Yes we know it’s not easy to optimize an entire ecommerce site – but a few best practices and blunders have given us hints on “easier” ways to approach optimization. Here are some tips on what NOT to do.

1. Recommendation Implementation Galore: Slow down with the recommendations! Very often when you see a page you assume that 50 different elements can be changed or shifted in some way, shape, or form. And that may well be the case. But, SLOW DOWN! Don’t go changing everything at once! Record what changes you want to make to a specific page, then make a single change and monitor the progress before moving to the next change.

2. Optimizing blindly: A recent client told me that they make changes and as long as the conversion rate doesn’t go down, they think it’s okay! But we don’t want to make changes for the sake of making changes – we want to track positive progress of the site! Although conversion may not decrease, there are other factors to consider.

Monitor your analytics closely to see did exits increase on the page? Did users navigate to one particular page more than the other as a result? Did firefox visitors respond differently than those who browse with IE? Were users using different browsers more likely to go to specific pages more than others?

Everything must be calculated and analyzed before determining how the change impacted your site’s performance.

3. Optimizing without testing: Testing goes hand in hand with any implementation. Never implement if you don’t have a plan for detailed plan to deploy A/B or multivariate testing. Without testing you’re limiting the results. You can never see what change works best, what change out performs the next, and you can never explore the different scenarios for a single element, etc.

We are advocates of light testing, meaning, rather than testing millions of scenarios, do your homework so you only have to test a few scenarios (minimizing the process tremendously in order for you to see results faster!). Your few scenarios, though, have to be well thought and calculated in order for you maximize the effectiveness of the changes. The point is, even if it’s minimal testing, it’s crucial.

4. Listening to your Colleagues over Gut: If you’ve done your research and taken all the steps mentioned into account, move forward with making changes on your website. Some team members may argue, and don’t get me wrong constructive feedback is great because you can always tweak your changes and limit testing scenarios. But ultimately, when you’ve done your homework, hold to your guns and don’t waiver!

5. Accepting A Half-Ass Job: Never accept something half done, or not perfect. If you specified an element to be designed or implemented in a certain fashion on your site, don’t accept the second best thing. The design must be done EXACTLY the way you specified before moving forward. In our experience, outsourcing work can be daunting, especially to countries outside of the US mainly because of language barriers. But more importantly because outsourced employees (like any employee) may try to get away with work done poorly as a result of limited monitoring capabilities from your end. Never accept anything but the best.

Conversion optimization is tough but necessary for any website. Your website is like a basket with holes, you will continue “dropping balls,” if you don’t patch it up. It’s a continuous effort and should never be ignored. What are some best practices that you’ve noted?