In the lead-up to hack week, any staff member can pitch an idea. These ideas might take inspiration from our roadmap, involve trying out an emerging technology, or simply aim to resolve a pain point. Some years have a theme; this year, teams were encouraged to build both for and with AI.
Hackers – including engineers, data scientists, and QAs – form small groups and adopt these ideas, or come up with their own schemes. Each group has four days to work on their project and prepare a short video presentation. Judges are appointed from across the company to pick their winners in three categories (most wanted, most innovative, and best video). There is a live audience vote in two further categories (best team name and best hack).
2023 saw an exceptional crop of hacks with some very impressive videos. Let’s get to it.
Hack name: HappierKat!
Team name: That feeling when you look down expectantly for the second half of your sandwich only to realize that you’ve already eaten it
The winning hack in the most wanted category takes data submitted in surveys, pages and forms and stores this as insight data. This unlocks various capabilities, such as being able to customize the email sent to form respondents. By pushing submitted data into an insight data collection, we can also use this data in program decision nodes to set marketing preferences, based on checked values in the submission.
As well as being picked by the judges, the Team That feeling … sandwich also saw off strong competition to be voted best hack by the audience.
Hack name: GPTranslatortron: Automate translations for in–app help, app translations, campaign translations, and perhaps even chat
Team name: Lost in Translation
Our platform is already available in 15 languages, but can we use WinstonAI to automate translation as we add more features? Even better, can we offer translation for the text that merchants input into Easy Editor? Can we enable translation in chat threads? GPTranslatortron delivers a resounding Yes to all of these questions.
Hack name: Dotdigital Automation Alchemists: Using natural language to automate Program generation using GPT4, RAG, and custom models
Team name: Team Automation Innovate (Team AI)
Inspired by the idea that program building could be automated to save time and cost, this hack explored the use of retrieval-augmented generation to teach ChatGPT about our programs. The team used Microsoft’s Azure AI Search platform to store and index vector embeddings, which then allowed WinstonAI to interpret a prompt containing a program description like:
At 1 pm on the day the contact is created, send the message “Hello there General Kenobi you are a bold one.”
That feeling when you look down expectantly for the second half of your sandwich only to realize that you’ve already eaten it
Arguably the longest and silliest team name in the history of hack week, That feeling … sandwich swept the board with over 40% of the popular vote.
Hack name: Watercooler Roulette
Team name: Culture Catalyst
A worthy runner-up in the best video category, this hack was all about team bonding and morale for remote teams. A series of Slack messages prompt interactions and conversations between employees who might not have the opportunity to meet in person around a physical water cooler.
]]>Split-testing allows you to compare and contrast the impact of various elements of your email marketing campaign. By measuring your subscribers’ reactions to these changes, you can optimize your campaigns to get the results you want.
Testing allows you understand which tactics get the best results, while helping you keep your marketing fresh.
Depending on what you’re looking to improve, there are endless opportunities for you to discover what drives your subscribers into action.
While there are endless variations of tests you can perform on your email marketing, there are five core elements you should be regularly testing to ensure you’re getting the best results possible.
Definitely the most common element tested by email marketers, subject lines are the first thing a reader sees when your email lands in the inbox. You should be regularly testing this part of your marketing to avoid stagnation. You don’t want readers to get so used to your subject lines that they stop engaging them and driving them into action.
Ways you can test subject lines include:
The pre-header text is one of the most underutilized elements of email marketing. As the first line in your email marketing campaign, pre-header text works in partnership with your subject line to drive subscribers to open your email.
By utilizing pre-header text, you can give readers a “sneak peek” at the in-email content, or personalize it to complement the offers in your campaign. By split-testing these tactics, you can see whether your pre-headers have an impact on a reader’s decision to open your email.
Landing in the reader’s inbox at the right time determines whether or not your email gets opened. If you always send your email newsletter at 12pm on Friday afternoons, you’re likely to be contending with a large number of other brands.
Varying when and on what day your audience receives your marketing prevents your marketing from becoming stagnant and also gives you a clearer insight into your audience. Maybe your target audience is more receptive to emails waiting for them as they commute to work or back home at the end of the day. The only way you’ll know is to test.
Getting your call-to-action (CTA) right in your campaign is essential to getting the results that drive revenue and growth.
In your email marketing campaigns, you need your CTAs to stand out. Thanks to split-testing, you can test the size, color, and text of your CTA buttons. Doing this will help you discover the style and copy that drives readers to convert. If you’re looking to improve your click-through rate (CTR) this is an essential element to be testing.
The content in your email can be changed in so many ways, so understanding what works with your audience is an insight you can’t underestimate.
To unlock this information, you should be testing your content regularly:
Every email marketer knows that split-testing isn’t just splitting hairs. It’s not done just to check a box. It serves a powerful purpose that can improve your marketing results ten-fold. But for you to reap the rewards, you need to ensure your tests are set-up for success
What are you trying to improve? Before you set-up your test, you need to decide how you’re going to measure your success. This may include:
Depending on what results you what to see, you may decide to test very specific elements. If you’re looking at open rates, subject lines and pre-header text are the obvious elements to choose from. Testing CTAs will help you improve your CTR, while testing the content and send time may deliver the results that help you improve your revenue.
