Meanwhile, discover strategies to boost conversion rates, transaction size, and customer engagement in our Idea Guide.
We've said it before and we'll say it again - personalization is here to stay. We could cite a million reasons why this is the case (and in fact, we did right here), but the main thing to note is if you're not investing in personalization, you will eventually be left behind.
If you feel like you're scrambling to catch up, you're not alone. 96% of retailers have encountered some challenges to their personalization efforts, including IT bandwidth, identifying the right partner, and difficulties aligning internal teams. Yet 74% think personalization should be a bigger priority in their organization.
One surefire way to generate more interest (and investment) in your ecommerce personalization efforts? Prove out the ROI.
In this guide, we'll help you understand:
Before we talk about how to measure ecommerce personalization ROI, let's get down our definition of what is ecommerce personalization. Personalization in ecommerce can be defined as any enhancement to your shopping experience based on a customer's actions, behavior, or intent.
Ecommerce personalization can be driven by:
Virtually any element of your ecommerce experience can be tailored to a customer's individual shopping preferences, but here are some of the more common areas that brands provide personalization in:
Despite the wide range of places and ways to integrate personalization into your ecommerce experience, the strategies to assess ROI are pretty universal.
The good news is, there are lots of ways to determine the ROI of your ecommerce personalization tactics. Before we dive into those, however, let's talk about measuring.
Specifically, the easiest way to understand whether your personalization efforts are paying off is through testing.
There are a few different ways to approach ecommerce personalization testing, but in general, all of them are variations of A/B testing.
One simple strategy for understanding the impact of personalization efforts is to look at how things perform before and after implementing a new personalization technique. Depending on what you're testing out, you might see a significant improvement immediately after installing a product or setting up a new test. If you're going to go this route, it's important to have metrics you can track against for the previous 14-30 days, at a minimum.
However, there is a major drawback to this type of testing: Seasonality. Virtually every business has seasonal factors that affect their sales. As an extreme example, comparing December sales to January sales often shows a stark drop-off in revenue.
But there are much smaller-scale ways your data can be skewed from month to month or even every two weeks. For example, in the first two weeks of one month, you might have an influencer mention your product in a post, or you might receive a restock of a hotly anticipated product. The following two weeks your sales might not be quite as high, even if you implement personalization in that time period. This doesn't mean your personalization isn't working, however.
If you are going to do before and after testing, it's recommended you look at things across a much longer timeframe - possibly even annually - and factor in any unique PR or product events that happened in that same time period that could have led to an inconsistent sales spike.
The most common way to measure the success of your personalization strategy is to run an A/B test with a control group, and an experiment group. The control group will receive the exact same 'vanilla' experience as your site's current default, while the experiment group will receive an experience with some form of personalization. Your customers are automatically and randomly sorted into either group, and you'll get to directly see whether those with the personalized experience are seeing an uptick in conversions, order value, or some other metric.
Again, you'll want to make sure you run this type of test over a statistically significant time period for your website, typically based on the amount of traffic you receive. A high-traffic site might only need a few days to a week to be able to measure whether something is working, while an emerging brand might need to run a test for 3 to 4 weeks.
A 'with' and 'without' test is one standard type of A/B test. But you can also run an A/B test where you run two simultaneous experiments to see which one is performing better.
For example, you could experiment with placing 'Bestseller' product recommendations on your home page versus 'Trending' product recommendations in the same spot. Your A/B test will show which type of product recommendations are driving more clicks, conversions, and higher order values. In this case you're not measuring whether personalization works, but what type of personalization works best.
It's important to note you can run multiple A/B tests for separate things at once, but this may end up skewing your results. If you're interested in testing multiple elements of an experiment, you may want to consider A/B/n testing instead.
Standard A/B testing involves comparing one experience to another. A/B/n testing allows you to stack multiple tests into one experiment - the 'n' stands for an infinite 'number' of tests. Generally speaking each variation should have just one different factor from your base or vanilla experience, or an equal number of differentiating factors.
