Introduction to Product Optimization
Webster’s Dictionary defines Product Optimization as… wait, no, it’s not in the dictionary yet. Product Optimization is a relatively new discipline as it relates to ecommerce, and so we’ll take liberty of quoting Google’s Featured Snippet on the matter:
Production optimization is the practice of making changes or adjustments to a product to make it more desirable.
For a basic definition, this is certainly workable — but what does it mean in practice? How do we decide which changes and adjustments can safely be made to increase, rather than decrease, our Average Order Value (AOV)? Rocking the boat on a shop that seems to be selling at a stable rate can be a little nerve-racking.
The answer is data-driven insights. Data-driven insights are revelations made about consumer habits and product performance derived through comprehensive monitoring and processing of a whole lot of data points — like the actions of everyone who visits your shop or, more specifically, everyone who interacts with the item you intend to optimize. Some data points are absolute and concrete — like time-on-site stats—while others require more synthesis to construct meaningful conclusions about how inventory can be optimized.
Some of those data points might include:
- Time spent on a page
- Origin of the visitor (another page on the site, or another site on the internet, or perhaps even an app or ad)
- Number of times an item was added to a cart
- Number of times an item was actually purchased
- Which product variations are most frequently interacted with
This is a small sample of concrete, high-volume data points that, when averaged and observed, can lead to strong indications of where optimizations can be made.
Testing your items with the DAR.WIN Product Highlight
The brief list above outlines some statistics that are always available for observations — but what about generating more direct data about a specific item? Setting up tests can be crucial to improving your Product Optimization insights.
In short, DAR.WIN is a tool that can create personalized product recommendations for ecommerce shoppers using data-driven insights.
We discussed the value of A/B testing and its big brother Multivariate Testing a bit more in a previous article. This kind of testing is fantastic when isolating variables in your marketing to determine what kind of messaging and product presentation gets your desired outcome most effectively. In this article, we’re looking at how to conduct additional tests within your own store using the DAR.WIN Product Highlight feature.
In short, DAR.WIN is a tool that can create personalized product recommendations for ecommerce shoppers using data-driven insights.
When a shopper visits a DAR.WIN store, they see product recommendations throughout the site that automatically adjust, and optimize, to fit their pattern of behavior more accurately. The products being recommended are automatically selected and ranked by DAR.WIN based on previous shopper interactions to increase the likelihood that a shopper will find a perfect item, add it to the cart, and check out.
The products being recommended are automatically selected and ranked by DAR.WIN based on previous shopper interactions to increase the likelihood that a shopper will find a perfect item, add it to the cart, and check out.
For the sake of this topic, we’ll refer to these recommendations as algorithmic recommendations — meaning they’ve been automatically selected and promoted by a DAR.WIN algorithm instead of being handpicked by a shop owner.
In addition to their inherent dependability (big data crunched by machine learning = smarty-pants insights with reliable outcomes), algorithmic recommendations make a great backboard for Product Optimization tests in your store. While A/B testing actual site pages and product listings can be much more cumbersome than A/B testing ads, on-site Product Optimization can be done quickly using DAR.WIN. Here’s how.
1. Define your test
Product Optimization is best done on a case-by-case basis to eliminate variables. Select a product from your inventory to optimize. The optimization we’re ultimately looking for is increased interaction with a product (as compared to interactions without the test conditions in place). This will help us spend less money on marketing by targeting more specifically interested shoppers, in addition to increasing the overall number of item purchases.
This is our example store: DAR.WIN Gifts.
And this is the product we’re testing: Hot Toddy Carry-on Cocktail Set
This product makes for a good test because we can see how it’s appealing to several profiles of our typical visitors (for more information on how DAR.WIN profiles visitors, check this out).
Our test will be to see whether this item sells better with shoppers we’ve profiled as Wine & Spirit Enthusiasts or shoppers we’ve profiled as Travelers.
2. Use the DAR.WIN Product Highlight
We framed algorithmic recommendations earlier. The way the DAR.WIN Product Highlight works is by allowing shop owners to hand-select products to place in the recommendation bar that shows up for each unique Visitor Profile along-side algorithmically selected products. We’ll call these highlight recommendations.
…a brief period of artificial promotion can help us conduct tests and achieve short-term goals.
