There’s no doubt about it - site speed matters.
Nothing is more frustrating than a slow website. Not only is it bad for end users, it’s also very bad for website owners. It affects your traffic, page views, conversions, sales, and your overall reputation. It also adversely affects your SEO.
Here are our top tips for increasing the speed of your website.
#1 Minimize the amount of filters and filter values used
The backend is doing an aggregation on each of these values, which takes time, so the more filters and filter values enabled you have the more time it will take to generate responses.
A great way to find out what filters are most popular, and which ones are used least, is to check the Analytics section in your Findify dashboard.
#2 Reduce the amount of data returned by the API
Carefully choosing fields that need to be returned from the API will help in reducing the size of the response, so that it will be transferred faster. This means the serialization\deserialization process will happen faster as well and will take less resources both for the server and the browser. Findify’s recommendation is to evaluate these already in the customization process so that you can set it up to your needs. Make sure that only what you need returned from the Findify API is activated. By enabling less you can increase the speed.
#3 Minimize the number of the products returned per page
The fewer products that are returned per page, the smaller the response time will be. This makes the data transfer faster. You can easily change this on the Findify dashboard under Advanced setup / Pagination settings / Number of results to display per page. Our recommendation is to have 24 or less.
#4 Use different image sizes for different screen sizes
Minimize the amount of data transferred for images depending on the size of the screen. The less data is transferred the faster the page will be shown.
This is especially effective on mobile devices that might have a slower cellular connection. Many ecommerce platforms (like Shopify) supply an interface that allows you to do this easily for your products.
#5 Minimize the amount of scripts and styles in the page
A website can have many scripts and functions that load each time a visitor is on the site. It is important to manage these continuously and make sure to remove scripts and styles that are unused or obsolete. Some scripts and styles can be combined or minimized, if needed. This helps with the data transferred and the amount of connections created.
In addition to this, if you are a Shopify user, you should keep an eye on the number of apps you have installed. If you have more than 20 then you likely aren’t using them all. If you have too many, this could slow down your site. Routinely go through your installed apps and remove any you aren’t using.
#6 Lazy load images
Findify provides a way to lazily load images in the search \ collection pages, so that only images that are currently in the viewport are loaded. This not only minimizes the amount of data transferred (instead of 24 you get 3-6 images), but also the amount of total requests on the page (there are limitations on the amount of connections the browser can have).
To activate this you can do it directly in the customization or contact us at firstname.lastname@example.org and we can help you to set it up.
#7 Load scripts in Asynchronous way
A good general rule to have on a website is to load most of the scripts in an asynchronous way. This way you are not blocking the page from being rendered. The faster the page renders the sooner users will see the content, and Findify will start processing. It’s important to understand how your scripts and functions are being loaded on your website as it gives you insight into what parts are taking too long to load. Making changes to this should be done with caution, however, as it might cause problems.
#8 Optimize images (sizes, quality)
Optimize images on the site for size and quality - use exact image sizes as in design (don't load an 800 px image to show it in the 200 px preview) and don't forget to optimize your PNG images.
We recommend using the tool TinyPNG, or something similar. This will help reduce the amount of data transferred on the page and help it load faster.
#9 Make sure DOM is loaded as fast as possible
There are great resources on the internet to help you improve site speed. One we recommend is Google Page Speed. This and similar services can identify what can be done in order to make sure that DOM* is loaded as fast as possible. The faster DOM is loaded the sooner your users can start interacting with your site and the sooner Findify can start operating, helping shoppers find the right products for them.
Site speed is very important to the overall shopping experience. At Findify we work continuously on making our service as fast as possible. But we also know that we are just one part of the whole experience and that we can only be as good as the entire shopping journey. That’s why we work with merchants to not only make sure Findify is operating at max speed but that your site is also. Please get in contact with us if you would like to discuss this more in depth and have us look at your current setup to identify speed improvements.
Adding stickers to product cards is a well-established technique to help shoppers make buying decisions, thereby increasing store conversion.
This incredibly powerful ecommerce tool (which is also sometimes referred to as the displaying of product badges, labels, tags, or markers) is utilized by all major shopping portals, including Amazon, Walmart, Etsy, and AliExpress to boost sales.
