Marketing is often seen as either a nice to have or a quick short-term fix but it is important to recognise the long-term value of a solid marketing strategy and the return on investment (ROI) it can deliver. In our experience, our clients often have a clear idea on what they want to push, how they think they want to push it and what the messaging should be; this is great as it provides a structure for the campaign, with clear objectives outlined; however some campaigns also require a bit more consideration when it comes to desired audience response. To help you to ensure you are maximising the return on investment from your campaigns, this article focuses on split-testing and how it can be used to ensure your marketing campaigns are spot-on for your audience.
What is split-testing?
Split-testing is usually defined as the process of segmenting a database into smaller groups and sending different marketing messages to these groups. Both should have the same desired outcome e.g. enquiries or registrations for something, so that they are directly comparable. For the purposes of split-testing, it is important that the database is split at random; systematically splitting the database e.g. by gender, age, location can create skewed results and this is the process of targeting messages rather than split-testing.
How does it work?
Split-testing can be used on almost anything if you think a little laterally; for example, although it doesn't use a database, you could split-test your adverts to maximise the return on investment. You could design 2 adverts and test 1 each month in a series deal. Each should have a code to allow you to measure the response rates. The more responses, arguably the better return on investment (Please note this is very generalised as we shall explore later in the measuring ROI section). The same is true of email marketing; take the same base copy, change the email subject, the order of the information and/or the emphasis of the stories and send these to half your database each; alternatively, you could try one with benefit statements and one with a discount to see which delivers more revenue.
Why should you do it?
Split-testing can be used to maximise the return on investment for your marketing. This could be of direct monetary value in terms of revenue e.g. increased sales or reduced costs e.g. lower expenditure; or it could be an intangible benefit e.g. increased brand awareness, trending, higher website stats, increased social media followers etc. which could in turn leverage more sales.
How is it measured?
As any marketer will tell you, measuring return on investment is difficult and is by no means an exact science! How can you measure the value of more website hits, or twitter followers in the same way you measure the sales from a special offer advert? Well it isn't easy, but it's not impossible either; provided that you view this as one stage of the sales process, rather than the whole sales process.
Split-testing with a campaign can be a great way of enabling better ROI measurement. The basic rule is change as little as possible, keeping much of it constant, only changing specific things. Once you're happy with your messaging, decide on a time-frame, then run these for a set time. Using dedicated 'codes' can help with measurement. At the end of the time period, collate all the information you can, then work out the increase in XYZ (e.g. sales / enquiries / sign-ups etc) versus the expenditure. This will give you a ‘broad brush’ ROI from which to make future decisions.
An example
Company A is launching a new product; they want to know which message will help them sell the most products - an 'introductory price' or a collaborative offer with another product. They are sending out emails to their existing database with the objective of generating direct sales.
Both the emails should have the same 'base' copy, introducing the product, an image (if applicable) and the key unique selling points.
- Email 1 - Introductory price - stick to the base copy, keep it friendly and encourage purchase. Add the introductory price, with a unique code. If possible, it would be good to use a trackable url, or email software with tracking to determine the number of people that opened the email and clicked on the link, with an option to measure the number of codes used to redeem the offer.
- Email 2 - Collaboration - again, stick to similar copy, but change the call to action to relate to buying two products together, using a second unique code for your measurement. You can monitor this by looking at the number of sets sold, as well as the number of codes redeemed.
Working out the value
In this case, measuring the ROI is not as simple as Email 1 or Email 2 delivered more sales; they will need to take into account the cost to discount against sales.
For example, the new product costs £20.
- Email 1 offer costed the company to send the email to half of their database (£20) plus £5 off per product. The offer sold 100 units.
- Email 2 offer costed the company to send the email to half of their database (£20), gained them the sale of two products together worth £40 originally, but with a total discount of £10 making the actual sales value £30. The offer sold 80 units.
In this example:
- If we look at unit sales, Email 1 is more effective because it sold 100 units versus the 80 sold by Email 2
- If we work out revenue generated, Email 1 sold £1500 and cost £520; Email 2 sold £2400 and cost £820
In this case, although Email 1 generated more sales, Email 2 generated more revenue. Short term, the costs for Email 2 were higher, however the company had already accounted for this revenue loss by offering the discount in the first place. The intention is that these one-off purchases will turn into loyal purchases and future revenue at the full price of £20 will follow to some degree.
Without the split-testing it is impossible for a company to place a value on different offers. Split-testing can really have some surprising results!
Top tips:
- Your split-test will only be valuable if you have a good database. It isn't worthwhile to test it on 10 people for example
- Make sure you can afford the offers you are doing. Don't undermine the value of your product by offering rock-bottom prices. Make sure you are giving people the feeling of a good deal, without making the full price seem extortionate
- Make sure you can measure it. Add codes, tracking links, specific calls to action and tags whenever you can
- Do a cost-benefit analysis. If you were never going to do a discount, don't split-test for the sake of it.
- Only use split-testing as part of your existing marketing campaign to ensure you are delivering what your customers actually want; the more elements of your marketing toolkit you use, the more effective your brand building can be and the more relevant a split-test will be. That said, a one-off split-test can also be effective for delivering relevant statistics on a specific marketing element to make that as effective as possible, even if you don’t have time for anything else.
- Make sure you are prepared. If it proves successful, you have to be able to deliver on your promises otherwise customers won't be impressed!
Comments