Product data is the backbone of every fantastic e-commerce customer experience. Whether you're a seasoned e-commerce executive or just stepping into the world of online retail, mastering the art of product data tagging is essential for attracting customers, enhancing discoverability, and driving conversions. This comprehensive guide to truly effective product tagging will delve into its intricate nuances, from understanding the importance of attributes and properties, categories, and keywords, to optimizing product descriptions and leveraging advanced tagging strategies. By the end of this guide, you'll be equipped with the knowledge and strategies needed to elevate product tagging so that your customers will discover and purchase the items from your store that they love more quickly.
Product tagging in e-commerce means attaching descriptive labels, attributes, properties, and keywords to products, making them easily searchable and categorized. These tags include elements like size, color, and brand, helping customers find what they're looking for. Categories and subcategories organize products, while keywords aid in search, recommendations, filters, and SEO. Tags can also convey pricing, discounts, availability, and compatibility information. It’s virtually limitless what kind of product tags you can apply to a given item or product line; it all really depends on your industry, product types and variants, and how buyers search for what you’re selling. Product tagging streamlines the shopping experience by simplifying product discovery and providing essential details at a glance.
While all of these types of product tagging would ideally be standardized across an entire store’s catalog, there are frequently instances in which they aren’t. A lack of internal cohesion around how to categorize items – whether it be by category and type or spelling and classification of those tags – can lead to inconsistent tagging of items that would otherwise be considered the ideal search result from a customer query.
Additionally, storefronts that source their products from multiple businesses or manufacturers may run into a similar issue where product tags are inconsistent. In the absence of tedious manual data analysis by merchandisers – cumbersome work, to say the least – untouched product tagging can lead to disparate descriptions of items that should be more closely aligned with other products that a customer with those specific qualifications in mind would find appealing.
In e-commerce environments, product tagging is the key element enabling merchandisers to display the appropriate items at the right time to the right people. Similar to the information a visual merchandiser in a brick-and-mortar store gains from viewing and physically interacting with products, product tags give e-commerce retailers the insights they need to understand where and when to serve product recommendations to online shoppers.
Merchandising, whether online or in-store, is designed to anticipate the tastes and needs of customers. To a certain extent, this can be done by analyzing broader trends within a given store’s industry and reacting accordingly. However, in-store retailers and their sales associates have the added benefit of analyzing live customers in person – watching and interpreting their interactions with products as well as asking questions about their preferences.
For e-commerce retailers, the absence of those in-store benefits must be taken into account by putting significant effort into product tagging that will match its customers’ interests and, more bluntly, keyword searches. This allows the online shopping experience to be as nimble and sometimes more so than it would be in person by constantly evaluating both customer behavior and product data to put the right products at the forefront in a timely manner. However, more often than not, the ability to achieve this goal is easier said than done.
Both retailers and e-commerce customers have experienced the pain of poor product tagging and data hygiene. When data is not clean, standardized, and tagged correctly, customers’ expectations aren’t met. This leads to confusion, frustration, website bounces, and fewer conversions. And that’s never the goal.
Yet we still repeatedly run into companies making mistakes that are preventable. These are five examples of poor user experiences that can - and should - be prevented with better data hygiene and enrichment practices.
In an ideal world, you can land on an e-commerce site and tell it exactly what you want. However, shoppers often don’t know exactly what they want, so they need to perform “inspirational searches.”
Inspirational searches often contain descriptors as well as a category name. For example, “fun socks” is a subjective inspirational search. On most e-commerce sites, the search results for “fun socks” would simply return all socks. This isn’t ideal, because the user already has something vaguely in mind – not necessarily concrete, but less vague than the entire scope of products available.
In a traditional brick-and-mortar environment, a sales associate would be more likely to contextually select items that match this description and recommend the products accordingly to the customer. On an online store, this is not necessarily achievable in real-time. This is where an online merchandiser’s job comes into play. Although tagging certain socks as “fun” is subjective and challenging, it’s extremely valuable to that user.
To mitigate the time restraints and uncertainty concerns around online merchandising, e-commerce teams use AI like Crossing Minds Data Enrichment to do this automatically. It adds product tags that are derived from the existing image, title, and description data, ultimately enabling more inspirational searches. Paired with AI that analyzes and accounts for customers’ behavior, it’s a powerful approach to online merchandising that can power more conversions.
