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.
Content marketing continues to be one of the top forms of marketing out there. Businesses around the globe utilise content to communicate with their audiences, share their expertise, and encourage conversions.
Why Use The Power Of Content Marketing?
Content marketing is the act of creating, publishing and sharing a variety of content across different channels. Its purpose is to communicate with a targeted audience and is not normally intended to directly promote. It works on relationship-building and raising brand awareness. Therefore, content marketing should have a strong focus on sharing helpful, actionable advice.
The Most Effective Content Marketing Techniques
There are different types of media to use in your content marketing strategy including:
Today, we're going to look at some of the most effective techniques you can incorporate into your overall content marketing campaign.
1) Optimise Your Blog Posts For Search Engines
Blogging may be the most obvious form of content marketing out there but it's also one of the most effective. Running a blog enables you to build an audience and following by sharing your knowledge. As an expert in your particular field, your audience will expect you to have answers to their questions and to be able to give them sound advice.
Just like your webpages, your blog content needs to be optimised for search engines. Search engine bots look for relevant, up-to-date information on websites and blogs so that they can present the most helpful search results to their users. As such, consistently creating relevant and up-to-date blog content will help it rank higher in search engines. But, there are also other necessary steps for you to implement:
Brainstorm - Keywords are what search engines use to match your content with search queries. Begin by brainstorming possible terms that people will search for in relation to what information you're offering. For example, if your blog post explains how to lay a carpet, possible search queries may be:
These are the types of things people search for when they're looking for an answer or instructions.
Use a Keyword Planner - Once you have your list of possible keywords, use a keyword planner like Google Keyword Planner or Ahrefs to pinpoint which ones to include in your posts. Both of these platforms will show you the search volume of your keywords, along with how much competition there is for them.
Choose Your Keywords - On both platforms you will be shown a list of similar, alternative keywords. This enables you to select the most relevant one that will be easiest to rank higher for.
While the examples given above are all very similar, one in particular may have a considerably higher volume of searchers per month. Although more searchers may seem like a positive thing, you need to take into consideration the amount of competition for each keyword before selecting them.
A keyword with high competition means that there are already lots of websites that are ranking for it, and some may be very well established. This would make it very difficult and timely for you - as someone just starting up - to rank high using those keywords.
For example, you wouldn't be able to overtake Amazon or eBay in search engines. Both these sites have a considerable amount of traffic, along with lots of funds to put into their SEO campaigns. They are too well-established for smaller businesses to compete with in search engines. The ideal combination for start-ups are keywords with high search volumes and low competition.
This doesn't mean that you should avoid the high-competition keywords altogether though. Ranking high for high-competition/high search volume keywords may take much more time and effort, but in the long run they could bring you significantly more traffic. Therefore, it's worth including a mix of keywords in your content.
Search engine bots crawl websites to find and index pages. They can enter your pages and posts through these links, making the process faster and more effective.
Include links in every blog post you write to direct your audience (and bots) to other relevant posts that you've written. Your internal linking system will encourage your audience to visit more pages on your site by providing them with an easy means of further reading. You should also include links that lead to high authority, relevant websites to increase your SEO.
In addition to the linking system within your blog, work on link-building outside your blog to direct people and crawlers to your posts. Guest posting is a great way to do this and to reach a wider audience. It is the act of posting a unique piece of content on another blog or website and including a natural-looking link back to your relevant blog post.
Search engine bots look for quality and relevance in websites. Ensure that your posts are informative, accurate, and free from spelling, grammar and punctuation mistakes.
To be considered high-quality, your text should be broken up with subheadings, bullet points, and images. Make it easy to understand with short sentences and non-complex words.
Keep it Simple
Don't overload your blog with widgets, plugins and large images. These will all add to the page-loading time and result in a slower blog. Google's ranking algorithm takes speed into consideration and a slow blog will impact your SEO efforts.
If your blog is slow try compressing your images, clearing your cache, and removing any unnecessary elements. If you're using WordPress, install plugins that are dedicated to clearing your cache and increasing your site speed.
There are also plugins to help optimise your blog for search engines, like Yoast and SEMrush. Yoast, in particular, is designed for users to improve your ranking even if they're not very SEO-savvy. The free version goes through all the basics and advises users on aspects such as:
2) Inject Personality into Your Brand and Content
Consider if your business was a person. What would they be like? How would they speak to your audience? Think about the type of words and language they would use.
3) Create and Distribute Tutorial Videos
Be specific in your videos and offer relevant, practical advice that people can take away and put into action. Videos should always be high in quality with a clear picture and sound so that you are viewed in a professional capacity. A poor-quality video will reflect badly on your brand.
Add your brand name, logo, tag line and a gentle call-to-action in your video, but don't make selling the main purpose. Request likes, follows and advise your audience where they can find further related content if they'd like to learn more.
