POST CATEGORIES
September 21, 2018
Which e-commerce site search platform can help your retail business drive more online sales conversions?
What’s the best e-commerce site search platform to help your retail business drive conversions?
Good question. And an important one. Here’s why.
Boosting online sales conversions continue to be a challenge for retailers. While rates vary by device and by country, benchmark studies from Monetate Ecommerce Quarterly indicate that, heading into 2018, average conversion rates continue to hover around 3%.
And so a more immediate, more fundamental question is this one:
What can companies do to increase conversions?
Have you tried enhancing your site search experience? It’s an underrated improvement that can have a direct and dramatic impact on conversions. Consider the following:
- Up to 30% of e-commerce site visitors use the search box.
- Sites with an effectively optimized search box convert at a rate 2 to 5 times higher.
Think what a difference improved site search can mean to your business.
Of course, effective site search is more than just a clear and well-used search box. The customer needs their on-site search to deliver relevant results.
Relevance is a vital part of a personalized experience because 90% of customers who use search don’t go past the first page of results. Users will often abandon their search altogether if they can’t find information quickly.
And when it comes to search, many of the larger online businesses think Endeca and BloomReach, among others.
Endeca (which became Oracle Endeca Guided Search after Endeca was acquired by Oracle in 2011, then simply became known as Oracle Commerce) and BloomReach Search are among the options that large midmarket and enterprise retailers turn to for site search technology.
Each platform takes a unique approach to how the site search and browse experience is managed and controlled.
This article will serve as a guide to help large midmarket and enterprise retailers decide which approach may be best for their business.
First, a few words about Endeca
Endeca is a keyword-driven, on-premise search platform.
It has configurable and extensible dynamic merchandising abilities that put more power in the hands of business users.
The Experience Manager component allows the business user to configure more intelligent customer experiences through product spotlighting, personalization, and segmentation, using a company’s existing product content.
This technology requires a hands-on approach for support and maintenance, to ensure the software performs to its full capacity.
There are a number of features that can be configured to provide customers with a tailored shopping experience in terms of the content they see, and the context in which it’s presented.
How search relevancy tuning is handled with Endeca
As outlined in RealDecoy’s Search Tuning blog post, search relevancy tuning is a manual task that should be completed regularly, at least every 6 months or after every meaningful catalog change, whichever is more frequent.
Endeca has a sophisticated system that allows for granular control of search result relevancy, with different modules that can be combined and re-arranged as needed.
The process of search tuning in Endeca involves development work as well as business team involvement. The business team evaluates and scores the current search performance based on what it believes a relevant set of search results would look like for a given set of search terms.
A trained technical specialist will then update the Endeca relevancy strategy to obtain the highest scores. A relevancy strategy, in this context, refers to the strategy by which Endeca arranges search results in order to have the most relevant results appear at the top of the list. This is done multiple times to tune specific searches without affecting others.
The process requires the involvement of both business users and technical specialists. There are multiple steps in this process, which include:
- Business users gathering sets of search terms (usually the top search terms on the website).
- Business users reviewing the corresponding search results for each of the terms identified above, and scoring them based on relevancy (the higher the score the better the search relevancy).
- A technical specialist updating the Endeca relevancy strategy to increase the overall score, and business users re-evaluating once those changes have been made. This process usually repeats two or three times to get the best tuning score for search results.
- Business users working with the technical specialist to determine what value is being driven from search relevancy, and then considering how to improve the search results across the site and how to benefit from levering search features to assist with that.
How personalization is handled with Endeca
Personalization is achieved using the Experience Manager component.
Experience Manager is a separate component that can be purchased and implemented with Endeca. Companies can attain very detailed levels of personalization using Experience Manager, but they need a dedicated individual or team in place to support this level of customization.
The team must have a deep understanding of the target market segment: who the customers are, their motivations, behaviors, what they might be looking for, and what types of campaigns might draw them in.
Armed with this intelligence, a trained business user has granular control over page layouts and the components returned, based on their own pre-defined set of rules.
The business user can apply changes and manipulate configurations in Experience Manager as needed to deliver a personalized experience and targeted marketing for website visitors.
