Hello and how are you in this economic climate?

Whoa dude! I've been away for a bit since my last post - has anything happened?

It sure has, global economic meltdown, banks being nationalised, the high street looking towards a bleak christmas and analytics vendors convincing themselves that sales will still come flooding in because "now is the time customers need optimisation more than ever'.

I do see that point of view. I also see the point of view of customers that says "We're spending a lot of money, in this climate I have to seen to be cutting costs effectively - do we need that piece of software?". I think the coming months will see more challenges to the ROI of software and consultancy in the web arena.

In a recent poll of data miners there was a mixed response regarding what the current situation might lead to. Across industry as a whole I suspect there will be mergers, and unfortunately job losses in the near future for many companies. In a cold hearted way this might be good for consultants like myself given that companies will become unwilling to offer permanent headcount, in bad ways it could mean I'm out of a job next week as a result of cost cutting in my own organisation!

My good friend Simon Metcalfe says

"Talking to clients they are all saying that there will almost certainly be a drop in Marketing budget but they realise that the best way to get more out of what they have will require analysis and insight.

It's also worth remembering that Direct Marketing as we know it now was born out of the last recession in the early 90's as a cheap way of getting out to customers. Today we now have the web, so I can only see more requirement for our services not less"

Marketing will not get dollars to buy shiny new tools, they will have their budgets slashed but with it they will be asked to come up with more inventive ways of becoming more effective.

I believe the skills of a data mining consultant who's skills are focused on analysing customer data on the web will become ever more useful - thats what I'm telling my bank manager anyway!

Engagement - if you need a definition, here it is

Engagement has been a buzzword in web analytics for probably a few years now. It came along with those other buzzwords like Web 2.0 , RSS, Rich media etc.

Currently people still seem to be banding the term around and trying to find a definition for it so they can impress their bosses. So here it is.

Measuring engagement is the process of segmenting your customer base by the activities they perform and counting how many customers fall into each segment.

There, that wasn't so hard was it?

The key issues so far with engagement has been similar to those that have bedevilled other terms in the online arena, simply that for different websites, different things will be positive and negative. Examples of this would be "stickiness" and "Time spent on site" for which an increase in either can be a positive or negative metrics depending on the goal of the site.

So why has engagement hung around and is it really different as a concept to anything else before it?

I dont think its new in marketing but one thing I would say is that it is something new for the web analytics vendors out there and thats why its creating a buzz.

On the whole the web analytics vendors are still used to dealing with metrics based on site activity - X many visits to this section, Y number of purchases in that category etc. What they are not used to is measuring things from a customer or user perspective and thats what engagement is all about. For example, knowing that you get 12 leads in 100 visits is not the same as knowing that you have two customers who provided 6 leads each.

So lets look at that definition again.

Measuring engagement is the process of segmenting your customer base by the key activities they perform and counting how many customers fall into each segment.

Lets hit the main problem head on. The activities you chose to measure within engagement will depend on your site, your goals and your business - I cannot tell you them, other than to say you will probably have defined what these should be in previous discussions regarding the equally vague "success events".

Next problem is how do you segment your user base? Well again, you can do all sorts of clever stats involving dozens of metrics and fancy algorithms talking about "distance from cluster centres", "neighbourhoods of attraction" etc. here but lets keep it simple.

You could for a given key metric

a) Set some arbitrary thresholds up, such as "any customer who posts more than 3 times is call an engaged user"
b) Look at the spread of activity and split customers into significant groups such as "we have a 100 customers who post anywhere between 0 and 100 times a month, lets sort the customers by their posting activity and split those customers into quartiles (4 equal groups) and then see what the thresholds exist and label the groups Highly engaged, Quite engaged, Slightly engaged and Not engaged at all" - (note if this example were real I would probably have a separate segment for all those customers who never post and then split those who posts between 1-100 times into the quartiles)

If you have multiple metrics, then simply add them together for each individual customer creating a score and then again chose approach a) or b).