Using our home page product recommendation 'type' experiment above, another thing that could be tested is the placement of the boxes. You could even layer in with / without testing. An experiment using all of these parameters might include:
Through this test, you should be able to determine:
Once you've determined what type of personalization tactic you're going to test, and how you're going to set up your experiment, the final step is to understand what exactly you're measuring. In other words, what are your KPIs (key performance indicators)? How will you know when one experiment is outperforming the other?
Here are some of the main ways of measuring the success of your experiments.
The most basic way of understanding whether your personalization efforts are working is whether you see an uptick in purchases, period. When customers start using LimeSpot's personalization suite, we typically see a 2-5x conversion rate increase through product recommendations.
It's important to recognize that conversion rates are a relatively macro metric, meaning that a lot of factors may go into whether you're driving more conversions or not. As such, conversion rate is a metric to pay attention to over a longer period of time.
Some types of personalization - like customizing your home page to a customer segment - may not be as obvious at driving conversion rate increases in a short period of time. But if you look at year-over-year, or even quarter-over-quarter measurements from when you first implemented this type of personalization, you should see a positive trend line.
There are two ways to track conversions. The first is conversion rate increase (expressed as a percentage or 'x-rate'), while the second is incremental conversions (expressed as a number). Average conversion rates vary wildly year over year across the entire ecommerce sphere, but ballpark around 2% (+/- 1%). Post-personalization, a brand might expect to see a minimum conversion rate increase of 2% (or 2x) for a total conversion rate of 4%.
If we were measuring incremental conversions, you would look at a site's traffic. Let's say this particular site has 300,000 visitors per month, with 6,000 shoppers converting (2%). Their incremental conversion post-personalization would likely be an additional 6,000 conversions, for a total of 12,000 transactions.
The other most common personalization metric is average order value (AOV). In other words, are customers not just shopping more, but spending more per order?
One of the perks of measuring average order value is you're typically able to see results in a shorter time period. Adding a cross-sell to the cart? It should be pretty clear in a basic A/B testing scenario whether you're driving more revenue on average with this experiment.
Keep in mind with average order value, you can measure the actual percentage increase, or the dollar value. For example, let's say a company whose average order value was $50 saw their AOV go up to $60 after the introduction of personalization. Their AOV increase could be reflected as $10 per order, or as 20%.
AOV is a macro metric. You should also measure the effectiveness of specific offers or approaches, and consider different experiments. Above, we mentioned how offering a cross-sell in the cart should give you a pretty clear indication whether it's boosting AOV. But with this one experiment, you can actually consider dozens of spin-off experiments and questions:
It's important to look at each individual offer to get a full view of where there might be missed opportunities to continue to drive up conversions and AOV across the entire site. Digging into this type of measurement may also be referenced as 'sales from clicks'; in other words, which specific actions are leading to which specific numbers?
At first glance, average order value and incremental revenue might seem similar. But AOV measures how much more you're making per order, while incremental revenue identifies how much more total revenue you're making over a control group. Need a benchmark? The average LimeSpot customer sees a revenue lift of 20-28% after introducing personalization.
This type of measurement is again ideal for tracking over a longer period of time, particularly for businesses with sales inflections based on collection releases, seasonality, or promotional considerations.
In short though, if personalization is driving more conversions and / or higher average order value, you should be making more money, period. Incremental revenue measures just how much. A company making $3M per year might see a lift of $750,000 after implementing personalization, should they realize an overall revenue lift of 25%.
Calculating revenue per visitor (RPV) is a combination of both AOV and conversions, but it can also be viewed as a subset of incremental revenue. Essentially you take the total revenue and divide it by the total unique visitors. Or to think of it another way: RPV = (AOV x conversions)/total unique visitors.
A store that has 600,000 visitors in a month and generates $7M in sales would have an RPV of $11.66. Personalization should see this number trend up, as personalization typically drives bigger basket sizes (AOV) and more conversions, positively impacting RPV values.