While there are a number of third-parties offering product sticker services, Findify clients have the additional benefit of sticker creation without the need to download additional apps.
Why are product stickers important?
As we all know, online shoppers have a very limited attention span. This means you only have a few valuable seconds (between 5 and 8 seconds, to be exact) to grab their interest. Product stickers are a very effective way of doing this.
Stickers increase simplification of navigation and can quickly and easily highlight special attributes of different products (Is it ‘New’? Is it ‘Eco-friendly’? Is it a ‘Bestseller’?), helping the customer differentiate between products.
What kind of stickers can you feature?
Put simply, you can feature any sticker with any logic, image, logo, or text that you desire. You can customize every aspect of the sticker, from how it’s generated, what it says, how it looks, to what it does when a customer hovers over it.
Some of the most popular types of stickers are those in the ‘Popularity’ category - those which create a sense of trust in the brand and in the product.
When customers see products being purchased by a lot of people, it is a powerful psychological reason for them to also buy. This social proofing helps to build trust among shoppers, and makes them more likely to purchase.
Some examples of stickers in this category include ‘Bestseller’, ‘Top Rated’, ‘Editor’s Pick’, ‘Back in Stock’, and ‘Only x Left’.
Other types of stickers deal with the cost-saving aspect, such as ‘Sale’, ‘Bundle Saving’, ‘Buy One Get One Free’, ‘Limited Time Offer’, ‘Deal of the Day’, and ‘Free Shipping’.
Stores selling environmentally friendly products may want to add stickers like ‘Organic’, ‘Eco-conscious’, ‘Vegan’, ‘Ethically Sourced’, and ‘Fairtrade’, while other stores also benefit from stickers like ‘New’, ‘Members Only’, or a countdown timer.
When the stickers are set up, they will automatically populate based on your conditions. Perhaps you wanted every product with a certain tag to have a ‘Sale’ sticker, or you want every product under a certain price to carry a ‘Incredible Savings’ sticker, or for items with fewer than 50 in stock to show a ‘Hurry only x left’ sticker. Whatever the requirement, there is a logic for achieving it - one that requires little to no manual effort on the part of the merchant.
I’m a Findify client, how can I get my stickers set up?
First, have a think about what stickers you’d like to use on your products. While it might be tempting to add as many as you can think of, too many stickers will only confuse your customers. Only add stickers that will bring value to your specific store, and don’t include multiple stickers that all mean roughly the same thing. The purpose, at the end of the day, is simplified navigation.
Next, have a look at our support documents here. If you need help setting up your stickers, contact us by emailing email@example.com and one of our developers will then liaise with you on what stickers you want, how you want them to look, and what rules you want them to follow.
You can feature multiple stickers on one product if you’d like, and you can feature the stickers in your autocomplete, search results, and collection pages, as well as in your recommendations widgets. Our stickers are completely compatible with all our personalization software solutions - Personalized Search, Recommendations, and Smart Collections.
In terms of position, you can display the sticker as an overlay on the product image, or on a separate line below the product title, or anywhere else on the page. It is also possible to blacklist products to prevent the display of stickers on it entirely.
For more information on Findify’s powerful ecommerce tools, including personalization software and solutions such as Personalized Search, Smart Collections, and Recommendations, book a demo here. To ask about creating stickers for your store, email firstname.lastname@example.org.
The rise of mobile ecommerce has been nothing short of spectacular.
Annual mobile ecommerce sales are expected to exceed 3.5bn USD in 2021, while a recent Findify study shows shoppers using a mobile device accounted for more than 60% of all traffic to ecommerce websites.
Considering how popular mobile ecommerce has proven to be, it is increasingly important for merchants to put a lot of thought and effort into how their site appears and functions on mobile devices.
With that in mind, the experts at Findify have put together a list of their top tips on how to optimize your ecommerce site for mobile - starting with the hugely important autocomplete function.
The Autocomplete Function: How important is it?
When implemented correctly, a good autocomplete functionality serves to assist and guide users toward better search queries.