You need to pick a realistic sample size to get meaningful results. Too small and you won’t learn a great deal, but too large might take too long to get your results.
For small audiences, you can test on larger sample size, such as 10% or 20% of your mailing list. If you have a larger database, testing on the same amount of people will require a sample size of about 5%.
Once you’ve discovered what tactics are most popular with your audience, it’s time to implement them. What has your test uncovered? Do your audience prefer visual, image-based emails or insightful, copy-based content. Are they more engaged with subject lines that contain emojis than personalization? Whatever insights your test has uncovered, it’s time to optimize your strategy to drive your results.
The only way to ensure your marketing continues to land with impact in your subscriber’s inbox is by frequent testing.
Just because changing the time of day your email arrives works after initial testing, doesn’t mean it always will. If you don’t change your tactics, your results will stall as readers become bored of receiving that same thing, time and again. The only way to surprise and delight your customers is to keep your material fresh. Continuous testing will help you achieve this.
Testing and tracking don’t have to be hard. We’ve got an easy-to-use worksheet to help your compare and contrast your testing results. You’ll be able to easily identify which strategies work for your marketing database and spot trends in your audiences so you can continually improve your marketing. Download the fundamentals of testing today.
]]>So, the question on everyone’s lips: what are the components that can and should be tested in an email marketing campaign?
Here are the key ones to consider:
It’s an obvious one to start with – and possibly one of the most important elements of an email marketing campaign. Ultimately, whether or not your recipient opens your email rests on the language, context, and relevancy of the subject line. Remember, people’s inboxes might be full to the brim – so standing out is key. We’ve spoken previously about how to write the best email subject line, but the only way to really be sure what will work best is to test! Plus, don’t forget the preheader text which you can also test. This handy bit of extra info supports the subject line in convincing readers to open the email.
The call to action is the component of an email that propels the recipient towards a conversion point, usually via a click-through. But which call to action works best for your recipients? Test different options to see if they impact results differently.
Here are a couple to consider:
The issue of whether or not to include images in your email marketing campaign is one that divides opinion in the industry. If recipients are likely to have images turned off as standard then there’s an argument for using images sparingly in your messages. However, if lots of your recipients have added you to their safe-senders list, then you could get better results by increasing the number of images in your message. The only way to know for sure is to test. And don’t forget to segment your database depending on what you find out! In any case, it’s always best to populate the alt text with information related to the image – giving readers some context should images not automatically appear.
The ‘from’ address is another element that can have a big effect on open rates. Is it better to use your business name as the from address or does an individual, maybe your MD, work better? A known contact can be particularly successful for B2B companies, for example, because it’s especially relevant to the recipient. For higher education institutions or learning organizations, perhaps a team name would be more appropriate. Test a few different possibilities and see what gets the best results for your email marketing campaigns.
Personalization is an aspect of email marketing that has, to an extent, been overused in recent times. Sometimes its effectiveness is muted and, at worst, can look spammy. Try using different forms of personalization to see whether it works for your recipients. Don’t just change the odd word or piece of copy, change elements of the creative as well. Or better yet, test out your messages and content on different segment types, identifying what works well for whom.
Psst!
If you’re an ecommerce brand, Dotdigital’s RFM personas make it easy for you to target the right customer group – based on the recency, frequency, and monetary values of their total purchases. Our automation tool can help you test which RFM group fits better with your message, through split-testing and then sending the email marketing campaign to the group that engages more. Just set your engagement criteria, whether it’s an open, conversion, or purchase. And while our affinity score tool helps you judge which product recommendations are more suited to your individual customers, you can always test whether your recipients in general respond better to personalized recommendations or bestsellers, for instance.
What happens when your recipients click through? Do they then convert or does your landing page put them off? Try alternative pages for different segments, versus a standalone generic page, and see whether this has an effect. It’s equally important to test your content and creative on the same landing page to see what combination works best.
The day and/or time of sending can have a big effect on open rates and conversions. B2B emails generally work better during business hours, whereas high street stores may want to send an email on a Friday to tempt buyers to visit their store the following day. When it comes to ecommerce brands, we all know what has turned usual trends upside down this year. With lots of people working from home, a timely lunchtime or evening email could secure a sale any day of the week.
For those of you who have been using email marketing for some time, you can always look back at your previous campaigns and see if certain sending days or times had a better result. You’ve probably already done some of your testing!
Every company’s audience or recipients will have different preferences or behaviors, so testing and segmentation really is the only way to find out what will work best for your business.
Discover more email marketing tips in our fundamentals of testing worksheet.
]]>There are a wide range of items you can split-test, but for reliable results, each time you split-test you should only change one thing. Otherwise you’ll never really know which change made a difference. For example, let’s say you’re split-testing your landing page.