Now, a few words about BloomReach
BloomReach Personalization (which includes Search, Merchandising, and Insights) is a cloud-based solution that powers search, navigation, and personalization on ecommerce sites (desktop, mobile and apps). It provides customers with personalized search results, categories, and recommendations.
An additional feature is BloomReach’s ability to create a connection across all digital devices that a non logged-in customer might use throughout a search experience. This is accomplished by applying a series of machine learning algorithms, which gather data and assemble the results in a meaningful set that is ranked based on relevance to the customer’s purchase intent.
BloomReach’s Web Relevance Engine ranks results by applying metrics to potentially millions of products. The Engine then personalizes the results and rankings by learning from every interaction a customer has online and offline.
This technology allows a more hands-off approach, since much of the process of knowing your audience and tailoring your search platform to meet their needs is automated.
When manual adjustments need to be made, BloomReach provides business users with a Merchandising dashboard from which users can apply traditional merchandising rules such as curated results, anchored products, boost and bury, etc.
This technology requires a few weeks of learning for the initial pixel-based data collection to take place. Once the learning phase has been completed, much of the upkeep is automated.
How search relevancy is handled with BloomReach
Search relevancy tuning is primarily automated by BloomReach’s Web Relevance Engine.
The Web Relevance Engine is constantly learning in the background: picking up on new trends, data, and actively listening to the world both on and off-site. This enables it to continually tune its results and improve how those results are being returned to customers.
A business user has full transparency into how the search relevancy is determined and has control to make exceptions when needed.
How personalization is handled with BloomReach
As part of its 1:1 personalization framework, BloomReach is able to understand users’ long-tail search queries, synonyms, and concepts through its semantic understanding engine. BloomReach can also personalize results to individual users based on identifying product affinities, browse behavior, attribute patterns, etc. BloomReach actively evaluates each user’s site journey to provide personalized results.
Cookies and javascript pixels—tracking code that is added to the ecommerce site—are used to examine customers’ cross-device behavior and build up unique customer profiles. This includes browsing patterns, social media interactions, previous online and offline purchase history, searches, and personal affinities.
All this information is then coupled with contextual information about time of day and location to help serve up the best results for each customer.
A side-by-side comparison of Endeca and BloomReach
A detailed side-by-side comparison of Endeca vs. BloomReach is provided below. The comparison is based on criteria originally employed in RealDecoy’s E-commerce Search Platform Analysis for Retailers legacy blog post.
| ABILITY TO DRIVE CONVERSIONS | ENDECA (on-premise) | LUCIDWORKS FUSION |
|---|---|---|
| Levers your own data to drive relevancy
The degree to which the search platform is capable of levering data attributes to drive relevant search results
|
Technical specialists can manipulate the relevancy strategy being used, indicate what fields are searchable, and search other data points not in the index. Business users have a relevancy tool they can use to experiment with different relevancy strategies using data collected. | Technical teams are responsible for configuring the telemetry data that will be used in relevancy, for configuring searchable fields and the weight applied to fields for relevancy scoring on field matches.
The Fusion interface provides business users with the ability to conditionally boost and bury records, choose stages in the query process to ignore or include and establish business rules to impact relevance. Business users can also easily access analytics within Fusion to help drive decisions for improving relevancy strategies. |
| Business user control
The extent to which the search platform provides business users with the control they need to make updates and deliver timely, engaging campaigns |
Gives business users full control to drive a detailed and targeted search experience, but this requires expert guidance and initial training. | Business users can only affect search configurations for relevancy. A technical team is required for building targeted search experiences on the front end. |
| Ability to cater to unique personas
The extent to which the search platform enables business users to cater content to different personas |
Need a team or individual with a deep understanding of customers to identify and define different segments/groups. |
Segmentation is done by signals data, automating user to item, user to query, query to item and query to query similarities to present similar types of results (and auto-suggested queries) to similar types of customers.