According to experts from the Baymard Institute, when autocomplete queries are done well “they inspire users about the types of queries to use, teach them correct domain terminology, help them avoid typos, and assist them in selecting the right scope to search within”.
On mobile specifically, autocomplete query suggestions are also helpful in reducing typing during query formulation.
Extensive research carried out by the Baymard Institute on 409 test subjects over the course of two years reveals users struggle with typing on mobile more than they do on desktop. Thus, any shortcut that autocomplete query suggestions can provide offers users a way to avoid struggling with the mobile keyboard and potentially introducing typos.
In testing conducted during the aforementioned research, carried out by Baymard, 78% of subjects using mobile ecommerce utilized autocomplete query suggestions, which spared them from typing an average of 8 additional characters on their small mobile keypads.
Autocomplete Design: No more than 8 suggestions on mobile
As you can imagine, a poorly designed autocomplete interface makes it extremely difficult for users to understand the options available within the function, and can either overwhelm the consumer or lead to unexpected, and potentially undesirable, interactions.
The first rule of thumb when designing your autocomplete is this: ensure the generated list of suggestions is a manageable one. The number of options should be limited so as not to overflow the user’s viewport. While this rule stems from an ease-of-use reasoning, there is also a secondary problem caused by a long list of suggestions - choice paralysis.
Findify experts recommend, and researchers from the Baymard Institute agree, that autocomplete suggestions should be limited to a max of 10 on desktop and 8 on mobile. There should be at least 4 suggestions on both devices.
Autocomplete Design: Minimize distractions
When designing your autocomplete, on both mobile and desktop, you also need to think about how the overall page will look to the shopper - including the area outside the autocomplete box.
It’s important not to distract the consumer when they have already initiated a search and have started to engage with the autocomplete function. One simple tweak can ensure this: darkening the page background while autocomplete is active.
This gives the autocomplete a stronger emphasis, minimizing elements like ads, carousels, and other page content, that could interfere with the search process.
Adding a subtle border around the autocomplete function can create even more depth and make it easier for the user to focus on the suggestions offered.
Autocomplete Design: Highlight the differences between query and suggestion
Another best practice when designing your site’s autocomplete is endeavoring to stylistically highlight the difference between words that match the search query and words that are autocomplete suggestions.
There are two ways of doing this - either you highlight the characters already typed, or you highlight the suggestions being made by the autocomplete.
While both types of unique styling can reduce the visual burden for users, highlighting the query characters helps the user gloss over repeated words, giving them less to read.
In the latter alternative, it is the prediction that is emphasized, which helps draw attention to the different suggestions, allowing the user to ignore the characters they have already typed.
Both methods are effective ways of differentiating between the query and the suggestion, thereby speeding up the process for the user and helping them make their next decision quicker and easier.
Autocomplete Design: Ensure users can easily read and click suggestions
The final point relating to autocomplete design is a seemingly obvious one, but is an aspect many merchants accidentally overlook - readability.
The dangers of poor readability (and clickability) in a mobile autocomplete function is best highlighted by a recent study carried out by the Baymard Institute.
“During mobile testing, autocomplete suggestions styled in small font sizes made it difficult for subjects to view and select a suggestion. Further, when inadequate spacing and small font sizes were combined, it contributed to mistaps that left some subjects puzzled upon arrival to unexpected results pages,” they wrote.
To resolve readability issues within autocomplete, use legible font sizing, ensure suitable spacing between tappable elements, and provide hit areas of an appropriate size. Finally, presenting autocomplete suggestions in lower case or title case (or “headline-style”) makes for easier readability than all uppercase text.
Autocomplete Results: Include spelling tolerance
Now that you know what design elements to incorporate in your autocomplete function, you should be thinking about the suggestions themselves and how they are presented to your valued customer.
One of the most useful capabilities you can infuse into this process is that of spelling tolerance. Your customers are only human, and when on mobile they’re also struggling with a small screen size. Mistakes will happen. Therefore, your autocomplete really needs to be able to deal with the typos it will inevitably be faced with.
Since autocomplete plays such a key role in early search interactions, unexpected suggestions due to minor typos can cause users to change their product-finding strategies by seeking other browsing methods or reworking queries, and, in the worst cases, contribute to abandonment downstream if alternate product-finding strategies don’t quickly lead to relevant results.