Distribute your videos on social media platforms, emails, blogs, and your website. You could also create short video courses and distribute them to course platforms.
If you really want to get stuck into video-based content and your budget allows it, you can take advantage of various video ad programmes, like YouTube's. These are a great way to place your business in front of people who enjoy watching videos.
4) Showcase Your Testimonials
5) Launch Competitions, Polls, Special Offers and Giveaways
There's no better way to increase engagement than simply asking your followers to engage! Competitions and giveaways will always be a hit because people love to get something for nothing.
The exciting thing about competitions is that they allow you to get creative. The more creative and exciting your competitions, the more engagement you'll get. Here are some simple tactics to get your imagination going:
Competition rules and regulations vary from country to country. They also vary on different social media platforms. You need to check all of these before launching any competition to make sure you stay compliant.
6) Conduct A/B Tests On Your Ads
If you're running an ad marketing campaign as part of your content marketing strategy, A/B testing can be really beneficial for maximising your results. Even better, it's an easy task to carry out. A/B testing ads is the process of running two ads simultaneously, with one element different to the other.
You can A/B test:
Keep in mind that only one element should be changed at a time so that you are aware of specifically what has given you your results.
Some Points to Remember
Regardless of your specific content marketing strategy, keep these things in mind throughout:
Always add value - Your audience wants to feel like reading, watching, or signing up for something is worth their time and effort. Lazy content creation just wont do it. Take the time to make sure you're adding value to each piece you create and share.
Social Media is Your Marketing Friend - Social media, you either love it or hate it. But, regardless of your stance here on it, social media can be a tremendous tool for marketing. Every time you create a new piece of content, post it on social media. A new post on your blog? Let your followers know and give them a link for easy access. Writing a new ebook? Again, let your followers know. Tell them what they can expect and when to expect it.
Keep Creating Content - Content marketing is a continuous cycle. You need to keep creating, posting and sharing to build momentum and build your positive brand reputation. It wont happen over night but if you work at it, you will gradually widen your audience, gain more followers, and encourage more sales.
Picture a typical shopping journey for a brick & mortar store:
A customer is walking in the Summer heat. She passes a display of mannequins with cooling linen clothes and decides to check out the store. Upon entering, an assistant greets her and guides her to the right sections. The assistant also suggests additional items that fit her style and even finds them in the right size. Thirty minutes later, she walks out with a great new outfit.
Brick & mortar stores employ a variety of strategies to recommend products to customers, such as displaying hot-selling items at the store-front, suggesting items based on someone’s shopping history, or placing certain items closer to each other.
These tweaks not only increase the average basket size, they transform the entire shopping experience for customers.
For e-commerce websites the opportunities are endless - so what is the optimal strategy to grow your business?
The challenge for e-commerce
For digital stores, designing a good shopping experience is even more critical.
Not only do e-commerce websites present a completely different medium for shopping, online customers also tend to be unforgiving. It’s common for customers to abandon a shop because a webpage looks messy, or the images are blurry, or they can’t find what they want with the search bar. Hence, improving the online shopping experience, with the right technologies and the right strategies, is key to customer retention and conversion.
Each component of a website must be designed with the shopping experience in mind. For now, let’s focus on an important part of your online store: recommendation widgets.
As with every website component, there are many different types of recommendations to consider and many different ways to style them. And it’s hard to identify the best solution because the strategy you use depends on changing business circumstances.
We’re here to help. In this article, we’ll be improving our understanding of recommendations and covering some strategies for using them.
The purpose of recommendations
At the heart of it, recommendations are a way to present products that a shopper is not looking for, but is likely to buy. Recommendation widgets are commonly employed by e-commerce websites. For example, when you’re shopping on Amazon and you see a widget that says “Customers who viewed this item also viewed” followed by various products — that’s a recommendation widget.
To break it down even further, recommendations work by accomplishing one of three things:
Why are recommendations so important?
Let’s go back to a time when humans were hunters and gatherers, a time when choosing the right products wasn’t simply a matter of fashion or comfort, but of life and death.
Picture a young homo sapien who finds a berry bush. The fruits are brightly coloured and plump, but a homo sapien who reached for the fruit without a second thought wouldn’t survive for long: the berries might be poisonous!
So what does the homo sapien do? He relies on the testimonies of tribe members who have tried the fruit. Maybe they tell him that it is safe to eat, or maybe he sees other tribe members eating them. This gives him the confidence, and desire, to try the berries. If they turn out to be sweet and juicy, he’ll return for more.
These are the underlying roots of recommendations and branding.
We try new products (an unknown berry) based on the recommendations of friends and family (the suggestions of tribe members). When we trust the brand of a product (the shape and colour of the berries) to satisfy our needs (whether the berries are safe to eat), we become loyal to the brand.