|
| Speed of catalog refresh The speed with which the index can be rebuilt or updated |
Full catalog refreshes – required for updates to the spelling correction dictionary or changes to the faceted navigation – can take considerable time depending on the size of the catalog, or if a complex ingestion process is required. Updates to selected products or attributes are possible, but require manual programming and are limited in their capabilities. | Full catalog refreshes are required for updating Solr schema configurations and updates to spelling and auto-suggest dictionaries. Facet navigation does not require full catalog updates. Incremental updates are supported for both record modification and record additions. The time for update depends on the size of the catalog. However, this can be managed or reduced by distributing the index job among multiple nodes using Apache Spark and Zookeeper. |
| Data/index management features
The extent to which the search platform allows for incremental indexing and the auto-pruning of data without manual intervention |
Indexing is usually a scheduled job. Business users control the data in its source. Scheduled indexing will pull in updates from source data when run. | Indexing is scheduled and executed via Apache Spark. This supports both full and incremental updates. Data can be loaded/joined from existing collections in Solr. Backup and recovery also supported by data management/indexing. |
| Understanding customers
The extent to which the search platform offers comprehensive linguistic functionality, natural language processing, stop words, spelling correction, thesaurus, etc. |
Some manual intervention is required (by IT and business users) for phrases, thesaurus, stop words, and stemming. Natural language processing is not available. | Fusion supports both automated and manual development of dictionaries (synonyms and phrasing) and head-tail analysis of queries to improve search. Fusion server also supports natural language processing supported or named entities, sentence detection and part-of-speech tagging. |
| Reporting capability
The degree to which the search platform offers business users the ability to monitor customer search behavior |
Provides general reporting on top search terms, conversion rates, A/B testing, popular records, sorting options used, spell correction and alternate suggestions, business rules activated, dimension searches. Reporting requires an initial setup of the third-party analytic solution by IT. | Built in analytics tools to provide reporting capabilities in near real-time (from the Analytics Dashboards). Reports can be used to perform application log analytics, search analytics (which supports head-tail analysis), and A/B testing reports from experiments. Custom dashboards configured by technical teams. |
| Ease of search tuning
The ease with which the search platform can be tuned for optimal customer experience |
Manual effort with multiple steps, and a need to integrate with third-party analytics, reporting, and BI tools. | Search tuning can be done in real-time without the need to reindex the data set. Changes can be viewed immediately once applied. Leveraging signals from Fusion AI can assist with search relevancy improvement without needing extensive market research |
| ABILITY TO MEET BUSINESS USER NEEDS |
ENDECA (on-premise) | BLOOMREACH PERSONALIZATION |
|---|---|---|
| Total cost of ownership
A combination of license fee and the cost to operate, maintain and customize the search platform |
Per-query licensing that requires a specialized, trained team to customize the search platform. Third-party support typically required to implement and deploy. Employee costs to support and maintain platform. | Usage-based licensing model, with no on-premise software or hardware required. Initial IT costs needed for integration. |
| Personalization management
The extent to which the platform empowers business users to control search functionality and extend personalization |
Experience Manager, purchased separately, enables personalization. Templates allow business users to dynamically create pages with components, organized under various conditions (granular control). IT can create additional templates as needed to build custom pages. Users define category pages and build rule-based product recommendations. | Business users are armed with a dashboard which empowers them to augment the Web Relevance . Engine in search and category pages given their unique understanding of their business and trends. Customers can also incorporate personalized widgets such as “more like this,” “just for you,” or "best sellers" throughout the site. |
| Flexibility
The degree to which a search platform can be configured to business needs, and the level of effort involved to make changes |
Catalog + non-catalog content is available. Business users have granular control over the elements on pages and can define sets of rules on when those elements will appear. | Catalog + non-catalog content is available. Dashboard controls are provided for product grid curation in search results and category pages, based on various product attributes. |
| Data consumption
The extent to which a variety of data sources can be integrated, and the platform’s ability to overlay these data sources |
Manual: one-time set up by IT for process that gathers data and joins it together. Raw data can be imported from anywhere. |
Manual: A one-time manual integration effort is required, then automated using standardized feed processes and data formats. |
| Multilingual/multi-site
The ease with which the search platform can be managed and indexed across multiple languages, brand websites and/or product catalogs to display the most contextually relevant content |
Manual: IT must build separate indexes for each language, and/or each site (depending on business goals for multi-site). | Comprehensive multi-site framework to support multiple domains/brands/sites/countries, languages, currencies, and verticals. |
|
Support/maintenance The degree to which established lines of support are available |
Requires trained business users for Experience Manager, in-house IT trained on Endeca, and may occasionally need to outsource for specific implementations. 24/7 support is available. | 24/7 support is available. Premium services available for elevated levels of support. |
| Hardware footprint
The extent to which a search platform is linearly scalable without extra costs |
On-premise. Requires an infrastructure team to scale (an extra cost). | SaaS (Cloud). |
How to decide?
