We launched a new site. Please contact us if you run into a problem. See known issues.

Personalization: How to Increase Returns and Build Customer Trust

In Summary

The internet, at its core, is a faceless, sterile producer of HTML gobbledygook (letter vomit). It’s only through building great applications and collecting data that we can get a better picture of who our customers are and what they are looking for. But our efforts are only as good as what information the customer is willing to give to us.

The Case for Personalization

Let’s say your conversion rate is at 3%, and you would like to get it up to 3.5%. This represents a 16.6% increase in sales (provided your AOV stays constant). If you have a turnover of $20M/yr, you would net an additional $3M/yr of revenue. With a 10% net profit margin, you would line your pockets with $333,000 in cash. Take $100,000 of that and invest in continued personalization and optimization efforts, and you will reap tremendous benefits.

Keep reading as I’ll give you ideas (with numbers) on how you can achieve this goal.

The core of this article is focused on two key areas:

  1. Precise details of what happened. A video recording, evidence of problems, etc. The more you provide the less time the developer will have to spend.
  2. Exactly when this happened. The closer to the accurate timestamp, the easier this is to triangulate in the logs.

Free Offer: Identify ways to personalize your online storefront, for free

Before we dive into this subject (presuming you read chronologically), I want to make you an offer. If you use personalization on your website, ignore this. But, if you don’t, keep going. I am offering to do an audit on your website and present you with suggestions on how you can implement personalization. This is a service to drive value to you, and as such, it won’t cost you anything. The only request is that I’d love to write about our findings in our next article—that’s it (if this is not an option, I’m still happy to chat). The only thing you have to do is pick up your phone (not that hard) and call me: (913) 705-0004. Just do it. I can’t reach through the phone and bite your ear! 😀

What is personalization? Why should we care about it?

The action of designing or producing something to meet someone's individual requirements.

We put names on bachelor gifts, write notes on Christmas cards, and license plates (ESCUZME or TIMELESS on a Delorean). While we have built personalization-heavy websites of this type, that’s not what I’m talking about.

Here’s an example: we can differentiate between a pre-recorded political campaign phone call and a legit person on the other side.

Since we can’t replicate in-person personalization, we can do it in subtle but powerful ways:

  • Tweaking search results to boost results that are helpful to a customer.
  • Automatically applying filters for product listings to weed out unhelpful products.
  • Arranging slider / hero images on the home page to push them toward relevant sales.
  • Adjusting the product recommendations based on order or browsing history.
  • The best: highlighting rain products when the visitor’s physical location is raining (I’ve seen this, and it’s really good!). Or sending an email to visitors based on their geography and weather forecast.

What are the results of well-implemented personalization?

Your conversion rate will increase. Your average order value will increase. These are the magical numbers that almost completely represent the health of an online store.

But how we get there is through making the customer journey easier: 1) morph their experience into what they want at the time and 2) accelerate the process to purchase.

What are the results of poorly-implemented personalization?

I recently received this piece of mail. It honestly caught my eye as I saw my name on it—but Anna? I know several people whose name is Anna, but my wife’s name is Elissa. I applaud their attempt, but it fell quite short.

I can’t say I would have been tempted even if the name was right, but... this sealed the deal. Personalization done wrong will cost you sales. You will point customers to the wrong products, highlight the wrong things or make a mistake at this level.

Here’s another example of personalization gone wrong. I received this email:

Hi Joseph — I hope you are having a great week. It seems you are around the Olathe area, it must be mostly cloudy today in the day time (47 F). We're inviting you to an exclusive session below.

I like what you do, and I was impressed when reading about your company. If I understood it right, SwiftOtter seems dedicated to solving complex software problems. It sounds fascinating, and I look forward to learning more.

I'm David, Sales Consultant, at XYZ Company. We've helped similar companies recently, and reserved a few exclusive online sessions to present how we've helped others. Please accept our session invitation (attached deck here) to learn more.

Have a great rest of your week!

When someone pops into your website for the first time, it’s like going to a grocery store on vacation: you are thinking “where are those marshmallows?” Then, you do your best to group by category (overhead signage on each aisle) to locate baking goods. According to this example, the grocery store would understand that some of its customers are here on vacation and marshmallows (for those finger-lickin-good s’mores) are going to be important. This is difficult to do given physical space constraints.

