Thinking about ecommerce KPI’s

With website analytics tools it’s possible to measure just about anything, from average time on a page to micro goal completions. But just because you can measure it, doesn’t mean you should. It’s important to think about what goal you are trying to achieve and decide what are the best Key Performance Indicators for the particular business or website that will help you understand how to achieve the goal, and then choose the metrics that inform the KPI’s.

Start with a goal

Goals should be simple, easy to understand, and tied to business objectives. ‘Increase revenue’ might be such a goal, or ‘Increase Conversion Rate’. ‘Increase Average Order Value’ isn’t a good goal as it is too specific and isn’t tied to achieving an outcome for the business. Increasing AOV might be part of achieving a goal of increasing revenue, but it isn’t a goal itself. Goals and KPI’s can be closely linked, for example the goal could be ‘Increase revenue’ whilst ‘Monthly Revenue’ can be the KPI that shows whether the goal is being achieved.

Choose actionable KPI’s

Once you are clear on what goal you want to achieve it’s then possible to choose the Key Performance Indicators that will tell you how well you are doing towards achieving the goal. So, if the goal is ‘Increase Revenue ‘ then you might choose Number of orders, Average order value and Average number of units per order as your KPI’s. KPI’s need to be actionable. There is little point measuring something if you aren’t able to do something to affect it. It’s good to have more than one KPI feeding into each goal.

Pick pure metrics

Once you know what KPI’s you want to use to show how the website is performing, you can then decide what metrics you are going to use to inform the KPI. Metrics are pure numbers, such as number of units sold each month. KPI’s can be pure numbers or a calculation such as Average number of units sold each month. A number can be a KPI and a metric, depending on what you want to measure.

Metrics you might use to inform the KPI’s above would be Number of orders, Number of units, Value of sales. From these metrics you can calculate the KPI, such as Average Order Value, and see how the KPI is affected when you try things like encouraging customers to basket-build by offering incentives such as free delivery over a certain order value.

Create a funnel towards the goal

Creating a funnel for each goal helps to understand how each metrics is performing and how it is affecting other metrics.

It is important to understand the relationship between two numbers and how changing one might or might not affect the other. For example, to understand how many people visit a website and make a purchase we could track number of sessions, number of unique visitors, number of customers, number of repeat customers. This creates a funnel and a direct causal relationship between one number and another. So, if you increase the traffic to the site by 20%, does the number of customers increase by the same percentage? If not, you can then look into the reason and figure out whether the quality of the new traffic is low or if there is something about the website that is reducing the Conversion Rate.

Keep the funnels simple

Funnels need to have a clear, simple and causal relationship between each of the metrics. You could throw Bounce Rate into the mix and think that this is going to help you understand how many visitors to the website leave straight away and so have no chance of becoming customers. Knowing the Bounce Rate will help with this but it has a more complicated relationship with the other numbers in the funnel which it’s important to understand. Bounce Rate is about the visitors who left, so in a way you don’t care about them because once they have left the website (and so are included in the Bounce Rate) there is nothing you can do to affect them and move them from visitor to customer. All of the other numbers in this funnel are about people you are able to interact with, but Bounce Rate is the percentage of people who you can no longer interact with. This makes it a negative metric while all the others are positive, and this makes it a potentially confusing KPI to include in this funnel. It would be ok to measure Bounce Rate as part of a different funnel, perhaps one that has a goal of reducing the number of visitors who leave the website, but it’s important not to confuse the metrics.

Reporting on KPI’s

A single simple dashboard is often the best way to report on KPI’s. Be clear whether the dashboard is for reporting on the current state of the business on a given date or for showing performance against trends. Having both in the same dashboard can create confusion. The dashboard can be created using a simple spreadsheet or with more specialist software.

Achieving business objectives

At first glance, setting KPI’s looks easy enough and often it’s the obvious ones such as Conversion Rate that are chosen without a great deal of understanding about how to set KPI’s that are meaningful or how KPI’s work with metrics and goals to achieve business objectives.

Maybe you shouldn’t hire the best candidate

Interview three people for a job. One is ok, one is good, and the other is clearly a better candidate than the other two; better qualified, more ambitious. The clear winner gets the job. That’s understandable. The recruiter wants the best person for the job, but perhaps without knowing it, they are applying a kind of Nash equilibrium. They are assuming that the candidate, once recruited, and themselves have nothing the gain by changing strategies.

But, the situation can quickly change so that the Nash equilibrium no longer applies. Ambitious people are always looking for better opportunities, more pay, more responsibility. If they were your best candidate, chances are they’ll be somebody else’s best candidate. They’ll be offered another job and they’ll take it, leaving you to start again.

