Cost per influence is mentioned on only 3,430 webpages (according to Google by doing a quick intext:”cost per influence” search) . And some of these date back to 2005 so its not exactly a term setting the heather alight with online conversations and interest. And to be honest I think the whole concept may be wrong.

The great love that I, and lots of businesses, have for online marketing is that its quantifiable. I make a PPC advert up I bid 10 pence of every forty people I get a sale. The sale costs me £4.00 and I make a couple of quid back as profit.  If its not profitable I do some refinement, if it’s not profitable I turn it off or keep going with it accepting that revenue generation is more important and that other high return keywords will subsidise this one. With every channel its the same – we spent ‘x’ and we got ‘y’ as our return.

By starting to bring in concepts such as “cost per influence” everything beomces fluffy. Less empirical. Not based in numbers and we’re unable to define cost and benefit. There is a link between having strong PPC and a strong SEO campaign (PPC can beomce tactical and SEO more strategic by working together). But no agency goes in and says if you spend an extra couple of quid on PPC you’re SEO will benefit due to the growth in brand equity, and this means you can cut back on your SEO investment. And why? because:

• they know they can’t prove it

They know that banner advertising will always cost you a fortune and deliver only marginal results (marginal compared to PPC, SEO, email etc). Accept it. It’s expensive: that’s why most agencies will tell you banner advertising helps with ‘branding’ – my friend colin once described “branding as the excuse you give for not making sales” and I agree whole heartedly. Measuring brand equity online is a fools errand – it can’t be quantified. And if it can’t be quantified should we really be bringing it into online marketing. Just because your brand was mentioned 20 times in one tweeting day doesn’t indiciate that your campaign worked.

Cost per influence looks remarkably like the excuses made for campaigns that were for “branding”. Online marketing is about proving causality between action and result. I’m not saying that online advertising cant have a benefit in developing brand equity etc. Rather than you should not view these kind of fluffy, unquantifiable variables as something of use to your campaign.

Online campaigns generate return,  clients invest to make returns and achieve targets. Multichannel analysis may indiciate cross campaign pollonation and benefit. But an indiciation of something is not evidence, and saying that some one came through one channel and bought on another without a clear factual basisis wrong.

Here endeth the rant for today. Feel better.

1. Overview post on Google website optimizer – good starting place for anyone who’s interested in whats going on. Nice and easy introduction for all those new to onsite conversion etc.
2. Marketing Experiments set of links about Google Website Optimizer – I love this site. Its all testing and results driven – for evey online marketing channel you can think of these guys have probably done some cracking little test to give you some ideas on getting more out of it.  This is their post is about helping you find the best way to improve your tests – as its only with good tests you’re going to get any decent information.
3. Google Website Optimizer Cartoon – I was just going to use the image but I always liked the guys at GrokDotCom (one of the first places I ever heard about testing to convert more visitors if I am absolutely honest)

So here I am with a wee flash movie telling me about the PDFs I’d blogged about a wee while ago. But the flash movie aint a link! Yep when you click it nothing happens. Its telling me specifically about something which isn’t contained within the content of the page and its not a link.

Its not a unique thing to the BMM site all over the Internet people are making websites with nice images and flash movies that dont have links on them. I did some work last year for a hotel website where I found that over 40% of the inactive clicks, clicks occuring by people clicking on non-linking areas of the webpage, were on images. So the picture of the wedding was being assumed by people to be a link to the wedding section. By updating the CMS to get these images to be links we increased conversion by more than 10%. By not having these links functioning properly they were losing sales!

To find out how people are really using your website – and where they are clicking and expecting responses from the website have a look at installing Crazy Egg or Click Density. I prefer Click Density but that because I like to do sitewide tests rather than testing just one page at a time – giving better overall understanding of how the website is being used.

Please note that tracking of flash objects is a bit dodgy at times – with clicks not being effectively tracked. so remember to cross reference all your click data with your analytics click path data to get a better picture of how visitors are using/trying to use your website.

It’s something we’ve been developing for about three months. It’s kinda low level but it takes information about a user and then uses this to intuit their intentions when they reach a website. We thought it was great little thing and the more we’ve got into it the more excited we’ve become.

The problem with your website is…

Your website talks to each consumer as if they wanted the same things.

Every user gets the same experience – irrelevant of intention. It’s not great for the consumer and it’s not great for your business. You can go to Google, or a click on a link on another website, visit somewhere from an email and get the same experience as every other user. Inspite of the fact that I may not have the same intent as every other user.

In essence all that marketing budget you spend online is to ‘engage’ prospective visitors and get them to click to your website. And once they get to your website they will be given content/information/pretty pitcures and with this information they decide to buy your services or your products. But often a website can’t cope with the myriad of expectations which your website has. It just can’t.

If you think of it channel by channel you’ll see what I mean:

SEO: you optimise your website titles and content, get great navigation, get great images and you test and refine till you get great positions for your brand terms and some core generics. For a simple example have a think about a hotel chain. If you type ‘ramada jarvis hotels‘ into Google.co.uk you get the Ramada Jarvis homepage – makes sense that you’d find it in top position.

If you type in ‘weekend break hotels’ you’ll find them at about position 4 in Google.co.uk, you’ll get the same webpage and the same experience as someone who entered the brand term. Inspite of the intention of you, the user, being completely different you get the same experience as someone who is brand aware and knows what the site does. With the ‘weekend breaks’ query resulting in traffic the psychological link between the query and the resulting website are incorrect – in effect the website has led the consumer to ‘lose the scent’.