What could a brick/mortar store do? They could have someone standing at the door asking “what are you here for?” Some folks will choose to decline help, but there are others who would say “where’s the marshmallows?”

What would this look like on your website? Sites that feature car parts have a Year/Make/Model selector. Beauty brands often have a survey to ask questions about skin/sensitivities. MarlowWhite.com (a military uniform distributor) has a similar selector.

Website option selector

Popups are annoying, so please use those sparingly: only when the user has shown engagement with the website can they be deployed. Just below the header or integrated into the menu system are two great places to request this information. It will be seen.

A fairly-recent feature on Google is highlighting text on the resulting page. This is a collaboration of the Chrome (and Safari) web browsers with the search engine. It’s possible for a developer to fetch the highlighted portion of text from the value in the URL proceeding: `#:~:text`

Accelerate the process to purchase.

We, as customers, are not usually on a website for the viewing pleasure. Sadly, many merchants perceive that customers love their website so much that they wish to spend hours a day (in fact, some even build a mobile app to download).

The more a customer has to hunt on our website for the product they desire to purchase, the more likely they are to think, “Forgetaboutit! I’m going to Amazon.” And we just lost a sale.

Helping a customer find what they want as quickly as possible is a sure way to increase your conversion rate. How do we know what they want unless we ask them?

What are potential applications for personalization? / Where can we deploy personalization?

Category / Search

The most common location for personalization is search results. We want to help customers find the perfect product as quickly as possible. Once visitor preferences or information is available to the display engines, it can then positively affect results. Both Adobe Commerce Live Search and Algolia are examples.

Product recommendations

These are one of the most under-looked areas for enhancement. When someone lands on a PDP, chances are high that they won’t find the exact product for which they are looking. The couch will be too expensive or the color wrong or the fabric not durable enough. We risk the visitor clicking back, and the sale is lost.

Nosto and Adobe Commerce Product Recommendations are the two solutions that are capable of handling this.

Hero images / Content changes / Store promotions through banners

These changes are not as customary as the previous two—but they are just as effective and, quite frankly, are easy to implement.

Google Optimize is my go-to solution here. You can use the free version to implement up to 10 personalizations. Of course, you should first A/B test the changes to ensure that there is a significant improvement. The only downside to Google Optimize is that you cannot target specific Google Analytics audiences (which would be super nice!). Google Optimize 360, which has this feature, among others, weighs in around $150,000 per year.

If you are engaged with a development team, you can also work through these changes with them. However, it’s much better to test your personalization theories with Google Optimize. But, if you need data fed to Google Optimize, in order for it to work, you would need a developer to aggregate the data to turn on the personalization engine.

Promotions can be a powerful way to motivate a quicker response.

Marketing emails

What is this customer interested in? Send emails related to these interests. Will this take extra time, for sure! But is it worth it? Only time will tell thanks to your A/B tests.

Just looking at these once-in-a-while emails as your only marketing emails is missing at least 50% of the equation.

Oh, and don’t forget, the customer adds their email multiple times on your website. They should never have to re-enter it again. Seriously, this is the 21st century.

I did a podcast with Andrew Davis a while back, and he focused on customer touchpoints. When does the product ship (yep, covered)? When does the product arrive (sometimes “yes”)? When does the customer start using this? When are they most likely to be thrilled with the product?

But even when the product arrives, are there usage instructions? What is their onboarding experience? We all know how painful Ikea furniture is to assemble. Could that be the same story for your products?

You are likely thinking, “Wow, the time investment will be significant.” You are correct. Remember, this magazine is similar to those old Sears catalogs where you take an idea and try it—if you implement one thing from each of these Empowering Merchants magazines, you will see tremendous results (I hope you gain more than just one thing).

One of the more basic ideas of personalization is this:

Dicks Sporting Goods Personalized Promo Offer

Once the association between a customer’s email and a browsing session are made, your tools of choice should allow you to now target this customer based on their browsing.

Klaviyo is a great example of a tool that can do this very thing. It is easily integrated into Magento or Shopify. Pricing is super reasonable. You’ll be thrilled once you implement a solution like this. It’s somewhat possible inMailchimp, but I find that Mailchimp is less focused on ecommerce and is used by a wider group of businesses.

