Constructor’s benefits are only as good as our uptime and performance. For this reason, we have robust standards around both. During the holiday season and the peak demand period of Black Friday and Cyber Monday, we increase our standards out of recognition that it is the most important selling period for many of our customers.

Over the past year we worked continuously to drive even better performance and scalability, contributing to improved latencies and zero downtime.  We tested scaling to 2000% of current average daily traffic volume.  We used chaos testing to validate that catastrophic failure of the following supporting infrastructure does not impact critical features.  And on top of that from Black Friday through to Cyber Monday we:

• Freeze all deployments except for the most critical fixes
• Over-provision all infrastructure above and beyond typical scale-out policy
• Double our on call rotations for automated alerts
• Enter a state of elevated monitoring of all services

We take uptime and performance very seriously because the best conversion optimization and machine learning are moot if we do not deliver fast and stable service consistently. Please feel free to reach out to your Customer Success Manager if you have any further questions.

And now for the updates!  What’s new this week from Engineering at Constructor?  We’ve been cooking up more than a few things.  In fact we’ve made lots and lots, with some powdered freeze dried lots sprinkled on top. In addition to improving platform uptime and performance, smushing various minor bugs, and helping customers, we also:

• Added more data points to our core analytics platform for future merchant analytics needs
• Improved dashboards and monitoring of internal services in advance of Black Friday
• Upgraded support for quizzes in multiple client SDK's (Python, JavaScript, Android, Node)
• Experimented with running internal services on ARM based architectures for improved performance
• Added support for ingesting product catalogs directly from Salesforce using cartridges
• Prepared several tests to be run at a later date for various product ranking models
• Added facet to facet filters for recommendations
• … and tested a new backfilling approach for complementary recommendations

Bon appetit.

~ Engineering at Constructor.

Until today filtering recommendation results has only been available on the API, but now it is easier than ever for merchandisers to specify filtered results using our dashboard. Merchandisers can apply business logic on recommendation results by applying “show only” filters to return results only from specified facets or categories. For example, retailers might have a vendor partner agreement that requires them to show only products from their brand in certain contexts. Another example is matching sets, if recommendations are shown for a bikini top, show only bikini bottoms. 

The Interact example below shows how “show only” filters limit results. A “show only speedo branded products” filter returns only speedo goggles and no other brand in the recommendation results below.

This ‘Show Only’ filter is applied in the recommendations merchandising menu at the strategy level. Merchandisers choose an input attribute for the rule to apply and an attribute to which results will be filtered. 

The image above highlights the process of selecting categories or facets options to filter recommendation results. Merchandisers can search or browse through their category hierarchy and facet options to set the filtered result. 

For more information, please connect with your Customer Success Manager, contact us through support@constructor.io.

A key component of website merchandising for most folks includes preparing for product launches and promotional activities. Whether it is a one-day Flash Sale, a much larger Black Friday event or a new product launch these types of activities often require some level of site merchandising to ensure the experience is just right. 

This type of merchandising often includes rules that are time-boxed for the future - to kick in at the start of the sale or the launch and then end after a predetermined amount of time. Until today, those rules needed to be previewed in real time as they were being set up. Now, we’re excited to announce that there is a new, seamless way to preview the entire experience across multiple categories or queries as a holistic review of the final experience for your site visitors, while ensuring that all your rules and any future-dated products will come alive precisely when intended. 

Date picker available next to search input in Interact page.

Within Interact, you can now select a date in the future to see the effect of any future-dated rules apply to the result set, and if you’re using our Product Launch capabilities then you’ll be able to see future-dated products come into the result set as well. 

Within the calendar view, you can see a summary of when searchandising rules start and end, as indicated with the red and blue dot indicators so that you can easily see when you have new rules taking effect and rules ending.

Calendar view with indicators for rule starting and ending. 

For more information about future-dated preview, please connect with your Customer Success Manager, or contact us through support@constructor.io.

In our latest update we head to an area of the dashboard that isn’t often the topic of release notes but important nonetheless - today we’re talking about the Catalog Uploads Page within the “Integration” tab in the dashboard. 

We have made a couple of changes to this page including a UI refresh so that it matches the rest of the dashboard for a more seamless experience. We have also added in user-friendly index names everywhere that we reference a particular index. These names are the same ones that are set in the “Indexes” section of the dashboard so they’ll always be familiar and easy to understand. 

Updated UI for Catalog Uploads page in “Integration” section of the dashboard.