You want to increase the amount of people that fill in a form. Here’s some of the things you can change on the page:
However, you can only test one of these things at a time. That doesn’t mean you can’t test them all. It just means that you find the winning outcome for each consideration, then work that winner into the next test. This way you end up with the most optimized outcome possible.
Just because you can only test one variable at a time, however, doesn’t mean you can’t test multiple versions at once. For instance, you can test out as many hero images as you want in a test. Just make sure the variable is in the same category. This is why we don’t tend to say A/B testing at Dotdigital. We’re multi-taskers. And A/B/C/D/etc. testing takes too long to say or write. So, back to our original question, what can you split-test? For your marketing and engagement campaigns, you might consider split-testing:
Or even…
That last one might not seem like a big deal, but consumer credit reporter Experian found that using emojis in their subject lines increased open rates by an amazing 56%. These are the kind of results you can start to see when you start experimenting with split-testing!
Of course, all of these variables can be split-tested with ease within Dotdigital. See for yourself in your account, set up a demo if you don’t already have one or download our ebook today to learn more about the how small changes can amount to big results when you start split-testing.
]]>Facebook Messenger is a great channel to use for sending operational, transactional messages. Already, we’re seeing how companies are supplementing or forgoing apps in favour of using Messenger. Unlike push notifications, companies can see if the notification has been read. Customers can also reply to the message if they need additional support. Combined with the engaging nature of this fun and rich channel and high adoption rates of 1.4 billion and counting, this makes Messenger the perfect channel to use for guiding your customers to improve application completions.
Problem: As a recruitment agency, it’s your business to gather as much information as possible from applicants. After all, if your clients ask for something, you want to have it to hand. But this can make application processes long and sometimes arduous for candidates. Or maybe you are a large company who handles recruitment internally and sees a lot of traffic, but not many completed applications.
Cause: It’s possible candidates are filling out applications on the move, which makes having the necessary documents or information to hand tricky. It may be that they need a break from the application or get distracted by a task at home or at work. Or, they might have a question about the application, or can’t continue due to a technical area.
Solution: More than likely, you are already getting users to create an account. This is useful so they can revisit their applications and save their progress. Simply add an additional step. Include a button that opts them in to Facebook Messenger updates. Briefly explain the value of this, that it will prompt them on additional steps needed to complete their application, such as a passport number, reference contact details, or a proof of citizenship or address.
Users will also be able to reply to these updates. For instance, let’s say they can’t complete an application due to a bug on the site or they are missing a document. Later, they receive an alert to say their application is still in progress. Users can easily reply to the message to let you know about an issue they are having. You can also use Facebook Messenger to alert them about upcoming deadlines, new vacancies, or if they have been successful in their application.
Problem: As a financial company, you require a lot of information in order to process an application for a loan, mortgage or overdraft and assess eligibility. But while you see a number of applicants begin your applications, there are a low number completing.
Cause: Much like recruitment candidates, it’s possible your applicants are on the go. This makes it unlikely that they would have everything with them necessary to complete the application. There may be complicated or ambiguous financial terminology applicants are confused by. There could be a number of roadblocks you haven’t considered or are unaware of. Consumers are today empowered by both technology and a wealth of choice for their buying decisions. Without giving your customers a way to speak with you, you will never know what these roadblocks are. They will simply go elsewhere.
Solution: Include the opt-in button to Facebook Messenger updates early on in your application process. This way, you will be able to handle any issues your users are having when they reply to your updates or prompts. This will also give them access to their own adviser who can help guide them through the application. Not only does this improve the likelihood of completion, it also provides an exceptional customer experience.
You can also use Facebook Messenger to improve application completions for:
Any industry who has a consumer base that might require a nudge or feedback to complete an objective or goal will benefit from incorporating Facebook Messenger to their messaging strategy.
See also: Putting the social back into social media with Facebook
]]>The only way to really get an answer to these questions is to experiment by testing and tracking sends with different variables; this is called split testing.
Split tests are perhaps the easiest way to compare ever so slightly different versions of the same email campaign for effectiveness based on key stats like unique opens, clicks or forwards.
Pretty much any component of your email message, here are a few ideas:
Subject Lines: Why not try pitting a longer subject line against a shorter one? It might interest you to find out if the word ‘Free’ impacts open rates or if personalizing the subject line affects the campaign in any way.
Calls to Action: Experiment with the positioning, style, and frequency of your calls to action to see which option delivers the best results.
Timing: Try sending on different days and at different times in the day. You may even want to consider testing a weekend send against a weekday one.
From Field: Do people respond better when there is a real name in the ‘from field’ or is it better to send correspondence from your company name? A split test will help you decide….
Here’s how you actually go about performing a split test once you’ve pinpointed the criteria you’d like to test:
Split tests can arm your marketing team with powerful data that can help shape and improve your future e-marketing efforts.
However, remember to only test one variable at a time. If you send an A/B split test with varying subject lines at different times in the day you’ll not be able to identify whether it was the timing or the subject line that contributed to the better/poorer result.
]]>