Since spelling errors in search queries do occur with significant frequency, autocomplete’s relevance can, and should, be enhanced by mapping misspelled words to meaningful autocomplete suggestions.
Autocomplete Results: Display and update suggestions quickly
Interacting with autocomplete suggestions becomes relatively challenging for users when they’re presented with inconsistent or slow load behavior. Autocomplete suggestions should appear and update nearly instantaneously as the user types. Delays in the display of autocomplete suggestions mean that some users will never get exposure to the very guidance that such suggestions are intended to provide.
“[Improving speed] requires intensive optimization work to ensure that the search engine can produce autocomplete suggestions (ideally in under half a second, though faster is better, as the interface will feel more responsive to users), as well as focused work on lowering network latencies, and consideration and improvement of any other factors that may increase the delay between user input and autocomplete suggestions loading,” explained an expert from Baymard.
“This is especially the case for mobile sites, where slow-loading autocomplete query suggestions or interaction issues are magnified. During mobile testing, we observed some users typing so quickly that they never saw autocomplete suggestions. Others struggled with tap and hit area issues when trying to re access the search field. It’s therefore critical that autocomplete speed and load behavior is optimized for mobile sites.”
“Give shoppers the functionality they deserve, and they will convert more often”
Overall, testing has revealed most mobile users will struggle with issues that they don’t face to the same degree when browsing desktop sites. Despite this, the trend toward mobile e-commerce continues undisturbed - for most ecommerce sites, more than half of their traffic is on mobile, and some verticals have almost exclusively mobile traffic.
“As more and more ecommerce activity continues to take place on mobile, it’s more important now than ever before that the mobile user’s experience is a smooth one - from arrival on the site right through to successfully placing an order,” explained Findify’s Vsevolod Goloviznin.
“An ecommerce site’s autocomplete is a big part of that shopper journey, which is why it’s so important to perfect this capability’s design and functionality. We know that searching shoppers convert two to three times more than those who browse, so by lowering the barriers of their search journey - with rapid relevant suggestions, correcting typos and removing distractions - they will add considerably to the bottom line.”
This piece was written by experts at Findify, a leading provider of site search and personalization software, with infused learnings from the Baymard Institute. For more information on Findify’s powerful ecommerce tool, which includes solutions such as Personalized Search, Smart Collections, and Recommendations, book a demo here.
Check out the latest Findify clients
Experience the Findify effect
As all merchants can no doubt attest to, retail is an ever-changing landscape.
Those who have been in the game long enough may remember business in the 90s and 00s - where the biggest challenge (and opportunity) lay in the rise of ecommerce.
Today, with the vast majority of retailers embracing ecommerce, facilitating more than 3.5bn dollars in online sales every year (and growing), many experts believe the biggest challenge (and, again, opportunity) lies in the rise of mobile sales.
For the last few years, these experts have been expounding the virtues of mobile first, warning that mobile ecommerce would only continue to grow.
Now, new research and statistics, conducted and compiled by Findify experts, has revealed the magnitude of this shift, showing not only how incredibly important mobile ecommerce has become today, but also analyzing the tangible and measurable extent with which it has grown over the course of the last 16 months.
In a study of more than 4.2 million shoppers interacting with Findify solutions (on the websites of Findify clients), conducted over a one week period in April, 2020, shoppers using a mobile device accounted for more than 60% of all traffic.
For the same period, shoppers using desktop devices accounted for almost 30% of all traffic. Fewer than 5% of people were shopping on tablets.
While this snapshot of online shopper behaviour serves as an important reminder of the continued staggering growth of mobile ecommerce, the exact percentages of devices used fluctuate slightly when analysed over longer periods.
With this in mind, Findify experts conducted further research, diving into the shopping behaviour evidenced in 1.8 billion online shopping sessions.
In January, 2019, for example, mobile traffic was only just exceeding that of desktop traffic. Roughly 43% of all visits were from mobile, with roughly 40% from desktop.
The below graph illustrates how this has changed over time, with mobile traffic increasing significantly while desktop and tablet traffic continue to decrease.