And we’re not done with updates to this page, we have a couple more updates coming up on the horizon, including refreshing the pop-up experience when you click on a given catalog update for more information and adding pagination so that you can see more than the last 100 updates for any single index.  

For more information, questions or feedback please connect with your Customer Success Manager or contact us through support@constructor.io. 

Merchandisers have long been able to build bundles on our dashboard by manually slotting items in the complementary strategy. Now a new bundles strategy leverages Constructor's AI to automatically generate product recommendation bundles, a set of items that are frequently purchased together, making this shopping experience even easier to create. Bundles AB testing has shown increases in recommendation add-to-cart and purchase metrics by 2 - 4%, leading to a better shopping experience and increased business KPIs. The recommendations Interact view below shows what a shopper would see if they were looking at a lifeguard fanny pack with a bundles pod. 

The bundle strategy returns items that are most frequently co-purchased as a set with the seed product, the main product on the product detail page (PDP). We recommend showing the seed product, in this case the fanny pack, along with the visor and tee as a bundle with the option to add-to-cart all three at once on a PDP. The test results show strong shopper preference for purchasing well crafted bundles together. Bundles perform well when presented as a limited set to add all to cart together, but we still recommend using complementary or other preferred recommendations pods beneath these results to allow for additional discovery should the shopper not be in buying mode.

Bundles can be found in the Recommendations section of the dashboard under the list of available strategies. Deploying this new strategy is the same as adding any other strategy to an existing or new pod (see screenshot below). 

Quizzes are often highly experiential in nature and building out quizzes with that experience in mind can be tricky before you are able to see it come to life to make sure it is just as you envisioned - until now. 

We’re excited to announce the release of our quiz preview experience so that merchandisers can get a sense of what the quiz they are building out will be like before it is rendered on their site. This will allow merchandisers to make edits and adjustments and validate them in real time directly in the dashboard. 

Preview can be found in the top right next to the “save” button.

Example generic preview experience.

Within the preview merchandisers be able to validate all content (including copy and images), the quiz flow through different combinations of questions and answers, and the results returned at the end of the quiz for a given flow in a generic preview experience. 

For more information about Quizzes, check out our product page, connect with your Customer Success Manager, or contact us through support@constructor.io.

We recently released the slotting Rule Performance page to measure how slotting rules affect add to carts (ATC), and now the analytics are even more detailed! The new summary statistics and time series view will help merchandisers make data informed decisions to optimize business KPIs and merchandising rules. Clicking on the more information arrow on specific rules will allow you to view the detailed statistics on rule performance.

Below the summary statistics is a time series chart showing the daily conversion totals from slotted and unslotted results. The chart below shows that once the product was slotted (represented by the blue dots), the performance dropped off significantly from past performance (represented by the gray dots) when the position was not slotted. 

These additional insights will help merchandisers measure the impact slotting rules have on business KPIs. For more information, please connect with your Customer Success Manager, contact us through support@constructor.io, or read our documentation on merchant intelligence.

Power ecommerce product search and product discovery with Constructor and integrate with Akeneo for automated product catalog data updates.

We make the process of building and maintaining integration with Constructor even easier for Akeneo PIM customers. Introducing Constructor Connect for Akeneo – the App on Akeneo App Store, that makes it easy to connect and upload your Akeneo product catalog to Constructor. 

Now you can integrate all product data from Akeneo to power Constructor solutions, identify optimization opportunities, and empower your ecommerce team in real-time:

• In just a few clicks, connect your Akeneo catalogs for ingestion to Constructor (categories, products, and variations included)
• Done-for-you integration: no need to rely on engineering resources or third-party apps
• Up and running: integrate your product data in as little as the same business day 
• View catalog ingestion status anytime on the Constructor dashboard

Use our visual dashboard to validate your product data within the Constructor platform.

For more information, please connect with your Customer Success Manager, view our documentation, or reach out at partners@constructor.io    

Overview

Constructor’s conversion optimization and discovery benefits are only as good as our uptime and performance. For this reason, we have a robust process of performance validation and monitoring. During the holiday season and the peak demand period of Black Friday and Cyber Monday, we increase our standards in all of these areas out of recognition that it is the most important selling period for many of our customers.

Survey of peak demand for 2021

In planning our preparations in the run-up to the 2022 holiday season, we looked back to daily and peak demand changes during the holiday season and also reviewed how our baseline traffic has increased in the time since then.