Once you have your groups defined and labeled what next? Yes, the counting. Essentially, over time you need to see whether more customers are falling into the higher "engaged" segments. If they are, then great! They are more engaged and you have succeeded! If they haven't then you need to look at what things you can do to encourage customers to move up into a more valuable segment.

The next step after measuring engagement would be to look at "engagement over time" which essentially means tracking customers as they move from one segment to another and trying to see whether you can define their expected engagement over their complete customer lifetime.

The key thing is "engagement" is all about measuring things from the user perspective, not the site. This is something understood in other areas of customer analytics but we are still waiting for some bright spark to realise this and market it under some other clever buzzword (like "personas") and then for the web vendors to catch up.

No customer has ever asked me to validate their business plan

When I go and meet customers and try to gather requirements I always ask questions like "What is the web site for?" and I get a variety of answers but never a clear predefined plan of what is meant to be achieved.

I find this bemusing. I believe that every business venture requires a business plan and yet I see neither hide nor hair of them in my day to day dealings.

If we are talking about a whole site, then obviously business cases would be proposed at a high level "increased sales by X" etc. but I rarely see these plans and if they exist at a 'C' level they rarely filter down to the people who have to implement these things and therefore rarely by default into the analytics requirements.

This isn't just a high level problem. Whenever a new piece of functionality has to be added to a site, someone, somewhere has to make a business case for it but again, so rarely do I see this case when requirements gathering. Admittedly I work for analytics vendor, it could be argued I don't need to know, I just have to explain to the client "how" to measure things with the tool my company provides but I don't think thats the case. More often than not I'm asked my opinion on "what" should be being measured implying there isn't a clear business case stating what hopes to be improved by the additional functionality.


I do think there needs to be more clarity on higher level goals within organisations in general. Every time there is a planned change it should be communicated across the company whether it hopes to improve sales, leads, "engagement" etc. and over what period of time the improvement is being compared to. When doing this we must always take into account the seasonal difference, market differences, increased competition etc.

One of the other facets of this lack of clarity is that a site or a piece of functionality isn't usually designed to increase a single metric. Its usually there to attract and engage with a particular type of customer - maybe it's there to appeal to a younger audience, maybe mainly women, maybe mainly elderly couples looking for that cruise of a life time, whatever, there is usually a concept of any ideal customer.

The point is, if such attributes have been made clear in building the product or functionality, consideration needs to be given to how you measure the targeting of these populations. Is the site really attracting the kinds of people you expect for your product? Are they really performing as you'd expect? It amazes me how often these kind of considerations get lost between the creative product development department and the department in charge of actually making things happen online. I think this is often a reflection of the business goals not being communicated around the company and the creative team not taking responsibility to prove the benefit of their fabulous new ideas.

So what things should people be considering? Well often its a question of "To what segment is this new thing supposed to appeal?" and the challenge is to get a baseline of that "segment" performance now and after go live of the new technology. (I'm ignoring the concept of MVT right now because you dont always have the traffic or capability to test everything in isolation).

In order to get information on that segment think about the following

• Look at the demographics of people purchasing/registering (i.e. use the existing methods of data collection for analysis.
• collect ad-hoc survey information and tie that into user behaviour.
  
• Implement some additional functionality that encourages input of information you know is correlated with your segment. (i.e. teenagers are unlikely to respond to competitions for stairlifts).
• Analyse targeted marketing of your segment and compare to the random sample of marketing
• Buy data from other agencies for enrichment if it can be tied to your own.
  

These are all things that in my opinion should be considered from the beginning of any new venture. Remember, the simple question is

"What are we hoping to achieve and how are going to measure it??

And once your organisation asnwers that for any piece of functionality this should be communicated to all involved in the venture so that even if the specific details are not there all those involved are aware of the goal they are working towards giving much more focus to the
overall project.

How do you encourge your staff to improve your website?

Pay them to do so of course.

Last year I read an interesting book, Freakonomics where the author makes explicit many of the things regarding human motivation you probably implicitly knew in a very clear and entertaining way. This lead me to thing about the way my salary had been set in the past along with the targets of my colleagues.