But what might have happened if the recruiter didn’t use Nash and instead hired the second best candidate. They might not have all the skills and experience you want, you might have to put more into training them, but they might not leave as quickly. They could grow with the role and actually end up being the better candidate in the long run.

Working from home

Working from home is one of those interesting organisational challenges. On one hand, in an organisation that is short on space we are encouraged to consider alternative ways of working, but on the other hand, emails are circulated saying that working from home is a privilege that requires approval which will only be granted in specific circumstances.

In one conversation about working from home the point was made that years ago we were told we’d have paperless offices and yet we still print hundreds of sheets of paper a week. If that idea didn’t work then there is no way working from home could work. I can see the parallels. Both shifts require changes in organisational culture and expectations, changes in working practices, and most fundamentally changes in thinking.

I think thinking about it as ‘working in the office’ versus ‘working from home’ is a limiting and unhelpful approach. This kind of switch just swaps one set of issues for another and doesn’t really deal with any of the problems of working at either location. So, really it needs to be thought of as ‘working from anywhere’, be it on the train, in a coffee shop, in an office (not necessarily an office owned by your organisation), or at home. Since ‘working’ is about what the individual achieves for the organisation, ‘working from anywhere’ should be about where that individual achieves for the organisation, but also how.

It is the how that is the big mind shift. The older/existing dominant approach is to ‘move the workers to where the work is’. This is fine in an industrial age where the work is centred around a specific location with specialised equipment, but in a digital age where all a worker needs is a laptop and an internet connection, is it really the best way? A more modern/digital approach is ‘moving the work to where the worker is’.

Moving the work to where the worker is opens options. Of course the worker needs to physically be somewhere but not having a ‘move the workers to the work’ approach means that the worker could choose to work somewhere where they can achieve extra benefits for the organisation. Want to work more collaboratively with other organisations in your sector? Rather than monthly meetings at the office, just all work in the same shares space. Want to focus on a piece of work and not have any distractions? Rather than being at your desk where people can interrupt you, just work in a library.

It should be for the individual to decide how best to achieve for the organisation, including where to be, and the organisation should empower that rather than preventing it.

I took the John Lewis website survey

And this is what I thought about it…

Providing some good information at the start to give me confidence that my data won't be used in ways I might not want. And telling me that the survey will only
Providing some good information at the start to give me confidence that my data won’t be used in ways I might not want. And telling me that the survey will only
Does any website really make someone feel Excited, Hopeful or Delighted?
Does any website really make someone feel Excited, Hopeful or Delighted?
Do they really need to ask this in a survey? Isn't this what analytics are for?
Do they really need to ask this in a survey? Isn’t this what analytics are for?
Again, something analytics would tell them.
Again, something analytics would tell them.
Only 41% of the way through the survey and this page hits you with an overwhelming sixteen multiple choice questions and a free text box. I bet a lot of the participants drop off at this stage.
Only 41% of the way through the survey and this page hits you with an overwhelming sixteen multiple choice questions and a free text box. I bet a lot of the participants drop off at this stage.
For a retailer, I think understanding a customer's gender, age, and martial status for customers can help with creating personas so I can see the benefit of collecting this information.
For a retailer, I think understanding a customer’s gender, age, and martial status for customers can help with creating personas so I can see the benefit of collecting this information.
This seems like a good question to understand the buying behaviour and extended journey a customer takes before visiting the John Lewis website.
This seems like a good question to understand the buying behaviour and extended journey a customer takes before visiting the John Lewis website.
And just when you thought you had finished, one more question.
And just when you thought you had finished, one more question.


Survey complete, but the thank you message seems a bit dry and emotionless. I’ve just told you all kinds of personal details about where I live, how old I am, whether I have kids, how I shop and what I think about John Lewis, and all you can say is ‘Your input is greatly appreciated’. No reward, e.g. a discount code or prioritising my order to delivery more quickly. No telling me how all the information I’ve given you is going to help you offer a better service or improve your website. And no reiterating that my data is safe with you, that the cookie will expire.

Conclusion

It’s a very long and complicated survey with a huge range of questions asking for very different information in lots of different ways. I found it a bit confusing and not really offering any value to the customer.

Why doesn’t water come in cans?

Most major soft drinks are available in cans and bottles. Coca-Cola, available in cans and bottles. Tango, Fanta, Dr. Pepper, etc., etc., all available in cans and bottles. So why is water only available in bottles? Even sparkling water is only available in bottles and not cans. Why?