Last Black Friday our overall traffic increased 44% compared to Black Friday 2020, which was already 200% over daily baseline levels. Not only did we maintain our 100% uptime, but performance during the peak demand periods actually improved relative to equivalent periods (due to changes in traffic patterns). Since then the system has already scaled without interruption or degradation. For the past several  months our average daily traffic has been greater than last year’s Black Friday event.

Performance improvements over the past year

Over the past year we have worked continuously to drive even better performance and scalability, contributing to improved latencies and zero downtime. Some example projects and outcomes include:

• Decreased index update delivery times with incremental indices
• Rolled out new datacentre in Asia
• Continued to improve instance boot and data download time by a further ~50%
• Introduced parallel processing of requests
• Added more IP’s per service region
• Created independently scalable sub-services in each region
• Increased service dependency redundancy
• Improved our internal service monitoring platform
• Upgraded our operations and incident response platform
• Introduced IP based traffic prioritization

Scale-out performance testing 

We have tested scaling to 2000% of current average daily traffic volume, while validating the continued performance of the following:

• Database connections 
• Monitoring infrastructure
• Networking infrastructure
• Response latencies @ median, 90th percentile, 95th percentile, 99th percentile
• Response latencies for each customer, and each product used by each customer
• Data ingestion SLA times

Chaos and anti-fragility testing

We also use chaos testing to validate that catastrophic failure of the following supporting infrastructure does not impact critical features (primarily search, autosuggest, browse, recommendations, collections request/response times):

• Disabled MySQL
• Disabled index builders
• Disabled personalization queues
• Disabled supplemental ranking engines
• Availability zone and data center failures

All of the above is in addition to the rigorous performance test and rollout plan we use for every release:

• Full test suite on every pull request (incremental code change).
• Production traffic replay for all deployment builds (multiple times a week).
• Rolling, risk-adjusted deployment procedures across worldwide data centers.
• Canary deployment for deploys touching critical path request/response lifecycle.
• Automatic build failures if sensitive thresholds on result quality, latency, memory consumption, CPU consumption and more are breached at any of these levels. 

Standard on-call procedures                                                                                  

At all times we have multiple on-call schedules for the following teams:

• Front-end and client teams 
• Data science and result quality teams 
• Core platform and response performance teams 
• Each of these have multiple fallbacks and tiered escalation policies

Automated alerting

Alerting is automated across dozens of metrics to ensure we are aware of incidents within seconds. A few representatives:

• Queuing times
• Per-service latencies
• Memory and CPU consumption

Special holiday on-call procedures

In addition, we take special precautions during peak holiday shopping periods:

• We will over-provision all infrastructure above and beyond typical scale-out policy.
• We double on-call rotation utilizing the above-mentioned automatic notification and escalation policies.
• The entire account and product team will be monitoring throughout the Black Friday / Cyber Monday period, with elevated focus for other holiday periods (such as Boxing Day).

Code freeze during critical time periods

We will freeze all deployments except for the most critical fixes from Black Friday through to Cyber Monday.

Conclusion

At Constructor, we take uptime, performance, and service stability very seriously because the best conversion optimization and ML are moot if we don’t deliver fast and stable service consistently. The goal of this document is to provide our customers with a broad overview of our site reliability practices, as well as a specific view of our holiday readiness procedures. As always, please feel free to reach out to your Customer Success Manager if you have any further questions.

Building on top of our recent release of the quizzes builder within the dashboard, we’re already making improvements upon it to provide even greater flexibility in creating quiz experiences. 

Quizzes are often highly curated and immersive experiences which usually means including beautiful imagery to support the story telling that you are looking to achieve by incorporating a quizzes experience within your product discovery journey. This can look like having images that provide visual reference to the options you might be presenting, or a beautiful background image right at the start of your quiz to help set the mood and immerse a user within a quiz experience. 

Images to further convey the tasting note options within a guided coffee quiz experience. 

We’ve always supported images within our quiz experience, but now we’ve made it even easier by adding support for them directly in the dashboard on all question and option types. Any externally hosted asset (including images) can be added into a quiz, and can be adjusted at any time without needing to make changes to the rendering of the quiz. 

“Add an asset” for both the question and the option level on a multi-select question type.

Asset editor screen where you can reference in the hosted asset URL, alt text for accessibility and add a secondary asset if desired.

For more information about Quizzes, check out our product page, connect with your Customer Success Manager, or contact us through support@constructor.io.