One of the problems with asking your staff to improve your website is that just as you have trouble defining what is regarded as KPIs for your organisation, you then have to try and tie these things to individuals salary and targets.

An example of one thing I once saw was that the online sales people were given targets to increase sales and conversion as well as continuing to maintain the site with the regular BAU changes.

Increasing sales could be tough. In this situation, sales were separate to marketing and also to product marketing. All they really had within their immediate power was to change the website. One would think this a good thing, for them to increase sales there would need to be enhancements to the website.

However, two things were noticed by these individuals

• whenever a product was competitively priced in the market place, the online sales and conversion would dramatically increase regardless of changes to the website.
• whenever marketing made a big splash, sales would also increase but usually to the detriment of conversion.

The upshot of this was that they concentrated more on lobbying product marketing to produce "competitive products" as it was simply the easiest way for them to hit their targets than almost anything they could do onsite. If product marketing shifted in their favour, they would easily hit their targets, if they didn't, well they could hope that the general trend for increasing sales online would continue or else blame the "competitive landscape" and hope things would be better next year.

The point I'm getting to is that though their targets were directly tied to a key "KPI" for a retail website - its sales (and even its conversion), nothing about that KPI actually encouraged this team to do anything to improve the company website - nothing spelled it out for them.

Had it been my team, I would have liked to acknowledge that product sales can be price sensitive and also sensitive to marketing however this should not have stopped the team from being encouraged to change their site as much as possible during the year. I would have set lower thresholds for sales and conversion i.e. they must not go below X where X is linked to a combination of previous sales in previous period, offline sales of the same period (to provide indication of market performance) with the empahasis that they must have conducted a series of Y trials during the period as well as BAU changes.

The size of the trials is probably up for debate, obviously a bigger trial (on the homepage or registration form) could have a bigger negative impact on sales than a smaller piece of functionality but I would actually wish to encourage calculated risk taking through a formal test process and so would set an individuals size of trial and sales thresholds dependent on what contribution I believe their area plays on sales.

I would then assess the staff additionally from the perspective of the ease/number of tests
they performed, i.e. are they doing it effectively and efficiently combined with the revealing of any particularly useful insights. (things to consider would be number of tests, size of trials, impact of area changed)

So what I would be doing is encouraging a workplace where new ideas go hand in hand with accountability and individuals feel empowered to make a difference rather than having a lack of focus to update and improve their website.

Seth Godin's Meatball Sundae part 1

Seth Godin is clearly the new wunderkind of online marketing. So much so Omniture has invited him to speak at the upcoming Omniture summit.

I figured it would be interesting to take a look at his book, Meatball Sundae, and see what I can glean from it. Unfortunately I'm not one for buying every book that comes from the next big thing, if I did I'd probably be significantly poorer than I am now so what I propose is to look at Seth's excerpts and see whether it looks like a good buy.

Let me tell you what I hoped the book would be about after hearing the title. I assumed that it was an analogy to show how customers when faced with confusing content, messages or indeed food courses might not produce the desired outcome of a pleasurable eating/online experience. (as a side note, it is exactly this kind of mixing of foods combined with excellent execution that sparked the public interest in Heston Blumenthal with delights such as bacon and egg ice cream and snail porridge however I doubt Seth realised this when considering his title)

I'll also be honest and say that by writing about Seth I am essentially trying to ride his coattails a little but I believe this kind of reviewing is exactly what blogs are ideal for.

Lets begin with a quickie.

"The Most Important thing Understand that you don't get to decide what the market demands. The market does."

In my mind, if you take this as his first point, I think the second that Seth needs to add is that you are in control of how you react to market demands - and thats where marketers add value.

Does this mean marketing is always reactive to market demands? Of course not. Thats what market and product research is all about, giving you as close an view to whats really going on in the market place as possible.

However, what it does mean is that we have to put in place

• checks and balances to ensure we capture when our marketing begins to perform poorly. (measurement for management)
  
• processes that will ensure maximum return when the market does indeed swing our way (cross channel optimisation).