Conversion Rate Optimization Services

Conversion Rate Optimization Services 2014-11-03T16:18:44+00:00

You haven’t started A/B testing yet

I can support you with the following:

Already started A/B testing

I can support you with the following:

Analytics Plans review & creation

For you if

You have implemented or are thinking of implementing advanced web analytics tools like Mixpanel, Kissmetrics, Heap Analytics, Intercom, etc… These tools are event-based web analytics tools (by opposition to pageview-based tools like Google Analytics) and provide you with a much better understanding of what your visitors and users are doing with your product.

The challenges of events-based tracking

These tools enable you to track anything and everything, but many users are actually not tracking the metrics that will be useful for them to analyze in the future or are tracking the wrong events. It’s important to implement tracking correctly as early as possible in your implementation because:

  • most tools bill by events tracked: try to track everything you can think of, and not only will you have a dataset that is difficult to analyze and exploit but you will also be paying much more than you should.
  • you can’t track retroactively: that seems obvious, but I can’t tell you how many companies set up basic tracking, then when they need to do a serious analysis, they discover they weren’t actually tracking the metrics they need to decide on key business issues.
  • you need to set up tracking before you actually need to do the analysis: when you reach a point where you have to decide on key business issues, you need to get the analysis done right now. For that, you need to be confident your tracking plan is actually accumulating good data, from the start.

How I support you

If you already have an implementation in place, I review what you are currently tracking to make sure that:

  • the implementation is technically correct
  • all events that you’ll need to optimize your product and conversion rate(s) are tracked
  • no unnecessary event is being added to the mix, both increasing your costs and complexifying your data set
  • your tools themselves are optimally configured: funnels are well defined, etc.

If you are working on an implementation, I follow the following steps:

  • discuss with you your long-term business strategy and roadmap
  • analyze your website and product user flows
  • propose and validate with you all critical events and actions you should be tracking
  • send you a detailed tracking plan with all the events you should track, all the properties you need to push to these events, and I configure your tools for you

In both cases, I bring you value not only because I enrich and validate the analytics you will get, but I also save you precious developer time because my tracking plan details exactly what item to track and how to track it.

Testing benefit assessment

For you if

You are considering starting to A/B test some pages or elements of your website.

Why is this critical?

A/B testing is all the rage these days, and the mantra “Always be testing!” is everywhere. The leading testing platforms like Optimizely, Visual Website Optimizer or Convert Experiments all make it easy to start A/B testing with cheap plans starting at $19/month for 2,000 tested visitors per month.

Well, according to Optimizely’s own sample size calculator, if you can only test 2,000 visitors a month, and assuming your conversion rate is 5% (a very generous assumption), then you would need to design a variation which achieves a conversion rate of at least 7.5% (a +50% performance) to even see some statistically valid results.

If you don’t think you can design a variation that will get you this performance right from the first try, then you need to test more visitors, and you need to have the traffic to support more testing.

How I support you

It’s actually pretty easy to see if you have the adequate amount of traffic to support a testing program that even makes sense for you business-wise. First, use my A/B Testing strategy planning calculator for free. This will let you know how many days you’ll have to wait to get statistically valid A/B testing results, depending on your traffic and your current conversion rate, but also the revenue impact at different levels of performance from your A/B tests.

If you cannot get valid testing results in under 2-3 weeks, or if the revenue boost you’ll realistically get from testing is negligible for you, then you should focus on growth and come back to testing later. You’ll get a much better ROI by focusing on growth than you’ll get from A/B testing. Of course, there are exceptions, and testing with low traffic can be justified, but it’s rare.

What’s enough traffic? It depends of your traffic numbers, your current conversion, and the predicted performance of your variations against your current design. My calculators should allow you to determine the business value of A/B testing in most cases. But for more complex cases, or if you’d like to benefit from an expert assessment, just contact me. I will:

  • review all your web analytics data
  • prepare for your a testing prioritization plan that lets you know exactly which pages or elements you should test first, and how that will impact your revenue numbers.

Testing Consulting & Review

For you if

You’re currently A/B testing some of your pages or some elements of your website, but feel that you’re not getting the full value from your testing efforts. This is the point where the majority of my clients are when I start working with them. Good A/B testing is quite hard to achieve, and most clients, from SAAS start-ups all the way to e-commerce retailers with revenue above $400m are typically not running optimally their testing program.

