Just tell me, what does good measurement look like?
So what does good measurement look like, how do we define it and how do we know we are putting our effort into the right things?
We have been wrestling with the issue of good to determine what it is and certainly what it isn’t. In the image above what good looks like varies from one individual to another. Beauty is certainly in the eye of the beholder.
What it isn’t
Good does not always mean following what others are doing.
Sure it is good to get an external perspective of your competitors or the industry sector you are in, but basing your goals objectives or KPIs on someone else’s business is nuts. There are so many factors at play that puts your business at odds with theirs, the number of products you have, the user journey, even your audience segments or customers and where you spend your marketing budget.
Furthermore using “industry standard” metrics as the only way to measure or benchmark your business is also insane. Surely the nature of the industry means there can be no ‘one size fits all’ standard. We are not looking to standardise light bulb fittings here.
We have a diverse set of inputs and outputs:
- The number of visitors to the site
- The price of the products
- The uniqueness or the products on offer
- Cost of delivery
- Seasonality
- Spend on media
- Maturity of ongoing communications to existing customers
- The content on the site
- The design and look and feel of the site
The list goes on.
What good measurement looks like
Your own data gives you the best indicators.
You will have campaigns and content which are performing well and you will have others that are under achieving.
We for instance like to look at behaviour (visits). Let’s look at this from a combination of a) device and b) source of traffic.
It is the behaviour (visits) within this context that is important so that we compare like with like and don’t assume that we can get mobile ‘searchers’ to act in the same way as desktop email responders.
Bringing good to life
Avinash Kaushik describes successful measurement in 5 simple steps. In the following example we aim to highlight what good is and is not and how to measure it.
Look at the example below:
| Device | Sessions | Bounce Rate | Pages per Session | Session Duration | Add To Cart Rate | Conversion Rate | Average Order Value |
|---|---|---|---|---|---|---|---|
| Mobile | 1,962,046 | 40.93% | 2.69 | 00:01:24 | 4.63% | 0.74% | £ 28.91 |
| Desktop | 1,637,816 | 28.80% | 3.69 | 00:02:25 | 7.83% | 2.52% | £ 43.76 |
You can see how there is a big difference between activity not only in the number of sessions (page views) and session duration (time), but also the really important metrics such as conversion rate and average order value.
Secondly looking at channel (sources of traffic) we can also see the difference across these:
| Channel | Sessions | Bounce Rate | Pages per Session | Session Duration | Add To Cart Rate | Conversion Rate | Average Order Value |
|---|---|---|---|---|---|---|---|
| Direct | 671,944 | 62.29% | 2.38 | 00:01:14 | 3.13% | 0.79% | £ 56.60 |
| 2,388 | 17.59% | 4.43 | 00:03:22 | 19.93% | 10.30% | £ 36.99 | |
| Organic Search | 1,605,456 | 28.11% | 3.41 | 00:02:10 | 5.71% | 1.41% | £ 36.15 |
| Paid Search | 973,912 | 25.66% | 3.25 | 00:01:49 | 8.36% | 2.18% | £ 41.00 |
| Referral | 250,856 | 46.02% | 3.05 | 00:01:57 | 6.05% | 2.23% | £ 37.45 |
| Social | 95,304 | 40.95% | 3.17 | 00:01:43 | 9.89% | 0.82% | £ 22.69 |
Now when you combine the two tables, even though there is a longer list we can easily see there is difference even within a particular source.
| Device | Channel | Sessions | Bounce Rate | Pages per Session | Session Duration | Add To Cart Rate | Conversion Rate | Average Order Value |
|---|---|---|---|---|---|---|---|---|
| Mobile | Direct | 512,066 | 66.86% | 1.98 | 00:00:54 | 2.09% | 0.29% | £ 30.36 |
| 750 | 28.00% | 3.12 | 00:02:12 | 12.27% | 3.73% | £ 32.79 | ||
| Organic Search | 800,826 | 31.48% | 3.04 | 00:01:45 | 4.50% | 0.78% | £ 29.08 | |
| Paid Search | 505,144 | 27.55% | 2.79 | 00:01:22 | 6.57% | 1.10% | £ 29.11 | |
| Referral | 79,290 | 51.34% | 2.76 | 00:01:21 | 6.19% | 0.97% | £ 27.01 | |
| Social | 63,970 | 44.58% | 3.06 | 00:01:29 | 9.36% | 0.72% | £ 22.22 | |
| Desktop | Direct | 159,878 | 47.64% | 3.66 | 00:02:17 | 6.45% | 2.37% | £ 67.01 |
| 1,638 | 12.82% | 5.03 | 00:03:54 | 23.44% | 13.43% | £ 37.19 | ||
| Organic Search | 804,630 | 24.76% | 3.78 | 00:02:34 | 6.91% | 2.04% | £ 38.85 | |
| Paid Search | 468,768 | 23.62% | 3.75 | 00:02:18 | 10.30% | 3.34% | £ 45.21 | |
| Referral | 171,568 | 43.56% | 3.18 | 00:02:14 | 5.98% | 2.81% | £ 39.11 | |
| Social | 31,336 | 33.54% | 3.41 | 00:02:13 | 10.98% | 1.02% | £ 23.36 |
Once we have these contextual groups we are then able to understand and profile our audience behaviour appropriately and get closer to knowing what good looks like.
The next step for understanding behaviour is to then look at the different tasks which visitors perform on the site.
In our post What the Fu…..nnel? we identified the 3 key behaviours of visitors, conversion and value. We take the benchmark values for each of the these behaviours based on their signals.
From here we look for under-performing subgroups within each source & device combination and work out the revenue uplift the business might be able to achieve if they were able to get under-performers to their source & device average for each of the 3 behaviours mentioned above.
The Crank View
When you want to understand what good looks like, don’t do stupid things by using other people’s data to benchmark your own business.
Use it only as a guide to understand what the best in class is doing out there, but take into consideration their price, product offering and marketing activity.
Our three recommended steps to understand what good looks like are:
- Use your data to understand what is working for you and what could be improved on your site
- Use a combination of source and device to ensure you benchmark behaviour in context
- Identify the signals for behaviour for visitors, conversion and value and aim to get the under performers up to their contextual average.
Rinse and repeat to keep addressing under performers, which will continually lift your overall results up to your own ‘good’.
What’s your view on defining success?
