In a recent post on my blog titled 4 stages of growth hacking, I make the point that optimising your Growth Engine is often an overlooked step of the Growth process. Without having an optimised growth engine (particularly an optimised acquisition funnel & a low churn rate) you will have a very difficult time scaling your growth either because you cannot find profitable channels or your high churn rate means you’ll hit a growth ceiling.

However too often those responsible for optimising a Growth Engine focus on the best practices out there. They hear see the tactics and the stories listed on places like Growthhackers.com and think these are the things they need to be doing.

While all these stories are great for inspiration and can sometimes work for you, what works better is having an optimised process for uncovering what growth ‘experiments’ are most likely to work in your particular situation and carrying out those experiments in a minimal way.

To help understand this, I have documented and presented the framework I utilise for optimising Growth Engines.

1. Define a single metric to focus on

To begin with, a single metric to focus on must be identified. This is critical in order to ensure your limited marketing resources are highly focused on an area identified to return the highest ROI in the shortest amount of time. Using your analytics system you should be able to identify where the biggest blockage point is and make that the single metric worth focusing on. Often this is in your acquisition funnel (I.e. Signup – Activation Rate) but it can also be reducing your Churn rate or increasing referrals.

It is also important to set an exit criteria at which point in time you move on to the next key metric. This can either be by achieving the goal state or hitting the timebox limit.

• Goal state – A goal state must be set in order for you to map your progress. A goal state should be specific, measureable and attainable such as ‘50% Signup – Activation rate’.
• Time box – An exit criteria of a specific time may also be implemented. This is because testing and experiments generally suffer from diminishing returns, in that the closer you get to a perfect state the less effective your experiments become. Set a timebox limit that allows for enough experiments to be carried out but keeps progress in line with business growth projections. I know at HubSpot Brian Balfour and his team generally focus on 1 key metric in 30 – 90 day periods.

An effective way of setting goal state and timeboxes is to visualize the funnel in an excel spreadsheet and make modifications to the metric in question to see its effects on the rest of the funnel, and then comparing this to growth objectives.

2. Identify the blockage points.

Once a single metric is identified, the next step is to breakdown the steps/stages of that particular micro-funnel (For instance, if the Activation Rate is the identified metric then the ‘micro-steps’ to that would whatever steps you’ve identified as your activation steps).

By looking at this funnel (and deeper into the process if possible), you can identify the key areas where people are falling off and prioritize experiments in those areas.

3. Identify reasons for dropoff

Once the key areas for drop off have been identified, it is important to get feedback from customers as to why this is occurring rather thanjust devise experiments based on unfounded assumptions of what is happening.

There are multiple ways this can be achieved, some of which are outlined below:

• Call – A customer lifecycle analytics package like Kissmetrics will allow you to pull up a list of users who got stuck in an identified state (I.e. Have done event ‘Signed Up’ but have not done event ‘Activated’). Assuming your recording your customers phone numbers at signup you can potentially call the customers to ascertain deep levels of feedback on what went wrong.
• Email – Using the same list, you can potentially email those who got stuck and either provide them a link to a survey or better yet, just give them one or two questions in the email and get them to simply reply to you.
• Qualaroo – Qualaroo is a micro-survey tool that can display a ‘mini-survey’ to a user somewhere on the page. I personally find Qualaroo great for capturing feedback from people who I don’t have contact details, like those who visited the website but didn’t signup.

The questions you ask are going to vary significantly based on the problem you are trying to solve, but some generic questions I have asked before that help give insights at different parts of the acquisition, retention, referral funnel include:

• What are you hoping to achieve by using <product name>? – Ask to people who have just arrived at your site. Should give you insights into what messaging & copy to use to appeal to their pain points & goals and increase conversions.
• Is there anything preventing you from signing up? – Ask to people who have viewed a few pages on your site but not yet signed up. Should give insights into missing information or points of friction in the conversion process which can be addressed to increase the visit site – signup conversion rate.
• What other tools were you considering before choosing our tool? – Ask to people who have just signed up as a way of improving Visit Site – Signup conversion rate. Should give you insights into whether people are comparison shopping, who they are comparing you against and whether you need to take action (create comparison content, refine messaging, etc) to stand above the identified competitors.
• What persuaded you to choose us over other tools? – Ask to people who have just signed up. Should give insights into the key choice drivers behind your product, and can help you know what content, messages, etc to make prominent on our site to improve Visit Site – Signup conversion rate.
• What prevented you from <insert Activation actions here>? – Ask to people who signed up but didn’t take the necessary actions to reach Activation. Should give insights into the friction points preventing people from activating which can be addressed to increase Signup – Activation rate.
• What prompted you to upgrade your account? – Ask to people who have just upgraded to one of your paying accounts. Should give into the triggers that cause people to upgrade which can help you devise initiatives to increase Signup – Billed conversion rates.
• What prompted you to downgrade your account? – Ask to people who have recently cancelled their account or downgraded to your free plan. This should give insight into the triggers that cause people to downgrade which can help you devise initiatives to reduce churn.

4. Develop falsifiable hypotheses and define experiments

Once the dropoff points have been identified and reasons are known, you can then create falsifiable hypotheses and design experiments that when implemented, should hopefully enhance the key metric.

If you’re not familiar with the term Falsifiable Hypotheses, they are hypotheses that are specifically worded so that they can either be ‘proven’ or falsified’. A good example would be ‘Adding customer logos to the signup page will increase trust and improve signup rate to 20%’.

As you can see here, this one sentence defines the experiment you are going to run as well as the expected result. Ash Maurya provides a brilliant template for creating and documenting falsifiable hypotheses and i’d thoroughly recommending checking it out over at his blog.

5. Carry out experiments

This aspect will differ largely depending on the problem you are trying to solve and the experiment you’ve devised, and could be anything from running A/B tests on your website copy through to revising the onboarding process of your application.

The key thing to remember here is to do the minimal thing you possibly can to validate your hypothesis and then scale it from there if it works.

A good example of this is AirBnB’s infamous photography hack.  Their initial hypothesis was that properties with professional photos would get more listings, but instead of going out and hiring a team of photographers to take professional photos of every listing, they recruited a few freelance photographers to snap some shots of a few places and put them on the listings. They then measured how many bookings these properties got against their booking history and similar properties in the area and found that properties with professional photography got 2-3x more bookings.

So when carrying out experiments, remember to not to go all out designing big, complex features and systems. Instead, think about whats the minimal thing you can do to test whether your hypothesis is correct or not. Often this may be unscalable things that involve manual processing, but these can always be automated and scaled later once you know it’s going to be worth investing in.

6. Record results and learnings

Regardless of whether an experiment returns a positive result in the key metric, it is important to document the results and any learnings you may have received from it. These learnings help you develop your customer knowledge, which in turn helps you create more insightful, knowledgeable hypotheseses that are likely to have a positive result in the future.

6. Iterate and repeat

The final step of the process is to use the learnings and results from the previous experiment to iterate on the overall plan, modifying any upcoming experiment or changing priorities based on the new information.