Most of the time, marketers are optimizing their ecommerce experience to drive more revenue - which makes perfect sense. But softer metrics like time on site and bounce rate can be directly impacted by your personalization efforts, particularly with things like tailoring home page content.
The idea is that the more a customer feels like they're in the right place, the more likely they are to stay on your site. More time on site equals more clicks - personalized product recommendations often help shoppers find their 'next click', as another element of personalization. And more clicks typically means more sales.
The long and the short of it is, if you're not seeing revenue-based results with your personalization strategy, take a look at time on site and bounce rate metrics from before and after implementing a personalization experiment. It could be you've found a way to highly engage customers, but something else is keeping them from converting. Consider looking at your price point, shipping, return policy, or promotions.
Personalization is about much more than just your on-site experience. How your ads are personalized will also have a major impact on your click-through rate (CTR), conversion rate, and return on ad spend (ROAS). With LimeSpot's Shopping Ads for Google product, we enrich listings with valuable data points to improve the likelihood of showing up for free, as well as showing up for the right audience at the right time. The results are ROAS improvements of anywhere from 2-26x. Reducing your customer acquisition costs (CAC) by boosting your ROAS is a clear signifier that personalization is working.
Again, personalization can (and should) be applied to well beyond your main website experience. Embedding 1:1 product recommendations into emails is just one example of an easy way to test whether personalization is working. With customers that receive personalized product recommendations, you should be able to measure increases across:
The same metrics could also be considered for customers that are opted-in to receive texts and / or push notifications. For example, let's say you're offering a store-wide sale of 25%. You could run an A/B test where the control group receives a blanket notification you're having a sale, while another group's notification spotlights the specific brand or product category they're most loyal to within your store (e.g. '25% off Nike' or '25% off camera equipment'). Again, it should be pretty clear whether you're seeing more clicks and conversions from the group with personalization.
The other metrics listed in this guide are all focused on seeing gains and improvements across common retail KPIs. But it's also worth assessing the ROI of whatever personalization tools you're using.
9 out of 10 marketers report returns of at least $1-$2 for every dollar spent on personalization, with 43% receiving at least $6 in return per dollar spent. Calculating ROI should be simple, although incremental revenue is one of the more popular approaches. Simply calculate how much net revenue you've generated through a tool, then divide it by the tool's cost and multiply by 100. A product with a monthly subscription cost of $150 that generates $7,000 in incremental revenue has an ROI of 47x.
If your personalization tools aren't delivering the same results, it may be time to look for a new supplier.
You've probably heard more than a few marketers complain about 'not achieving statistical significance' with their experiments. For example, let's say you tested placing product recommendations right below a product description (Version A) versus at the bottom of a product detail page (Version B). In Version A, the brand might see a conversion rate of 3.18%. In Version B, they might see that number dip to 3.15%. Is 0.03% worth optimizing toward?
Before we answer that question, the most important thing to recognize is the rate of what is considered significant varies wildly. A 0.5% increase might be nothing to one business, while it could be monumental to a behemoth like Amazon.
Secondly, you need to look at whether adopting a practice has any other drawbacks; for example, are you seeing any other metrics trending down by placing your boxes below the product description? Have you received any customer feedback about where they're placed?
Finally, it's worth considering volume. It might be worth it to run your experiment for a longer period of time than you initially planned to. Longer time periods equal more exposure. More exposure equals a bigger sample size to see if that gap grows or closes up.
There's also the consideration of users adapting new behaviors over time. A shopper might get accustomed to seeing product recommendations at the bottom of the page, and over time, that gap might close up as customers get conditioned to simply scroll down when they're hunting down more product recommendations.
Now back to the question of whether it's worth it to pursue incremental gains. If you've gone through the considerations above, the last thing to do is some modeling. For a super high volume store, a 0.03% increase could equal major revenue. For a smaller business that's still just working to provide a better shopping experience, it's likely not worth it to change up what you're doing. In other words? Onto the next experiment!
If you learned one thing from this guide, it's that there's certainly more than one way to implement personalization on your site, just as there's more than one way to measure its value. Your key takeaways and lessons from here?