Using a platform like Klaviyo would allow you to:

  • Send instructions X days after the product has shipped.
  • Send a review request X days after the product has shipped. The goal is to have the email in their inbox at the moment when the customer is most excited about the product.
  • Abandoned cart email. Try to add value to the email.

I know this is tough, but do your best to keep the marketing content to a minimum. Instead, focus on “how can we empower this person to be the happiest user of our products—ever?”

If I could harp on a pet-peeve for a moment (this one is legitimate, I promise), please do bear with me.

Why is it that when I place an order, I get an email from `[email protected]`? Does that sound approachable? Not in the slightest. Send your order confirmations from a real email. If a customer has a concern, all they have to do is reply to the email, like they would a concern regarding any other email they receive.

How to build personalization

Personalization only works as well as the data it is fed. There are very few (if any) platforms that give insights from one website to another. Thus, when someone arrives on your website for the first time, the amount of data you have about them is next to nothing. As they interact with your website, you begin to build a picture of who they are.
Beyond this, rarely is personalization a good choice through just “machine learning.” While machine learning will augment your choices, it is not a substitute.

Where to start?

The most important question is “what’s wrong?” Why do you want to change something?

The easiest place to start is with customer feedback:

  • What is returned most often?
  • What do customers say they have the hardest time finding?

You will notice that some of the above can be solved by clearer information on product or category pages. But these are often bigger problems.

Personalization can solve problems. This is easier because you are working within an existing infrastructure.

But another place to look is what hasn’t yet been stated. Customers won’t tell you that they don’t think the order of the product listing is useful. They won’t tell you (maybe they might?) that they wish they had quick access to instructions. They won’t even consider how your website tells them whether or not this product will work for their use case.

Instead, they will happily (or maybe not happily) browse the website.

This is now whitespace—a blank canvas. It is up to you, the artist, to fill in the blanks. My goal is to inspire you with what to start drawing.

I suggest you ask the question: “What does it take to provide a personal experience?” If you have a brick/mortar store, see how your salespeople interact with the folks that come into your store. If you don’t, work through this list, asking this question: “How can I add a personal touch to these key places?”

  • Marketing emails. How can you make this more like an email from a salesperson to the customer, instead of what the customer knows is a blanket email. Review the Dicks Sporting Goods email a few pages back. If the customer has ordered before, consider thanking them (don’t we all want to support companies who are “hungry” for our business?). Klayivo can even predict when the customer will purchase next—thus you can preemptively get in front of them.
  • The "thank you for your order" screen. Instead of feeding them upsells (who is going to buy more now, unless it’s all free shipping), how about taking this opportunity to direct them to being effective users of the product, assuring them the product will arrive on time, and that you stand behind it?
  • Transactional emails (order confirmation, shipping confirmation). You can stuff more products in front of the customer. There are occasions where this is appreciated. However, if you inscribe wedding party gifts, and the customer ordered knives for their groomsmen, consider highlighting an article about wrapping + gifting them correctly (if this is a thing).
  • Home page. Highlight the categories that customers have shown the most interest in through the hero image. Present similar or related products.
  • Customer account page You know what this customer loves—share content with them that they will love.
  • Category/Search pages. I put this one last because it’s the most obvious. Customer information and their behavior should at least partially reorder the results that are shown on these pages.


Before we jump into this section, customizing based on customer input takes development effort. Most solutions are geared around actions as this is predictable from merchant to merchant.

I’d like to take the previous example of MarlowWhite.com. To be frank, they may be using personalization on their website as you won’t ever find an indicator saying “you are enjoying a site that is tuned to your needs + tastes.”

Marlow White is an excellent example of requesting data from the customer. They could build out the selections into their menu system, but instead, they give customers the ability to select what matches their criteria.

No matter how you get this information, save it. I’d say it again if I can. If they aren’t logged in, save it to their session. If they are logged in, save it to their account. To ensure GDPR compliance, make sure the customer can edit it in their account if they are logged in.

