SES London Keynote – Avinash Kaushik – #seslondon

SES London Keynote – Avinash Kaushik – #seslondon

21st February 2012

It’s almost two years to the day that State of Search was launched and I heard Avinash speak at SES. He was truly inspirational and some of the ideas he presented that day I still use on a weekly basis today. Needless to say I am a huge fan and was really looking forward to the keynote.

He started with a favourite quote from a Kenyan Farmer, “information is powerful but it is how we use it that will define us.” We spend a lot of time making love to data but very little happens as a result. The key is trying to find out how to move beyond that information to drive change in your company.

Re-imagine greatness
We have access to more information than God ever wanted us to have yet most people use this in a very poor way. Most companies measure clicks, impressions and visits which is all high quality garbage like page views. Don’t measure trash, HITS is an acronym for How idiots track success, that’s why we suck if we use it.

Think about the customer and match their intent. The quality of the landing page seldom matches the user intent of the searcher and that is massively important if you don’t want to suck. Match the mobile user experience too, Sainsbury’s doesn’t manage the mobile experience at all while Amazon kicks ass. No wonder why Amazon has its dominant market position.

Bounce rate shows you how much you suck. I came, I saw, I vomited. He then found 3 major mobile insights providers in the UK and not one of them had a mobile version of their website. The mail online was another failure due to the length of the page and had 784 links on the page, simply not usable for a mobile device.

If your strategy was to leave page views and push visitor loyalty your strategy you would stop sucking. It emphasises people more and technology less. Looking at interest over time brand trend graphs will give you a share of search for your customers. If you are going to do a survey don’t make it complex and don’t start it straight away. The Waitrose survey is tortuous popping up straight away and has several screens. The most critical pieces of information for a survey, what was your goal in visiting and did you achieve that, if not why not?

Super awesome reporting contains loyalty, regency, conversion rate and not page views.

Conversion rates
The top 50 ecommerce websites convert only 2% of visitors to customers. You end up analysing the hell out of the 2% but ignore the 98% who aren’t customers. Your website should optimise for micro and macro conversions. Measure the sales and conversion rate but also engagement and the other metrics. Figure out the economic value of those minor goals.

For John Lewis you may not buy the expensive scrubber on your first visit but you may sign up to the catalogue which results in a multiple revenue sale in the future, this should be measured. The same happens with gift vouchers as they will be part of the loyalty scheme. Instead of just scrubbing yourself you can be a lifetime customer that eventually chooses John Lewis as their wedding gift registry. Add a value of each goal that contributes to the complete impact your effort has on the business. True company impact is a sum of all the goals, not just the revenue goals.

You can then segment the data across all your traffic channels and define how each traffic channel performs. This gets you far closer to the complete business strategy to drive success.

Explore and Evolve
Traditional marketing is shouting at your audience and hoping that someday they will listen. Digital marketing is a two way conversation, use what insights you have from the potential customer’s intent and match their expectation. You no longer shout at your customer, you need to engage and have a conversation. This is a deep structural change. You can’t shout at people using Facebook, it’s not going to have any engagement so you need to change your approach and become more conversational. Disney sucks because they shout all day which results in 1,064 followers and no engagement. They all suck because they went to a conference and applied their traditional marketing techniques to social. A good example of it working is National Express, their Facebook account engages. Another good one is M&S, their social strategy is both reactive and proactive. They are not shouting at people, they are engaging.

Four things you should measure for social contribution:
1) Conversation rate – if you talk does anyone care? The number of audience comments per social contribution will help measure this.
2) Amplification rate – 70, 000 people follow Avinash, his second level of people is over 1 million people. # forwards per social contribution is a key metric to cover too.
3) Applause Rate – # positive clicks per social contribution
4) Economic Value – sum of your macro and micro conversions and how they contribute to your overall picture

He shared a positive experience for a credit rating company he found through social search. He not only got the best value but he then tweeted it thus amplifying the social coverage for that brand. If you don’t have a successful social graph you don’t have any way of performing within social search. Do your pages have the +1, Facebook Like and other sharing buttons? Social media isn’t about the major platforms, it’s about engaging with your customer to embrace the social web.

Intelligent Attribution Models
There are hundreds of cross platform brand touch points and how do you maximise the digital marketing spend.

Understand your current performance. Look at your cross channel conversions within Google Analytics which will show you how each of your marketing channels overlap. Work on your custom dimension paths and work out how clients interact with different channels to convert. Assisted conversions are key to measure. Move away from last click wins. First click also sucks because it’s like giving your first girlfriend the credit for marrying your wife. The Equal credit method is like saying everyone gets a participation certificate for participating in the Olympics, someone needs to win! Time decay works the best but you need to remove all sources that contribute beyond the average days to conversion value so that doesn’t skew the data. No model works for every business, you need to understand what model works best for your business.

Strategy for proper attribution: 1) Understand, 2) Test, 3) Be less-wrong.
Multi-channel advertising
Should we do 900 things? You can’t do everything all the time, you need to test for incrementality. You can run these tests by geo segmenting similar markets and turning off different marketing channels for each area. This will give you an idea how the channels work together and which contributes most to the profit. The profit metric is key in these experiments so you will need to be able to get those numbers out and work with your internal number ninjas to help measure that.

You need to use 10% of your budget to experiment with your multiple channel portfolio as you will be able to optimise your ROI without increasing your marketing spend.

Price, Cost, Share, Size
Ultimate measurement consists of measuring your Net Income. Your marketing should be impacting your market share and market size too. He shared a methodology which we will link to later. Work on the strategy before the data. It needs to be tied into net income for the business. If you tie into the net income you tie into the thing that matters the most.

All in all it was another amazing session from a truly inspirational speaker. If we were all as passionate about measuring what matters we would all deliver far more ROI for our clients.

This post in part made possible by a sponsoring from Majestic SEO who have the largest Link Intelligence database in Search.


Written By
Louis Venter is the founding director and CEO of MediaVision, a Search Engine Marketing (SEM) company specialising in all areas of search. His particular interests are organic search marketing, paid search marketing, conversion strategy and online PR.
  • This field is for validation purposes and should be left unchanged.