Spike in mobile ecommerce traffic following Coronavirus lockdowns
In the above graph, two recent spikes in mobile traffic can also be identified - one taking place in December 2019 representing the Christmas season, and the other taking place in April 2020, representing an increase in traffic following worldwide Covid-19 lockdowns.
Both instances show an increase in mobile traffic specifically, with a corresponding decrease in desktop traffic, proving that there remains roughly the same amount of shoppers, but these shoppers are favouring mobile devices over desktop devices, gradually changing their shopping behaviour.
Does mobile ecommerce revenue also exceed that of desktop?
There’s no denying that mobile users make up the majority of an online store’s overall traffic, but are they just there to browse? Or do they actually buy?
According to the stats, they actually buy. In the below graph, based on 16 months of revenue generated by more than 1.8 billion sessions on hundreds of online ecommerce stores, mobile sales were consistently between 16% and 20% higher than desktop sales.
Findify Insights: Ecommerce merchants need to commit to mobile first mindset
So what does this mean for ecommerce merchants? According to Findify CCO, Joakim Amadeus Olsson, it means they should be allocating more resources towards ensuring the online version of their store truly is mobile first.
“We interact with many merchants, many times a day, and we’ve realized that while a majority recognize the importance of mobile, very few are truly mobile first in their priorities and initiatives. This is of course extremely understandable, given that desktop has been the dominant window for many years now, yielding amazing results with capabilities still far beyond mobile. Change is always hard to truly implement and you don’t necessarily want to fix something that’s not currently broken,” he said.
“But the stats don’t lie - consumer behaviour is changing and will continue to change. More consumers are shopping, and buying, on mobile devices these days than on desktop, with the gap between the two behaviours continuously increasing. That’s a fact. Retailers need to occupy space where their customers already are - and that’s on mobile. In my opinion, the businesses that not only recognize this, but take action on it, will be the ones seeing the most long term success in the coming years.”
This research was conducted by experts at Findify, a leading provider of site search and personalization software. For more information on Findify’s powerful ecommerce tool, which includes solutions such as Personalized Search, Smart Collections, and Recommendations, book a demo here.
Check out the latest Findify news
Experience the Findify effect
Findify can personalize all the major touchpoints on your store: search, category navigation, and multi-page recommendations. Of this stack, one of the most complex touchpoints to personalize is search — and it’s also the most essential one.
Search users are customers with a high level of purchase intent, and to be able to harness that intent by delivering a truly unique experience to each and every user is an invaluable competitive advantage. There are thousands of knobs you can tune to make search results better, but at Findify we’ve been teaching the machine to automatically make the best ranking decisions for you.
Read on to learn the story of Findify’s technological (r)evolution.
How generic search works and what's wrong with it
When a new customer visits an online store and decides to search for headphones, she types 'headphones' in a search box, hits a button, and then receives a list of matching products for her query. Nothing could be more straightforward, right?
Think again. According to our statistics (collected from more than 1,000 merchants over the course of a one week research period):
In other words, the way search results are ordered is EVERYTHING. Hey, it’s 2020, and shoppers don’t have the patience to scroll through search results they perceive as non-relevant.
A standard search engine embedded in your shopping platform uses proven, open-source technologies such as ElasticSearch (Bigcommerce) or Solr (Shopify) — which, under the hood, use generic algorithms [TF/IDF, BM25] to rank search results. These algorithms match terms in a search query only to terms found in product descriptions, nothing more. But an online store has much richer metadata about their products and customers:
All these valuable snippets of information are typically ignored by standard search engines like ElasticSearch, giving customers less relevant results.
Improving search quality
Before taking steps to improve your search quality, you need to understand how that quality is measured in the first place. In the search engine industry, two groups of metrics are used:
Using these two simple rules to re-rank products in search results yielded, in general, a 5–10% improvement in user search experience in terms of both relevance- and business-related metrics.
If such a simple approach gave such measurably positive results, imagine the dramatic boost to search quality and merchant revenue if we go further, if we integrate all the data we collect from a merchant’s user activity into a single model. Enter the “learning to rank” family of machine-learning models: RankNet, LambdaRank and LambdaMART.