  • Upon registration, the customer’s selections are transferred from the session to their customer account. This should be pushed out to any relevant platforms, like Gorgias (support system), MailChimp or Klaviyo.
  • Your development team should include this information for third-party tools like FullStory/HotJar (tools to watch your visitor’s sessions to identify usage problems), Gorgias or LiveChat (help desk solutions), or Algolia.
  • Update your workflows and automations to adjust data.

If you sell products geared to the person (usually one ring-finger size per person, one military rank per person, one clothes size per person, etc.), consider warnings to alert the customer if they are straying from what is normal. The warning should not prevent action but ensure they are aware of proceeding.

This is a huge benefit to you as the merchant: you reduce the number of customer inquiries and returns if the customer makes a mistake.


This type of personalization is founded on what actions the visitor takes. They clicked on Add to Cart on this page. They viewed this product. They placed an order.

Don’t get me wrong; these are helpful. But if the budget exists, do your best to personalize based on input provided by the customer.

These actions paint a broad picture of what the customer is looking for: if they are looking at couches, you’ll see this. Here’s where it falls short: if you believe this person is just looking for couches, perhaps you would show them more couches? But what happens when they have already purchased the couch, and you keep showing them more recommendations? No effect (or they are bothered: Wayfair is a perfect example).

You will most likely start here. If someone is interested in couches (and you have control over resulting personalization), think of what else they might like next. Be one step in front of them.

Person-based personalization is ideal because you are working with specific selections made by a human. This gives you far deeper insight into who they are.

Tools for personalization

Website personalization tools

Google Optimize


I love Google Optimize. It’s easy to get started (drag and drop). Reporting and metrics is tremendous. This ties nicely into Google Analytics reporting so you can apply segmentation on your tests—this is a powerful feature.

If you wish to pass in additional details, your developer can easily customize this. You can have 5 personalizations and 10 tests. The only disappointment is that you have to pay $150,000 a year (minimum) if you wish to target Google Analytics audiences. That’s a lot to swallow.


Starts at roughly $3,000/mo (or more)

We have used this in the past and was very impressed with its performance. Since they eliminated their capable free tier, we exclusively use Google Optimize.

For reference, Optimizely is more than $36,000 a year. There are a couple of other products, that I have never used, which have more palatable entry points:

  • Convert: $700/mo (with yearly commitment)
  • Proof (useproof.com): $80/mo

Here are some other options you can review:

  • Qubit (no pricing listed)
Adobe Commerce Personalization

If you gain enough value from the features that Adobe Commerce offers, then the sticker shock will be overcome. You (and your development team) must analyze these features and make an internal commitment that you will use them. If you don’t... you’ll end up with a significant, yearly donation to Adobe—and you have better things to do with your money.

We know Adobe Commerce inside and out, so it’s critical to call out these features.

Customer segmentation

Every customer is assigned to a customer group. Each group has many customers, but each customer is only in one group. I bring this up because many people confuse a customer group and a customer segment.

Customer segments are calculated. A customer can be in multiple segments.

Segments are powerful. They allow you to identify customers that match conditions:

  • Customer address
  • Customer attributes (name, customer group, newsletter subscription which ties into Mailchimp)
  • Totals of cart items
  • Product purchase history
  • Totals of orders

With this, you can target dynamic blocks (along with cart / catalog price rules) at those who match the customer segment’s conditions.

You can also export the list of customers so you can use it for targeting in third-party applications. Be careful of applicable laws.

If you don’t have Adobe Commerce, you can get much of the same functionality by using Mirasvit’s “Customer Segmentation Suite for Magento 2.”

Product recommendations

Recommendations are the first of Adobe’s move toward “microservices” for the Adobe Commerce platform. This feeds activity-based data to Adobe Sensei and the recommendations are returned.

There is no full equivalent of this for Magento Open Source... except for some third-party offerings which we will discuss below.

To be honest, I can’t say whether third-party offerings are better or worse than Adobe’s as I haven’t been able to do a head-to-head shootout yet. I will say that while most of the third-party systems will accept additional data inputs, Adobe Product Recommendations won’t.

Adobe Live Search

This is a brand-new feature that was just released. This seems to be a competitor with Algolia. At this point, I’d recommend Algolia. This will likely change over time, but Live Search is too new to say exactly what to think.

