websites with a large amount of traffic or repeat visitors are better suited for personalization. Algorithms require a lot of data to iteratively get better.
Besides websites, where are we seeing personalization applied?
Personalization is very commonplace. From streaming services to email providers to apps, personalization is what drives repeat usage. As a rule of thumb, any digital platform that offers a “my account” or equivalent is personalization.
How do you identify audiences for personalized experiences?
Also, how does personalization differ by industry/business model? For example, is a personalization strategy different for an eCommerce site vs. a B2B lead gen site?
If a visitor is identified as a person with some unique identifier such as email or phone number, etc., personalization becomes possible. Similarly, with several services available today that help identify a visitor without disclosing PII or customers data, this enables personalization. This implies that personalization is possible for any type of digital service. The effectiveness lies in delivery, i.e., how well the brand is implementing its experience based personalization infrastructure.
What are some of the necessary components to get started with personalization?
The most important first step for personalization is a hypothesis that should be tested. Once a hypothesis is validated and there is proof that a personalization variable will be effective once implemented, only then should brands embark on spending budgets on building the infrastructure.
What types of skills are required for a team to implement personalized experiences?
Experimentation, data capture, data enrichment, segmentation, and creating and optimizing algorithms.
What does the process for implementing personalization look like from a high level?
There are three cyclical steps to the process of implementing personalization. They are hypothesis testing, data collection, and deployment.
Do personalized experiences require ongoing maintenance? What is required to keep personalized experiences performing well?
Yes, everything needs maintenance. Formulating new hypotheses, consistently evaluating data, finding edge cases and solving for them, iterating algorithms, and sorting new data are some of the activities to keep in mind to ensure personalization is performing well.
As website users start to expect personalized experiences, is there a risk of not implementing personalization?
Risk is only in doing what one does not know about. If brands have tested whether personalization works for them or not, then there is no risk in pursuing personalization or abandoning it altogether.
Can you explain why it is hard to accurately calculate the ROI of experimentation and personalization programs?
Calculating the ROI of experimentation is actually not difficult. All experiments where a change does not win and hence is not deployed effectively save cost and avoid damage. Without experimentation, teams make deployments because individual and team performance is gauged on output.
Thus, experimentation helps avoid damaging output and sieves through only the changes that will lead to positive results. Similarly, if an experiment showcases positive results for a personalization hypothesis, the ROI can be known and will only be positive, in line with the experiment’s result.
What are some of the typical success metric teams use for personalized experiences?
Uplift in conversion rate is the most effective success metric.
If a better customer experience is a key goal of personalized experiences, can we relate that into ROI for business executives?
As long as better customer experience can be quantified, it can be related to ROI.
What are some of the difficulties translating the performance of specific personalization campaigns into revenue?
Gut-based deployments for better or more effective personalization are largely immeasurable because of the noisy and chaotic environment. But once a personalization campaign is A/B tested with a KPI measured, the revenue impact can be easily known. Uplift multiplied by the number of conversions (baseline) multiplied by ARPU (average revenue per user) gives you the revenue uplift.
Instead of focusing on revenue or ROI, what are some ways that experimentation teams can measure the impact of personalization?
As a business, removing focus from revenue or ROI is not something I recommend. Although, I understand sometimes measuring these may be difficult and convoluted, perhaps even due to sparse data. In such cases, metrics that lead to revenue can be valuable. These may be macro conversions (like the number of conversions, ARPU, etc.) or micro-conversions (like clicks, pageviews, bounce rate, etc.)
Can you give some examples of highly successful personalization programs that leveraged the VWO platform?
Food delivery apps do a tremendous job of personalization, especially with the CRM programs. Similarly, eCommerce websites and OTT platforms are very advanced with personalization. Facebook and Instagram thrive on personalization!
What common characteristics do you see in successful personalization programs?
I have seen successfully executed personalization programs mostly on B2C websites and digital properties, especially when there is a login required. The more information a user shares with the brand, the better the personalization. The more data (at large) that a brand has, the better personalization it can deliver.
When budgets get tight, personalization might seem like a luxury that can be cut. Can you explain why personalization is a necessity that will pay for itself many times over?
If a personalization related hypothesis is tested and the results are positive, that is reason enough not to deprioritize personalization efforts. But suppose an experiment or A/B test shows personalization has a negative impact.
In that case, it should either not be pursued further or a follow-up test should be run, to verify whether the implementation was the best or not. Businesses should experiment far and wide to gain experience so that they get practice, and eventually be capable of knowing when to pull the plug on an idea, a program, a process or anything else.
Siddhartha Kathpalia (email@example.com)
Product Marketer, VWO
Siddhartha Kathpalia is a Product Marketer at VWO. He has almost ten years of experience and has worked with various organizations (both B2B and B2C) across different marketing functions. He is also a drummer and plays for a progressive rock band called Time Throttle.