Crafting Your Web Analytics and Attribution Strategy – Do’s and Don’ts [Podcast]
Boiling the Ocean for understanding Web Analytics and Attribution
These are best of the times and these are the worst of the times. We live in a world better connected than ever before. The average number of devices/ user is expected to hit 4.3 by 2020. While this open up new touch points for user communication, this also is the beginning of marketer’s biggest nightmare – web analytics and attribution.
John Wanamake’s quote “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.” doesn’t seem to get old enough.
As cliche as it might sound, all optimization tactics eventually drill down to attribution and web analytics – understanding which channel works best. Users follow a specific process when it comes to buying.
- Identifying what needs to be bought
- Gathering information about it
- Finding out of alternatives
- The Actual Purchase
- Post Purchase Behaviour.
Everyone inherently follows the same process. What differs in the completion time and the multiple other touch points.. This is what the Google throws up when you search for – “Men vs women shopping behaviour”
When the users behave in a linear process, attributing their decision making to a channel or a medium is a cakewalk.
With access to smartphones, this process is further skewed and complicated where in user in the middle of a cart check out process, bounces off to check out seller reviews, after sending a tweet and seeking social feedback, amidst which coupon sites start targeting user and user selects the coupon codes. At this point, the end user has walked into a physical store to buy groceries and eventually decides to complete the check out process later in the day using table. All this for buying one product.
While attribution sounds simple, user journeys such as these are an attribution nightmare. This becomes even more daunting with the whole Last-Click-Attribution and Cross Device challenges. Website, Mobile Web App, Native Mobile App, Physical Store, Registered Sellers on Marketplaces.Attribution becomes more difficult when the user identity varies across channels and mediums. Right from an anonymous cookie on Desktop, Device Id on Mobile Web, Physical Location with the offline store and email id with the coupon site. Such a problem paves way for an extremely strong attribution platform which can associate or capture the true one view of a customer.
Attribution as a problem that every marketers in the e-commerce faces sooner than later. Especially on the performance marketing side. Everybody wants a solution. There have been attempts by Google’s and Facebook’s of the world who have tried to solve this problem.
Is this problem solved yet ?
No. And won’t get solved by a far shot in near future.
How best should marketers approach this problem.
The biggest challenge associated with marketers is their ”Search for Perfection”. Marketers end up boiling the ocean. Here is how boiling the ocean process looks like –
- Understanding existing channels
- Researching available solutions
- Understanding the scope of your requirement
- Evaluating solutions for your requirement
- Multiple rounds of discussions with potential vendors
- Closure – Contract and commercial
- Actual Integration
- Optimization and impact measurement
In a large organisation, the above-mentioned process can take anywhere between 6-12 months.
The end result often is frustration. We often don’t need to have something perfect to start with. Stalling progress and movement only makes you lose value.
To give an example, Google analytics is free. There are sophisticated solutions such as Omniture out there which provide sophisticated tracking but something like a google analytics is a good starting point to start figuring out attribution and web analytics. To start with, you can URL pass through as long as your system now looks at saying from x campaign id and then you can just start looking at that user name or the user id associated with that campaign id.
It is not rocket science to sort of start figuring it out, you could essentially figure out that how many conversions came via which campaign / channel. You can definitely start creating cohorts for each channel and start tracking their journey and seeing whether or not do they participate or engage more. Google analytics, recently released a very interesting tool called User Explorer, allowing you to potentially start looking at just one user id and see how they behave across the system. Is it perfect? No, it is not but does it’s about something better than nothing so as naive that as it may sound from an attribution standpoint, I anything better than you have that 0 is still a better one to go by .
Graduating to Actionable Analytics from Pure Analytics
Again, Google analytics does not give you data of your user but it gives the data of a user. That’s something which a lot of people struggle with, people often look for actionable insights for the specific user. This creates a lot of fiction in the marketing campaign system wherein how do you reach out to that specific user and engage him. So what are the tools available out there to solve this problem statement?
Marketers do potentially want to track the particular user and then figure out what message or what campaign worked for that user. However, my personal opinion is that while personalization at scale is something we talk about, it is still not sometimes worthy effort to spend so much energy saying what is this one customer or what is this one user doing and how I can serve him/her better? I think it is something we should still be figuring out at scale. For eg. instead of figuring out what to do with this user, I could still start giving it on a channel or category level.
It is not the perfect answer but I think trying to do it individually for each user is both the technical challenge especially around privacy wherein all the ad platforms and all of your google analytics will analyze data and they will never allow you remarket or retarget that individual user. Here is a growth hacking way of how it is possible to personalise at a channel / category level and generate fair amount of value.
For instance, you have 1000 users coming through paid media channels. Can you treat these users in a specific manner. Can you give them a different messaging? For eg users coming from a display channel or a social channel should be treated differently. Similarly there are multiple ways of responding or customizing your user interactions.
There are ways like especially for B2B products, tools like Hubspot, Mixpael which allow you automate your marketing activity for each level of user activity.
For instance, when a user responds to your campaign and lead is generated then Hubspot Workflows capture the user data and actionand a rule based engine kicks in allowing you to personalise everything basis user actions ( events ).
There are multiple methods of ways of doing so. Even in case of re-targeting, marketers can essentially map out a funnel and identify the key criterias basis which retargeting for a user should be done. For instance users spending less than 5 mins and with Cart Value of less 1000, should not be retargeted for. There is no perfect tool which is being used at scale where marketers could just optimize for individual user. Instead, optimization should be focused on behaviors. So if i you 100 users coming to your site, there is a good likelihood that 10 of them would behave in a similar fashion. So you can optimize to cater to those 10 users together instead of those 10 individuals separately.
Here are some popular tools that are used by marketers at different stages of business scale :
|Tool||At What Scale
|For Whom||At What Cost ?||Remarks|
|Google Analytics Free||Anything up to 10Mn Users||Both B2B and B2C||Zilch.||You should definitely get to learn on how to use GA effectively. Across campaigns|
|Google Analytics Premium||Above 10Mn ++||B2C||Starts at 1Mn INR per Month ( India )||Allows end user level identification and mapping. Gives one view of user. Best suited when a brand has multiple touch points for end user.|
|Omniture||Above 10Mn+||B2C||Starts at 1Mn++ INR per Month ( India)||Comes with a full blown suite of Adobe’s Tech stack covering analytics, DSP, DMP etc. Deep Integration. Most sophisticated product out there.|
|Upto Mn Users||Typically B2B but works well for B2C upto a specific scale||Starts with 150 USD and goes upto 2000 USD per Month||Mixpanel and Kissmetrics offer actionable analytics allowing marketers to take the next step of “So What”|
Salesforce Marketing Automation Cloud,
|Less than 100K Users||Classic B2B||Enterprise version starts at 2400 USD per Month||Full blown marketing automation solution that makes every user action trackable and actionable. Extremely powerful|
This is the part where marketers need to look at their basic metrics and define their bottom line. Marketers can capture relatively smaller and simple opportunities and get incremental gains, instead of aiming at full blown personalisation.
Listen to the entire podcast here