When customers are choosing between products, they often use comparison charts to make the best decision. But when product data is inconsistent or incomplete, the experience renders product comparison frustrating and inaccurate.
Without comprehensive and accurate information, users are left grappling with a myriad of problems. Incomplete data often leads to confusion, as users are unable to make fair comparisons between products. Key specifications may be missing or inconsistent, making it nearly impossible to determine if a particular product meets their unique needs or tastes. Inconsistencies across product listings further compound the issue, leaving users uncertain about the reliability of the information provided. This lack of trust erodes confidence in the retailer, potentially leading to lost sales and a damaged reputation. Worst of all, users may end up making uninformed purchase decisions that force them to return products due to unmet expectations. All of these factors negatively impact conversions as well as customer loyalty and retention.
Pounds or kilos?
H x W x D or W x H x D?
Inches or centimeters?
Motion rate or Hz?
40 inches or 40”?
If a customer is shopping for a TV, they need the e-commerce site to understand their criteria. Yet if they have to select various checkboxes that really mean the same thing, they run the risk of not discovering the full range of options.
For example, if a customer wants a TV with 120 HZ, it’s ideal to make one selection and be done. Requiring “120 HZ” and “120 motion rate” is redundant. A poor experience.
In many industries, products sold may be supplemental to a customer’s previous purchases. For example, a customer might need a replacement AC adapter for their wireless speaker. While searching for after-market replacements, the most critical consideration is whether or not the AC adapter is compatible with the brand and model of the wireless speaker. If these product data tags or incorrect, non-standardized, or missing altogether, the product recommendation is rendered useless to the consumer. Similar to the improperly aligned comparison chart scenario, a customer will either forego the purchase altogether or become upset when purchasing something and realizing when it arrives that it isn’t compatible.
Poorly tagged product data can lead to mismatched or irrelevant product descriptions. For instance, if a product image shows a complete set of dining chairs but the product description and tags refer to a single chair, users might be confused about what they're actually purchasing. Similarly, if the dining chairs come in different colors, but these are not accurately or uniformly tagged, it can create further confusion. This misalignment can erode trust and lead to unsatisfactory purchases.
If properties are missing in general, this renders delivering relevant results to specific user queries virtually impossible. Say for instance a customer is searching for a pillow with a fuzzy texture. The store may have multiple products that fit that description; however, if that property value is left or blank or doesn’t exist at all, it would be extremely difficult for the site to surface those items in a timely manner.
Appropriate product tagging is crucial when it comes to the homepage and PLP. Remember when we talked about “inspirational searches”? Sometimes a customer knows that they want something in particular, but aren’t completely sure about all of its specifics just yet. This might lead to a search query that a typical search engine may find vague – in this example, “patterned” Birkenstock sandals. While the PLP does have filtering options on the left-hand side, and the search does surface patterned products to view, it unfortunately displays sandals that are not patterned as some of its top results. This can frustrate customers.
The cleaner your store’s data is, the easier it is to deploy recommendations to customers that are highly personalized to their behavior and preferences. An AI platform like Crossing Minds is able to analyze both product and customer behavior enables these extremely satisfying customer experiences. By matching a user’s tastes and movement across the site with all of the product data that makes items unique, surfacing items they’ll love is achieved sooner, driving more sales.
The hygiene of e-commerce product tagging is an excellent opportunity to tap into customers who either haven’t visited your site directly or perhaps have never even heard of you before. If a customer has a need – say, a ruffled green sundress – but isn’t quite sure where to look, they might start on Google. If your products are properly tagged and have lots of rich attribute data, it increases the likelihood of them discovering your product and store organically and making a purchase.
When using an AI platform to achieve your product tagging goals, it can also be leveraged to improve your SEO, and thus, organic discovery as mentioned above. Machine learning can analyze the search terms customers use most to surface your products, and update its properties and attributes accordingly so they match the most accurate and frequent search queries. This will improve the customer experience on-site as well as boost organic store discovery.
Sometimes, a store will cater to multiple different audiences, and in an ideal world, certain subsets of that audience would only be served certain product recommendations or have others excluded as results.
With clean product tagging, this task becomes infinitely easier even in the absence of third-party data or manual filtering by a customer. Based on the customer’s behavior on a site, for example, an AI platform may be able to estimate that the person searching is a child or tween. Paired with accurately tagged, data-rich products, the child’s browsing experience would naturally limit the discovery of items with mature themes or those geared towards adults.
