Be clever…measure!

I guess we would all recognise this four-step path to improved online sales performance:

1. Plan an optimisation project which explores the needs and wants of your target audience
2. Commission a UX partner to generate a whole heap of customer insights and recommendations to improve your conversion rate
3. Implement the recommendations with the design and technical teams
4. Monitor the improved sales performance of your target process

Hey ho, your conversion rate increases by 10%: job done. Or is it?

What exactly have you learned? Which changes made the most impact on your conversion rate? Did everything that you implemented have a positive effect? You have your overall conversion increase, (which more than pays back the cost of research, design and build) but relatively little idea what’s caused this. Could you have achieved an even higher conversion rate improvement with the same resources?

By using a structured roll-out of your site improvements, you can measure the impact of individual changes and use those learnings for future optimisation. In a typical project, not everything you do to improve the sales process will have an immediate, positive effect on conversion (nor should it, you’re also thinking about wider factors like brand preference and customer satisfaction). But it’s really useful to know which individual improvements create the largest commercial effect. My advice is:

• A/B test big changes with your process

• Multi-variate test pages that have a number of changes

• Work on one page at a time

This way you will understand the effect of every change you make. You might be surprised at how much more you can squeeze out of conversion rate improvements.

Plus you’ll look a lot cleverer if you understand why you’re now exceeding your sales targets, rather than just being able to read out the numbers.

24/05/11 Stuart Lamb said:

Nice blog article Nicky. As a seasoned measurement professional I wholeheartedly agree with your advice! So many clients run headlong into 'improvements', often personally liking a new design and feature and think that all consumers are like them - then wonder why it doesn't deliver the uplift. 'Test and Learn' is definitely the way forward! Stuart Lamb Manager, Measurement Practice Syncapse Social Media Management

25/05/11 Lauren Robertson said:

I agree, a much more measured approach so that changes can be analysed is much more useful. Clients are very hasty to want to change a vast array of items (for what benefit I ask a lot of the time) without taking much consideration to the long-term impact or monitoring the change. It's much more powerful to have real analytical evidence to back-up your changes rather than a vague number! Good post!

26/05/11 Daniel de la Cruz said:

Great post Nicky! Good to see people talking sense and reminding ourselves that there is a reason why we do all these things. I work in the space of corporate communications and I often ask myself: 'How are we going to measure the success of this website?' I guess it's more clear-cut for transactional sites... but maybe one day I will find a clever way to measure success of a corporate website too. Keep up the great posts, Nicky!