CRO: Why startups should prioritize conversion rate optimization early


I am lucky enough to have worked for companies that ranged from corporate giants like Uber and Coinbase to smaller startups that were run out of private residences in Silicon Valley. One of the largest differentiators between these companies was their respective emphases on conversion rate optimization (CRO).

The initial focus for smaller startups is typically on growth pillars, such as paid acquisition or starting up a lifecycle email program. By contrast, larger companies have dedicated teams in place for managing and implementing their CRO efforts, alongside all their other activities.

Bringing paid acquisition costs down when funds are tight makes a great deal of sense. Similarly, starting email marketing campaigns to improve performance through the funnel can be equally important. However, what many startups do not realize is that CRO can help lower paid acquisition costs and push users through the funnel as much as, if not more than, the other pillars.

As a founder, how should you spend more time on CRO and what strategies best help you establish a CRO function? After I review my experiences of what works best, you will understand how to better prioritize your time.

Think of CRO as a grand supplement to all the other items the growth side rolls out in the early days of a startup.

Examples of CRO

CRO has historically been confined to running tests on landing pages, but there are many other areas to test, including app store pages, email campaigns and retargeting campaigns.

Basically, if you are testing methods to push more users through your funnel and subsequently improve their conversion rates, then you are running CRO experiments.

For our purposes, I’ll go through the specific CRO tests you can start running for your startup’s landing page. Below are some of the largest areas to test:

  • Messaging
  • Images/videos
  • Module additions
  • Module placements

Most startups already test messaging on paid acquisition campaigns, but testing on a landing page is another area to experiment with. When I consulted for a product that appeared on “Shark Tank,” we ran dozens and dozens of weekly CRO tests on the website onboarding questions to find which answers brought in the highest propensity users, a very high rate of testing.

It has been quite surprising at how much I have been able to affect the conversion rate thanks to the repeated testing of different modules, such as testimonials, or FAQs, and where they finally appear on the website. For example, I have found that placing testimonials or press logos above the fold and not requiring the user to scroll down to locate them has always increased the conversion rate.

Setting up a testing framework

Before you begin your CRO journey, it is important to set a foundation to streamline efforts so that you can stack-rank tests and have a consistent measuring methodology.

  1. Create a list of hypotheses.
  2. Implement a stack-ranking methodology.
  3. Establish a primary metric for each experiment.
  4. Set up a baseline method to determine winners.

Once you have a list of hypotheses to test, it’s time to stack-rank tests so that a testing schedule can be set. I highly recommend a methodology such as the one I’ve created, R.L., which takes a score for reach and likelihood.

These two criteria are given a score between 1-5 (with 5 being the highest) on the “reach” of the test, the “likelihood” you think it will succeed and then added together.

Example of stack-ranking using R.L. methodology.

Example of stack-ranking using reach/likelihood methodology. Image Credits: Jonathan Martinez

In the above example, Hypothesis #1 has the highest score and should be tested first, given its combination of reach and the likelihood that the test will succeed. Even if you are already planning to run only a few tests, a simple methodology like this one will help you with test prioritization.

Prior to running any test, you must first define the primary metric that will determine if the test is successful. For example, if you’re running a test on landing page module placements, then your metric will likely be the conversion rate to a website form fill. After launching your test, the easiest way to call winners and losers is by using a calculator to determine statistical significance, such as Neil Patel’s stat-sig calculator. It’s typically recommended to use 80%+ statistical significance to call any test successful, but whatever percentage you use, it must be consistent across all your tests.

Cross-functional collaboration is key

The primary reason that most startups don’t prioritize CRO testing early on is a lack of resources. Additionally, performing that many tests requires cross-functional team collaboration. The types of teams that typically need to get involved are product, engineering and sales.

For all B2C startups, it’s crucial to involve product and engineering teams for robust tests and tracking. At Postmates I regularly worked with the product team to hypothesize, stack-rank and build experiments involving changes to the app and our onboarding flow. If I hadn’t worked closely with all members of these two teams, our timeline for launching tests would have been significantly slower.

For B2B startups, it helps to include the sales team in the mix so that their hypotheses and ideas can be shared with the growth team. As the sales team already experiences a close-up view, thanks to their frequent direct customer interactions, they’ll be capable of helping with ideation on growth CRO tests. In the past, I have witnessed small startups that ignored the sales team in their testing, and this always turned into a serious error.

As a solo founder, cross-functional collaboration will not apply, but it’s something you should consider as you expand the team.

Advancing CRO efforts

Once you have built that solid foundation for CRO testing, then you can begin introducing some exciting new tools, optimizing more down funnel and measuring additional metrics, like incrementality.

A few tools that are extremely useful include UserTesting, VWO and Unbounce. UserTesting allows you to gain insights from real people on your onboarding journey and product. VWO makes A/B testing much easier to launch and track. Finally, Unbounce is a landing page builder that makes testing variations a relative breeze. This is not an exhaustive list by any means, more of a starting point to get your mind thinking on the types of tools you can leverage in your CRO efforts.

Once you push beyond testing upper- and mid-funnel events, you can start to run experiments that involve analyzing revenue and incrementality. These types of tests typically involve more data resources and time to run, since they require user behavior that can take weeks to matriculate. During my time at Uber, we were constantly running tests to move the needle on funnel conversion rates but set the north-star KPI on metrics such as predicted First Trips (pFT) and other deep conversion events.

The biggest challenge to CRO is simply getting started, and my urgent advice to all startups is to do that sooner rather than later. Try to think of your efforts on CRO as a grand supplement to all the other items the growth side rolls out in those early days of a startup.


Leave a Reply

Your email address will not be published. Required fields are marked *