It’s been ingrained into our psyche since primitive times to rely on recommendations. Those who lacked this instinct didn’t survive.
How should I employ recommendations on my e-commerce website?
“Would you tell me, please, which way I ought to go from here?”
“That depends a good deal on where you want to get to,” said the Cat.
— Lewis Carroll, Alice in Wonderland
Knowing that something needs to be done is easy, knowing what to do is hard. There are many different strategies for leveraging recommendations on your e-commerce website, and the strategy to use depends on your business circumstances — who are your customers, what are your products, what’s happening around you.
Navigating the possible strategies can be dizzying, that’s why we’ve prepared a flowchart to help you pinpoint the appropriate kind of recommendations for your business needs.
Let’s take a closer look at two recommendation strategies on this flowchart: ‘#1 Hot Sellers’ and ‘#8 Frequently Purchased Together’.
Leverage purchasing trends with ‘Hot Sellers’
Consumer purchases don’t happen in a void, there’s usually an external context to them.
Sometimes customers buy new clothes to replace worn out ones, or warmer and lighter clothes when the seasons change. Sometimes there’s a sports event and customers want to display their support for their favourite teams, or maybe the holidays are here and festive products become popular.
There are two ways to perceive this. It could be a source of frustration for retailers to keep up with changing trends, but it could also be a source of endless opportunities. If you’re always pushing out new products or you have a versatile product line that ties well with current events and seasons, then each changing trend presents a fresh opportunity to sell different products to your customers.
‘Hot Sellers’ is a recommendation strategy that works on this principle by presenting trending items to your shoppers. Utilizing an intelligent recommendation system enables these recommendations to be rapidly updated based on real-time shopping trends, so that they remain relevant every day and every hour.
Sell more complementing items with ‘Frequently Purchased Together’
‘Frequently Purchased Together’ is a great strategy to consider when you have products with good synergy.
Think of luggage bags with neck pillows, salsa with tortilla chips, swimming suits with flip flops. Shoppers looking at certain products are likely to buy other related products.
If you have products that fit this pattern, this is an opportunity to let shoppers know that they can tick off more items on their shopping list while they’re here.
And of course, there are endless combinations between products, and some work better than others. It’s best to use an intelligent recommendation system that can study purchasing patterns and identify the products with the best synergy for you.
Where to go from here
These are examples of strategies that, implemented appropriately and with the right technologies, can greatly improve the shopping experience of your online stores.
If you haven’t been using recommendations, or you’re not sure how to proceed, the best thing to do is to start then iteratively improve from there.
Ultimately, there are endless opportunities to use recommendations, and endless ways to combine and style strategies. E-commerce is an evolving industry. To thrive in this environment, you must use the strategies that fit your current needs, and be flexible enough to adapt your strategies to ever-changing business circumstances.
Want to Learn More?
If you’re interested in learning more about different recommendation strategies and figure out which is the right one for you, please reach out to email@example.com
As we grow, we are increasingly customizing Findify to integrate seamlessly with the branding of each merchant site. We have changed everything from how pages are displayed to how shoppers move through Findify areas - and have even helped finish site development to get stores up and running!
"The one thing that really stuck out was Findify’s ability to work with us on customizing—meeting our demands." Richard Enriquez, Director of Marketing, White River
Findify’s four levels of customization:
We are often asked what the value of personalization is for an ecommerce store.
Here at Findify we talk a whole lot about personalization. We should - our artificial intelligence is built to personalize the online shopping experience – for everything from search results to collections pages and recommendations. So, we came up with this simple overview to show its effects:
Searching and browsing without personalization
The standard search engines embedded in shopping platform sites uses proven, open-source technologies such as ElasticSearch (Bigcommerce) or Solr (Shopify). Basic search results are generated from product tags or product descriptions only and collection pages bring up products in either the same order for everyone or a random order.
More advanced search technologies, like Findify’s base search algorithm, take account of further factors, such as how popular a product is and whether a merchant wants to promote a product when choosing what to put in search results.
The effect of personalizing the results
Findify’s artificial intelligence considers factors that are totally individual to the current shopper. Things like:
Based on Findify’s analysis, the search results look like this:
The results speak for themselves
As you can see, the personalized search results give a much higher chance of the shopper buying and converting into a sale for the merchant.
There are thousands of knobs you can tune to make search results better and at Findify, we’ve been teaching the machine to automatically make the best ranking decisions for you. Your customer John will see different products from your customer Jack because Findify shows the products that are most relevant to each.
Findify puts the right products in front of shoppers – in their search results, on the collections pages they see and in the recommendations that they read.
For more information about how Findify personalizes each individual shopping experience, get in touch and we’ll show you around.