Until companies will invest in a full-time role or team solely focused on A/B testing, this will likely be the case, since there are so many pitfalls to negotiate. Even as an expert consultant, I learn every week of ways to improve and streamline the testing program I run for my clients. It’s usually impossible to get this level of expertise in-house without focusing full-time on A/B testing.

The challenges of A/B testing

Listing all the pitfalls here would result in 50+ points at least. Most issues we see fall into these categories:

  • what page / element to test: the pages or elements tested are the best ones to minimize testing delays and maximize revenue impact.
  • how to conduct testing: the tests are set up in a way that will result in biased results; useless to take decisions.
  • how to run tests for: be very, very careful with the duration of your tests. I see many tests that show statistical significance, yet are statistically invalid and results are just random.
  • how to plan your tests so you optimize their combined impact: traffic is your scarcest resource for A/B testing, so you should conduct multiple tests simultaneously, provided you do it right.

Validating your testing practice and methodology is critical, because incorrect testing results are much worse than the lack of testing, from a business perspective: you’ll make decisions based on what you think is sound data, and you’ll spend a lot of time understanding what went wrong on the future.

How I support you

I offer 3 different levels of service here, depending on the depth you want me to go to, and the tools you have in place already. In general, I:

  • review all your analytics data
  • review your complete testing technical implementation
  • review your past test results and implementation
  • review your current tests, and your future test plans
  • establish a testing prioritization plan for your specific context
  • suggest additional analytics tools if justified

I provide a thorough review of your testing program, but you can validate most test results yourself using my free A/B testing results validation calculator. This will let you enter your raw testing data that you can get from your testing tool or from your agency, and will give you an overview of what it means and if it’s statistically valid. And yes, many tests run by agencies or CRO consultants do not mean anything, because they are not achieving correct statistical significance, despite what your agency says.

Variation Ideas to test

For you if

You’re confidently running a good Testing Program but are starting to see decreasing returns from it. You feel that another set of eyes to evaluate the variations you are testing or create new ones would provide you with a useful external opinion.

How I support you

Without delving into the technicalities of your testing program, you just show me what page or element you’re looking to test, and I provide you with specific feedback or suggestions on how to create or improve on your variations.

2 pricing options available

I offer 2 different options regarding my fees for this service:

  • time-based: I charge you based on the time spent evaluating your variations, researching alternatives, doing the wireframes, etc. My hourly fees are far from cheap, but you get charged for a fixed amount of hours.
  • performance-based: I do all the work at no cost, and my fees are based on pre-agreed performance targets. We both know what my fees are depending each target reached, and if my variations do not increase your conversion rate, then you’re not charged any fees.

I strongly prefer to charge based on performance, but some of my clients prefer time-based for various different reasons.

Fully managed Testing Program

For you if

You have already implemented an A/B testing tool on your website or application, but feel like you don’t have the internal resources to adequately run it. Then I can manage it for you, based on 2 very different levels of service:

  • simple outsourcing: you tell me what you want to test, I create the tests, run them for you, and analyze the results. Charged based on time spent.
  • fully managed testing program: you task me with running all aspects of your testing program, from strategy to implementation. Charged based on performance, no fixed or upfront costs.

Simple outsourcing

This option lets you concentrate your team’s time on your core expertise, and delegates to me the task of running your tests to an expert. Your team sets the strategy, the tests you want to run, but I do the implementation. This saves you time and is often cheaper for you because I am usually  much more experienced at setting up tests than your team is. It also gives you peace of mind that all results will be valid pieces of data you can act upon. I charge based on time.

Fully managed testing program

With this option, I basically becomes an extension of your team and run all aspects of your testing program. Not only do I implement everything, I also define the strategy, the variations to test, etc. You and your team validates everything of course, but you don’t have to spend time besides punctual validations. This lets you run a good testing program with no additional internal resources tied up by it.

The key to establishing a mutually benefiting alignment is to set our objectives to be the same. This is why I only charge fees based on the actual results you are getting from the test I run for you. There are no upfront or fixed fees. Only pre-agreed fees based on the performance targets I reach. This way, I am incentivized to get you the maximum performance as quickly as possible.

And because a good testing program takes time to do, I am also incentivized to keep you as a client, so I won’t take shortcuts to boost short-term results at the detriment of your long-term business objectives.