There are two key points that large midmarket or enterprise retailers must consider prior to selecting a search platform technology: control and team structure.
1. How much control does the business need over the search experience?
There is a spectrum between having complete manual control over the search experience and having the experience be automated by technology. Each technology that has been discussed here lies at a different point on that spectrum. Endeca lies close to the complete manual control side whereas BloomReach lies about three-quarters of the way to the automation side. There may be times when a business user wants control and other times when it wants the machine to take over (and many states in-between).
If the retailer needs to support both catalog and non-catalog content, and wants to control every aspect of the search experience with a detailed level of granularity, then Endeca may be a good choice.
Why Endeca?
Business users can create dynamic landing pages with elements that trigger under certain conditions. These elements can be set by time, keyword search, customer segment, and navigation selection. This allows retailers to employ more targeted cross-selling opportunities and catch customers in real time, to hit them with specific content to help boost average order values at checkout.
If the retailer needs to support both catalog and non-catalog content, and wants automation to do the heavy lifting with a business user in control, then BloomReach Personalization may be a good choice.
Why BloomReach?
Retailers can provide customers with personalized search results and category pages that are ranked based on their relevance to a customer’s need. The path to purchase is facilitated since the technology levers machine learning and natural language processing at every step of the customer journey. On top of all the automation, business users can create, curate and manage static pages, and add or edit dynamic category facets or dynamic badges to content.
2. What is the expertise and makeup of the team in charge?
Business users of these search platforms typically include merchandisers and marketers. These individuals are seldom trained to tune search relevancy for keyword-driven platforms, or configure personas, business rules, and page layouts for personalization. It takes time to not only train and develop these specific skills, but also to execute.
If the retailer has a business user or team dedicated to learning the admin platform, tools, and search analytics, then Endeca is a reliable choice.
Why Endeca?
The Experience Manager interface is complex, but it can be very powerful in the hands of a trained team or individual who can take full advantage of its functionality. This functionality allows business users to exercise granular control over page layout, search and navigation, merchandising, personalization, promotional content, and rich media, to name a few.
If the retailer does not want to devote full-time resources to manually manage the search experience on a full-time basis, then BloomReach Personalization is a reliable choice.
Why BloomReach?
The dashboard interface is intuitive for the non-technical business user. This means that, when needed, the business user can easily configure changes to the setup. Combine that with the power of machine-learning and natural language processing, and the retailer is left with a self-sufficient search platform.
The biggest consideration for top retailers isn’t if they’ll improve site search, but how.
Whether retailers are looking to improve search for the first time or are thinking about a wholesale change to a new search platform, they must determine if they have the capacity to manually control the search experience, or if they want to rely on technology to ensure search relevancy and personalization.
Ultimately, it comes down to where businesses lie on the spectrum of control for the search experience, and whether or not they have the resources in place to support that need. Either way, the best place to start is by examining their business goals. They must assess how (or if) these goals have changed since the last site search implementation.
What search platform choice is best for your organization?
At RealDecoy, we have the implementation experience and insight you need to make an informed decision about which search platform is best for your B2B or B2C business. We also understand what it will take to implement a solution, whatever it may be, so your ecommerce site is set up to drive ROI. Contact us to learn more.
Related Articles
Quality assurance is about more than just checking lines of code
Quality assurance is about more than just checking lines of code. It’s also about ensuring that the application executes every business requirement properly.
Read More
Amazon is bringing its "A-game" to B2B: How should you respond?
With Amazon set to pursue the B2B market, competition in the space is about to get fierce.
Read More
How to show faster return on your Hadoop cluster investment
Open-source framework Apache Hadoop has set a new standard for the distributed storage and processing of massive data sets.
Read More