This is often unavoidable due to the way that search engines rank pages: Ramada should be relevant for ‘weekend breaks’ their hotels are affordable and useful for trips away – but the search engine result isn’t 100% appropriate, and developing a new page for every term under the sun would make their SEO campaign hugely more expensive (and their SEO campaign is good – this was the third hotel chain I tested and the only one to have a generic term on the homepage that ranked well). If only the search engines were able to intuit more and give better results for the website owners they’d make more money.

PPC: Every man and his dog is (OK should be) using landing pages and deeplinking in their PPC campaigns. Specifically targeting pages with content relevant to the query drives better conversion. But with broad match on, even using a ton of negative keywords, you’ll find some irrelevant or unexpected traffic coming through. Try something like “Holidays in Scotland” – the result set you get may be different from mine. But if you click on the link to Hoseasons you get this page. (Yep we’ve just skewed up their PPC stats for the month as their onsite trackings just been given a dunt judging by the URL). But I can’t see anything on this page which is relevant for this query – they’ve probably bid so high on the term ‘holidays’ that they’re popping in all over the place. Their PPC agency should be shot but it highlights the point that the bids are going to waste.

Again the traffic you’re generating – what’s more you’re paying for – isn’t quite going to the page that is most appropriate for them – reducing your sales, reducing your competitiveness and making your online return on investment lower.

Affiliates: Affiliates help you make more money from your website. You pay them to get you clicks and should one of those clicks turn into a sale you pay them commission. But affiliates can be lazy – affiliate networks can be lazy – and your affiliate agency can be lazy. Take the Butlins affiliate program it has nice banners that say stuff like ‘Altogether more fun” but takes you to a page advertising “The big deal” – the disconnect is huge. You as a consumer had experience A in mind and what you get is something altogether different.

This is a bad for you – you’re not making the sales you could, for your affiliates – who will leave your program if its not making them money and ultimately , and most importantly, for the consumer.

The solution for your website is…

Make your website respond to what your customer has asked from it.

Let the behaviour of the customer before they reach your website and during their visit allow you to enhance their experience. I’ve heard people talk about website that do this as ‘Web 2.0′ websites, I call them ‘Direct Response’ websites and some call them ‘behaviourally adaptive’ websites. It’s not important what you call them it’s what they do that’ll help change your online thinking forever.

We’ve developed a very small system that does this – it uses three things it learns from users and enters beta testing in two weeks, we’re really pleased with it. The alpha tests were going well and then yesterday I attended a meeting in London. Was great – I got up at four, flew from Glasgow at 6.15 (which I only just caught), endured trains across london, got to conference venue at about 9.30 (I’m late), register (find that the early morning sessions are in one big hall and there are no seats left, the coffee doesn’t get served till 11.

I’m not a happy bunny at this point.

And then I go to one of these early morning seminars and from the first moment I felt like everything we’d been messing with to get cash out of our sites had been trumped. I was expectiong some stuff about how to target and test your website to make more cash. You know the usual A/B testing, multivariate testing that we all promise we’ll do and then I saw how strong a product from Omniture called “Test& Target 1:1” was.

It does your A/B and multivariate testing on the fly – to work out which configuration of your high traffic pages is best for users. When I say high traffic I mean really high traffic pages – you know the blue chip, high street, brand name webistes that we have probably all used at least once. It takes about 40 things about a user and seeks to derive their intention.

I’ll show you one cool little example of this – go to Google and do a search for “car insurance” – click on the directline.com result in natural search and the homepage will have the leader image below.

Direct Line Homepage image called in by Omniture Test& Target for the natural search traffic from ‘car insurance’:

Go back to Google and search for “home insurance” and again click on the directline result. You get the same page URL with different image presented.

Direct Line Homepage image called in by Omniture Test& Target for the natural search traffic from ‘home insurance’:

It’s cool – a huge UK brand is changing the content based upon your query. Not only that the system has probably tested three or four versions of that graphic to work out which one leads visitors to complete the goals most often.

What other ‘behaviours‘ can Omniture use to target websites?

If you think that this message can be changed by looking at your IP (to determine your location), your browser (as your browser can be a social calling card), you connection speed (Flash movie or animated gif depending on how fast teh files likely to download), which pages in the website you’ve already visited in the session, which pages you visited in previous sessions and its after midnight so I’m not going to put them all in. You can see, even from this short list, why I went WOW!

It was a bit dissapointing that during the Q&A that their seemed to be a lot of things that teh account handlers were ghetting batted back – as they hadn’t fully informed clients of what this ‘Test& Track 1:1′ could do for them. This thing should be flying off the shelves. Maybe it is maybe it isn’t how do you tell when the level of intimacy with the website is so subtle that you can’t see it unless you know that it is there.

Getting Test& Target isn’t going to be cheap (they never said as much). But when you think that you may need a set of 60 creatives for your homepage alone – rather than just one, the system itself will no doubt have a highish cost and you’ll have to mess about with your code a bit but if you’ve got enough traffic and your conversion rate is poor (they talked about conversion rates of below 4% at one stage being poor!!!!) it could be a viable way to go.

Until we have such a website @ mercurythread we’re gonna keep pluggin away with our little system and get better ROI for our clients by spending budgets effectively and efficently. And then when we go and sign up Sony/BMG or Hilton I know right where I’m going to be looking to get them to invest. I just hope by this stage Omniture have an affiliate program…

PS: Sorry for the epistle about behaviourally targetting websites but I think it’s damn cool. If you’ve got any comments – or know of any other systems that do this – drop us a line in the comments below.