Why you should be testing
Written by Matt Lawrenson
Why you should be testing (plus 7 helpful tips to get you started)
Your sales and ROI (Return on Investment) might be on the up now, but a little dash of pessimism in marketing goes a long way to fending off complacency.
Think of testing as a way you can improve what is already working very nicely! - who wouldn’t want that!?
Don’t know where to start? Looking for a shot of inspiration for your next test? It could be a new creative, offer, segment or format - or any one of the list of variables you can tweak to get a better response.
But how do you know your test worked? Was it that new strapline with your offering positioned differently, or a freak snow-storm in March, that cleared your shelves in-store?
Enter the ‘control’ – ‘the reliable one’ – ‘the banker’. Your control is the baseline against which all tests will be measured. It should be the best performing incarnation of your offer or message to date. As it will not be changed prior to the test, it is then used as a benchmark to measure how the variables you change, affect performance.
Measurement is ridiculously important in testing, not least because it allows you to know if your test was succesful.
Analysing your results sufficiently lets you know why and where it was successful and with whom – invaluable information that can influence your next test, for a cycle of incremental continuous improvement!
Hopefully you’re ready to start testing and seeing improvements in your results.
To help you on your journey, I compiled our top 7 tips below:
1. Commit to your testing strategy: If it isn't broken now, it might be one day!
2. To begin with, choose from the seven key marketing variables
According to the IDM Guide on Testing, there will be 7 key facets that make up any piece of your marketing. Stick to testing these, and you can’t go far wrong.
- Target Audience
- Format… check out some ideas on Creative Formats
- Creative (Information Design)
- Response Mechanism
3. Agree your objectives and consider other potential desired outcomes
Testing is not just about gaining increased response. Tweaking one of the variables above could have a range of effects.
If you tested a new format for mail, could you gain a better cost efficiency? Consider hybrid mail – you could send the same personalised, relevant content to your customers, but make massive savings on postage.
Test hybrid alongside your standard mailing, with the latter acting as a control, and if you your response levels are maintained or grow, either way you are winner!
4. Establish your control: What is your ‘banker’?
If testing was made illegal tomorrow, what would your go-to, best performing piece of marketing be? This is your control and the base against which all tests will be measured!
If you’re starting from scratch on a new product, or new venture entirely then case studies, awards and different industry sources like Royal Mail’s Mailmen resources page are a great way of benchmarking your industry and setting a baseline against which to judge your efforts.
5. Try not to test more than one thing at a time: “Something worked, I just don’t know what!”
If you have to conduct multi-variable tests, the key is ensuring the presence of a strict control. And if you are going to perform multi-variable tests, they should be conducted on a strict statistical basis with a rigorous methodology that lets you identify the variable that had the effect.
6. Make sure you can identify which responses come from the test cell!
If you’re sending directly to an individual, this will be more straightforward, as they can be can be isolated and identified easily, within segments, cells and lists.
If you’re conducting tests in advertising, consider how someone will respond to your messaging. Perhaps a unique URL, phone number or suggested search phrase could identify a geographic region, or other identifier.
7. Observe responses, analyse results & identify your key learnings.
Once your test is complete ask yourself these 5 questions:
- How did your test cell perform against the control?
- Can you spot any correlations between successful and failed tests?
- Have you missed some external force that affected results?
- How will this effect your next mailing?
And of course…
- What can I test next?
Embarking on your testing journey can appear to be a daunting task. However, when you’re tweaking or finessing one variable at a time it appears much less onerous. Consider it in terms of a process to achieve incremental gains, improving what you do in multiple areas by 1% each time, not revolutionise what you do overnight.
Now get testing!