Comparison charts are a tool customers love to help them make informed decisions, especially in stores that carry a wide variety of brands or similar items. In the absence of clean or complete product data, this tool becomes deficient, however. In the above example, one brand of vacuum has a wide range of descriptors for floor compatibility when it could be summarized as “All floors” as the other vacuums are. Meanwhile, a different vacuum is lacking concrete data for at least two of the specs.
Did product descriptions come straight from the manufacturer? Aside from the keywords they need to include, do they reflect the tone and voice you want to be making public? It may seem easier to just leave them as is, but a jarring difference from the rest of your brand identity can confuse or turn your customers off.
The above screenshot demonstrates the consequences of duplicates. Not only is it a bad customer experience to show the same product in different colors when there’s already a color selector bar, but these dupes create terrible analytics on the performance of the product.
An AI platform for data enrichment sweeps the store for duplicates like this so customers can see more different results quicker, and limits confusion on product performance.
Getting product data into Google Ads, Meta Ads, and other digital advertising avenues can be challenging when the data is disorganized. Cleaning and standardizing that data is extremely important.
Products like Feedonomics, a data feed analysis tool, help make the product data ready for those platforms by applying the right taxonomy and so forth. However, you still need data that’s standardized: the same format, free of duplicates, the same units of measurement – all of the things we’ve chatted about in this guide. Clean, organized product data will make this whole process easier. It’s a necessary evil to make running ads wortwhile, but it doesn’t have to be a manual nightmare if you have an AI e-commerce tool that can do the work for you.
Understanding the successes or failures of any given product or item can be tricky if there are so many variables. Was the dress popular because it was green, had ruffles, was a maxi, or some combination of those details? It becomes even more difficult when the properties used to categorize and tag these products are inconsistent across the store. Having uniform product tagging conventions and completely filled out properties will make this work much simpler to analyze, especially if you’re using an AI e-commerce tool like Crossing Minds’ Data Enrichment solution.
Data enrichment is comprised of cleaning product tags as well as adding additional product properties and attributes that can be used to further enhance a product’s tags, thus increasing the chances of surfacing relevant items to customers sooner. These two processes alone can be done manually by a highly competent e-commerce team, but it would be extremely time-consuming and costly, not to mention tedious.
An AI platform like Crossing Minds is able to achieve data enrichment using complex machine learning principles that can identify errors and gaps in data, as well as analyze a product’s images and other data to make improvements and additions to its product tags. Internally, this frees up the need for manual work on the part of e-commerce teams or contractors, leading to improved productivity and cost savings. Externally, it leads to improved customer experiences that can boost conversions.
Data feed analysis is focused on making product data usable and optimized for external channels, such as ad platforms, marketplaces, and so forth. This is of great benefit to e-commerce teams, but it doesn’t address issues with the product data and tags on its own site.
Data enrichment is focused on generating new data to optimize search, recommendations, and PLP filters onsite. In the example of someone looking to purchase a lamp, furniture and decor shoppers often want to browse by style. An e-commerce store would need relevant style tags, like “Minimalist,” “Boho,” and so forth consistent across an entire product catalog.
When merchants purchase products to sell on their sites and receive product data from the supplier or manufacturer, it’s often inconsistent between brands, or missing altogether. Data enrichment is the 1:1 analysis done with merchants to identify which properties will be the most valuable to add to their specific site.
The best way to intuit which types of data would enhance your products and the overall customer experience is to think like your customer. When your customers enter your site for the first time, how would they describe what they’re looking for? Think less about how you or your team refers to items internally – imagine yourself as a non-employee typing a search query. Your data should reflect those words.
If you are a storefront that only sells one type of item, you need really good data that differentiates those things from one another so that people can quickly discover what they’re looking for instead of getting an unbearable amount of results that are impossible to parse through.
Let’s use socks as an example. What are the socks’ texture? How thick are they? Are they knee-high, crew or ankle? Are they insulated? What is the material? Do they have rubberized grips? With all of these properties properly and uniformly defined, it suddenly becomes much easier for a customer to find a needle in a haystack.
What features or attributes help your customers shop? Are those features reflected in your product data? If so, are they consistent (“Blue” for all blue items instead of “Blue,” and “BL,” for example)? If you have any doubt whatsoever, you need to evaluate your product tagging data immediately. Crossing Minds is here to help – and for a limited time, we’ll do it for free. Get in touch with us today.