By the time a customer looks at the search results, we can already assume a lot about the search query itself, the products shown there, and past user behavior. This knowledge can be expressed as feature values, a set of numbers characterizing the importance of each small bit of collected data. The number of these feature values is limited only by the developer’s imagination. At Findify we use about 60 of them.
Through the feature values prism, our search algorithm sees the search results in this way:
This might look like a set of random numbers, but when you collect a month of user search activity you start noticing that some feature values carry more weight in results ranking for a particular query. For example, more product reviews lead to a higher click probability, but the review score itself has surprisingly little correlation with clicks.
You also notice that the picture above focuses only on product-specific feature values like product price and number of reviews. But there are two other important classes of feature values: user-specific and search-specific.
User-specific ones are dependent not on the products shown in the search results, but on user behavior, for example:
Search-specific feature values depend on the search query itself, for example:
Here’s a simple example to demonstrate the way we can weight these feature values to produce a better search ranking. For a merchant selling shoes, a new visitor looking for “men’s boots” got these search results:
Another visitor, looking for “winter shoes,” got another set of results:
You may notice that in both cases, users prefer clicking on a cheaper but more popular product. One response would be to then alter the ranking algorithm in such way that cheaper and popular products will get better ranking. But how hard can we boost them before we start ruining user search experience?
You can compute the best weights for price, popularity, and similarity using linear regression; this mathematical technique allows you to predict an event (in this case, a click) based on an input set of feature values. For the search results above the weights are shown below:
Thus we can assume that:
And just like that, we have built a simple — but real — machine learning model to predict user clicks on search results!
Using this model we can reorder the product for the first query, “men’s boots,” in the following way, the higher weight meaning a more relevant product:
In re-ranking these search results we’ve swapped the first two products, since the “Green boots” have a much higher weight according to our model.
Linear regression-based modeling is a good fit for simple cases, when you have only a couple of independent but descriptive feature values. In the real world, however, where creating more relevant search results means taking into consideration hundreds of feature values and millions of search results, you need a stronger machine learning algorithm — like LambdaMART.
LambdaMART is a learn-to-rank algorithm developed a few years ago by Microsoft. It far outshines the data-mining competition in producing high-quality search rankings, thanks to the following features:
LambdaMART’s approach is based on a specific way of building a collection of decision trees. A decision tree is a very simple data structure that models decisions — or states of being — and their outcomes. This one, for example, predicts passenger survival on the Titanic:
According to this decision tree (depth 3, with 3 feature values), if you were an adult male traveling in 3rd class, you probably weren't getting on that Titanic lifeboat.
As LambdaMART learns it builds a set of trees where:
Combining a large set of simple tree predictors into a single strong model is called gradient boosting; at Findify we use 100–200 trees, each of depth 6–8 and with over 60 feature values, to achieve truly remarkable search quality results.
But the machine learning algorithm itself is only one aspect of the full solution. To achieve a truly personalized, meaningful search for each unique user, the whole data analytics pipeline needs to work in real time. At Findify, when a user clicks on a search result or views a product page, it takes less than a second to receive, process and update his or her profile, and it takes less than 10 milliseconds to compute personalized search results.
The problem with most machine learning models is that they are opaque black boxes: you cannot reliably explain why they arrive at a particular prediction. At Findify we measure search quality, before rolling ranking-related changes into production, by using back-testing:
Back-testing simulation is a useful tool for performing a preflight sanity test, but it’s still a simulation. Real users can — and will — behave unpredictably. To measure search quality in real time, we monitor and collect these metrics:
At Findify we’ve spent months running personalized searches, and the general results are quite astonishing. As an example, for one of our merchants, conversion increased by 5–10%, average click rank went down from 9 to 7 (this is a good thing), and NDCG is constantly 15% better.
Another merchant saw 25% increased revenue per user, while others have seen an 18% increase.
Other merchants we’ve observed have seen similar improvements. Surprisingly, search personalization rarely improves average order value, but consistently improves conversion.
[Written by ML Developer Roman Grebennikov]
* Want to find out what Findify can do for you and your business? Book a free demonstration with us here.