The benefit of including highlight recommendations — products that aren’t from our stable of reliably selected algorithmic recommendations — is that a brief period of artificial promotion can help us conduct tests and achieve short-term goals. Short-term goals include things like clearing overstock inventory or promoting sale items. In our case, we’re looking to see how an arbitrarily selected product performs in relation to those produced by data crunching.
Here, we’ve chosen the Hot Toddy Carry-On Cocktail Kit as the highlight recommendation for our Travel Visitor Profile. We’ll run this highlight for a week and then run the same highlight, but for our Wine & Spirits Enthusiasts. It’s fine to run both in tandem but for our purposes, this will keep the results cleaner with less room for ambiguity about where traffic is coming from.
Now, any time a visitor behaves in a way that qualifies them as a Traveler, our Hot Toddy kit will show up as a highlight recommendation along side DAR.WIN’s suite of algorithmic recommendations. Let’s make sure:
OK, Travel Visitors are now being treated to a recommendation set that includes both highlight recommendations and algorithmic recommendations. We’re going to let that run for a little while and check back in next week to see what sort of performance this product is getting from these users.
3. Optimize your product meta and targeting (preview)
Technically, we don’t want to do this until our tests have been completed, but lets preview of our next steps. Assuming that our tests reveal that the Toddy Kit is more heavily interacted with by our Traveler Visitors, what should we do to optimize based on that information?
- Update your product meta and on-site framing of your target
Travelers are into our Hot Toddy kit, so lets start using keywords and marketing phrases on the page that really invokes that notion — “Perfect for your next trip”, “The ideal gift for friends with full passports”, “Jetsetting was never so delicious”. In addition to any SEO juice the utilization of these keywords precipitates, on-page copy helps clarify the value proposition of a product for shoppers who clicked through an ad (and helps you keep your Quality Score high in PPC campaigns). In addition, if you’re running a blog or a write-up series on your store, an entry about this product in the context of travel could be a great way to attract more eyes.
Where you may have intermittent general email blasts, you risk losing subscribers or interest each time one of those emails lands in an inbox and fails to generate any excitement or curiosity in your shop. Personalized messaging through segmentation is the answer.
2. Update your ad targeting
Ad platforms allow you to select target audiences whom you assume will be most responsive to your products. The upside is that there are millions and millions of people to advertise to when your audience is broadly defined. The downside, though, is that you’re paying a considerable amount of money in clicks or impressions to some people who aren’t as likely to purchase an advertised item than others in the group. Now that we know Travelers are a better audience than Drinkers for our ads, we can create a specific segment within our larger audience defined by their interest in Travel. Refining the audience for advertisements featuring our Toddy Kit will allow us to spend less money on more specific ads to a more responsive audience. In theory, this will increase our click-thru and conversion rates while driving down the cost-per-conversion and increasing our margin on ad-spend.
*Note: DAR.WIN has an exceptional feature called our Customer Exporter that allows you to export all emails associated with shoppers in your unique visitor profiles. Platforms that allow you to use emails to create lookalike audiences can leverage this list of emails to open your audience to a wider set of like-minded potential shoppers.
3. Update your segmented email messaging (and product embeds)
Just like the platforms outlined in #2, your email list can — and should — be as heavily segmented as your best information will allow. Where you may have intermittent general email blasts, you risk losing subscribers or interest each time one of those emails lands in an inbox and fails to generate any excitement or curiosity in your shop.
DAR.WIN users can embed the product recommendations for a visitor profile directly into any email.
Personalized messaging through segmentation is the answer. Using the Customer Exporter, DAR.WIN users can create highly-specific email lists within their subscriber-base to target with specific promotions and messaging (like, in this case, Travel deals). Furthermore, DAR.WIN users can embed the product recommendations for a visitor profile (including the product selected by the Product Highlight feature) directly into that email! Talk about synergy!
Wrapping up Part 1
We’ll check back in on our experiment soon. For now, we’ve:
- Learned about the basics of Product Optimization
- Differentiated algorithmic recommendations from highlight recommendations and defined the value for both
- Set up an experiment we can watch over time
- Made a plan on how to update our strategy to bring in more $$$ using personalization, targeting and Product Optimization.
When we return, we can explore:
- Comparing test results to identify where are best margins are
- Applying test insights into everything from product copy to dynamic pricing
- Test contaminants that make our data less valuable (and how to avoid them)