Algolia (Search / Category Display / Product Recommendations)

I have yet to find a better search tool than Algolia. It’s incredibly powerful but also reasonably priced. Its feature set is rich, and the results are quite customizable. The Magento integration just works “out of the box.”

We were working with a merchant who was using a different “bees knees” search provider. A funny side note: this search provider had six case studies on their website. Four of the featured websites had switched to Algolia. We also made the recommendation: installation was fast, and they were impressed with the results right out of the box.

Algolia can also take over displaying products on category pages. The benefit is you can use their merchandising and personalization tools to adjust the order of the displayed products.

Note: neither Algolia’s Magento module nor Shopify app supports product recommendations yet. This is likely 2022 or beyond.

Nosto / Vue.ai

These two solutions are pretty similar in their offerings (on the surface). They both offer product recommendations. Pricing is not available.

MailChimp / Klaviyo

We use Mailchimp for our email marketing. It does everything we need it to do and way more. Frankly, we don’t use automations like we can. If you don’t use your email marketing platform to the fullest, that’s ok. It should be on your to do list, even if you can’t get to it until next year.

Klaviyo is on our watch list. We are considering switching to it. Klaviyo is slightly more expensive.

For Magento, the Mailchimp integration works, but it can be hard on CPU resources.


We talked about collecting data from your customers in our last edition of Empowering Merchants. TypeForm was my recommendation. The forms are beautiful and are easy to build with logic. You can point someone in the right direction and then collect which direction they were pointed to.

Depending on your GDPR restrictions, your developer can customize the form to add anonymized details which can then be passed back to your customer data through a callback.

How do we know this is working?

The worst thing ever would be to invest time or money into personalization only to have it backfire with a drop in conversion rate.

My suggestion is to focus on one piece of the puzzle at a time. You can watch the metrics coming back from this source.

Start small. Personalization is a process, and it’s not a destination.

For example, your email marketing automations will tell you the revenue for each automation or campaign. Algolia won’t tell you revenue numbers, but your developers can add a flag to enable/disable personalization based on a flag from Google Optimize—then you can create segments in Google Analytics for the A and B versions and compare any desired statistics.

I will note that Adobe Commerce’s Recommendations / Live Search have limited reporting. These tools are either enabled or disabled, so it’s a matter of trusting that Adobe has done their due-diligence.

What types of custom data can we feed to the personalization engine?

We aren’t Facebook, so we are unable to get the depth of knowledge about each visitor. However, we can start collecting data as soon as they set foot on our website—and the longer we keep them, the more we get to know them.

One caution I present is GDPR. It’s up to your lawyers to determine whether or not you need to comply. But if you do, the data that you collect per customer needs to be viewable/editable.

Survey / form results

This is my favorite way. The key is to tell your visitor why you are asking, putting them back in control.

Go to MarlowWhite.com and roll your mouse over the Army menu. This is so apparent that an explanation isn’t even necessary!

But, what about putting a drop-down menu just below the header that says “To better customize this website for your needs, what type of business do you represent?” You won’t get 100% of people providing details, but it will be quite a few. Popups are also a possibility, but I wouldn’t recommend using them unless the visitor shows signals that they are engaged (they are viewing the page for 30 seconds or clicking into another page).

The Benchmade Knife Finder is also a brilliant use for this. It’s not as obvious, but Benchmade is quickly getting the idea of what this visitor is looking for.

Email address

The second your customer enters their email into the checkout, live chat or anywhere else, you should capture this. In fact, the Klaviyo Magento extension does just that. Once you have the email, you can use this to connect the dots across the board. Remember, your developers should rarely, if ever, write the email directly into the HTML source code. Instead, use an anonymized token for cross-platform communication.

Placed order

Once you have a customer, you now have a plethora of information about them. Do keep data privacy in mind. However, you have their address and can target them geographically in a reliable way (instead of relying on their IP address’ geolocation).

Website interaction events

You can combine these events (Add to Cart, live chat, contact form submission) with additional data, like what product or category they are on. Most personalization engines discussed here do this out of the box.

Other ideas

  • Specific features used
  • Type of device (this is not as ideal for person-based personalization)
  • Landing page
  • Product path
  • Products / categories viewed

Joseph Maxwell

President / Senior Developer at SwiftOtter - @swiftotter_joe