Thursday, March 13, 2014

Why are you wasting your money on an SEO agency?

It's been a while since I've posted (sorry about that) - I'm halfway through the middle of trying to rework out the purpose of this blog, who the audience should be and what it should look like. But in the meantime I thought I'd share some things that I've been looking at in the SEO world.

Recently I got an email from a nameless company about a specific page on my site that pointed towards their site:


It's possible that you've had an email like this recently too, if you are a publisher. Of course you might also have sent an email like this if you are an SEO person as well.

What is wrong with this email? This company is asking me, on my blog where I write about stuff to do with web analytics, to remove a link to a page to a jobs board that contains jobs specifically in web analytics. If Google thinks that this link is causing problems for this company then something has gone seriously wrong at Google. Of course they don't, they think it is a good link and probably upweight the page because of it.

Moreover this is also a web site about SEO, so you would have thought that I would know a fair bit about SEO and would actually think about this. Clearly the person who wrote this email either:

  1. Thinks that they are right and the link is bad for them
  2. They didn't bother to read my site, but just sent a boiler plate message 
  3. They just got told to do it by their search agency who are probably getting paid a fortune*
I'm not inclined believe number one on this, I am more likely to believe number two, but I'm most likely to believe number three.

(Of course this is likely to be the same agency that was doing link building for them before and caused them to have a number of bad links and resulted in them tumbling down the rankings and losing market share.)

And I'm not the only one receiving emails like this. Here is an article from Aaron Wall at SEOBook who has also been having similar emails. Notably his one was from a domain name company who had lots of thin pages indexed:

The big problem for Buy Domains is not backlinks. They may have had an issue with some of the backlinks from PPC park pages in the past, but now those run through a redirect and are nofollowed.

Their big issue is that they have less than great engagement metrics (as do most marketplace sites other than eBay & Amazon which are not tied to physical stores). That typically won't work if the entity has limited brand awareness coupled with having nearly 5 million pages in Google's index.
Of course the site that linked to me has more pages indexed than Buy Domains and many of them are also thin content pages with little engagement for the user. Incidentally, the page that I link to is indexed and is the top page on the site if you search for '[company name] web analytics'. You'd have thought they'd be pleased with the favour.**

I disagree with Aaron that this is something that is Google changing the rules and killing small businesses. A good SEO person will always come up with long term strategies to get a site to have good, long term content, that people will want to link to from relevant places and find ways to promote it to those people.

The problem is that in the past bad SEOers went around begging, buying and swapping links to poor quality content from places that aren't necessarily relevant in an attempt to satisfy clients who had poor metrics for a job well done. An SEO person should not be getting paid by the link, they should be getting paid for getting better conversions from quality visits from related search terms.

Bad Links

Whatever the past, the way I see it is that it is possible that you now have some bad links that have resulted in you dropping down the rankings. Emailing people asking for them to be removed is a waste of time. You're almost certainly going to use the disavow tool as well, so just do that.

But you should use the saved time to work out whether the links are bad or not. If you just disavow everything then you are throwing away good links along with all your bad ones. You've just lost yourself the good value as well as the bad ones.

Work out what is wrong

In 90% of cases I think that the problem isn't going to be bad links. You will need to think about user journey and how it affects the users. You want to look at providing value for your users and churning out boiler plate pages with thin content isn't the way. This incidentally was something that was also picked up by Google as being bad.

Spend time creating unique, interesting content. If you want to get visits to your web analytics jobs page, then you would be better off spending time creating some unique content on web analytics jobs than you are asking people to remove links to it. Of course this is something that you have to invest in because you'll need to update it over time. Maybe you could email a resource who has linked to your content and ask them if they would be able to help.

Google has always been aimed at providing good quality results to users. SEOers have always aimed to making their site be the top of the rankings. Find the ones who will make your content better for the site, rather than artificially promoting it through short term tactics (or of course both).

The reason that SEOers are now getting closer to Analytics is because the good ones are realising the above. If your SEO agency doesn't talk to you about what your conversions are, how you are obtaining them, what sort of reporting you have available, etc then you are talking to the wrong SEO agency and you should be getting rid of them.

If you SEO agency is talking to you about how they will get links, how they can identify bad links, how they can get you to number one in the rankings or how they can beat Google, then you are talking to the wrong agency.

This applies to small companies as much as it does large companies. There are good people out there who will spend time on your site if you are small and give you great advice on how to get more search traffic that will convert on your site. There isn't, I'm afraid, a silver bullet any more. The only way to do well at search is by having good quality content and through hard work maintaining and promoting it.

* Yes I do realise that I work for a consultancy whose job it is to tell clients what to do and get paid a fortune for the privilege

** No I haven't removed the link and I will be responding to the person in question

Sunday, November 24, 2013

The difference between Optimisation and Strategy

You'll have noticed that there has been a lack of posts on here recently and that is been because I've been quite busy in the office. What have I been busy with? Well one thing that has been going on is I've been working on some more 'strategic' projects. So I thought this would be a good time for me to talk about the difference between Optimisation and Strategy.

Optimisation

As a Head of Optimisation, I suppose it should make sense to start with what optimisation is and why it is important (although if you are a regular reader of this blog, I'm probably preaching to the choir).

What we do with optimisation is very complex process that we try and simplify as much as possible to make it more palatable. We take an existing thing, work out what it is meant to be doing, what a measure of success for that is and then use as much data as we can get our hands on to make it better.



We make a measurement framework like the one above. We derive a series of 'Business Objectives', work out the ways that the business is making these work on the site (your business unit's 'critical success factors') and then we work out the thing that shows that this is working or not - the performance indicators.

These indicators can be validated (in your more direct consumer world this is easier to do through CRMs or through market research where you don't have those direct figures) or they can be unvalidated if you either think you know enough or if you don't have the money to double check.

That's all very boring, but people like me will turn up at your bosses office and produce a load of graphs showing how you are performing against the target that you've given yourself and you'll all be proud about how good you are, or start making changes to see if you can become better.


If you believe some of the case studies, optimisation techniques can have dramatic effects. There are hundreds of case studies (here, here and here are just a couple) which should double digit improvements in conversion rates.

I personally worked on websites where our optimisation techniques improved conversions massively, where we came up with techniques of getting more people to do the things that we wanted based on the data that we had available and we came up with ways of increasing the volume based on search marketing techniques amongst others.

Is this stuff easy? No. It is hard, requires a lot of hard work and quite often requires lots of organisational changes. I've long since talked of maturity modelling to make sure that you are set up for optimisation.


These are things that you can do to set your business up to be able to do optimisation work. It looks at the processes of what you do, the people that you have to do it and the technology that they use. These are things that you should have your business set up to do, irrespective of whether you are doing an active optimisation process. When you are mature in these things then your business will be set up to succeed and the people in them will know what it is that they are doing.

This model should be set up for all of your activities when it comes to digital (and there are many in my organisation who are using this model in all sorts of channels beyond digital). Not just your conversion rate optimisation (CRO) your search engine optimisation (SEO or SEM), any email optimisation, your banner optimisation, your affiliate optimisation, your personalisation process, your A/B testing, your... etc.

This stuff is hard, takes lots of effort and will have real impact on results in the short term and in the long term on your business. If you are not doing optimsation now, you should be wondering why not and how you start doing it. And I'm not just saying that because I'm the Head of Optimisation at a consultancy (although if you do want to get in touch - there are some contact details on the left hand side).

Strategy

Strategy and Optimisation go hand in hand. Both have the same objective in the end: building the business to win.

When you do strategy, your job is different to Optimisation. Your job is to come up with the ways that will make you win the most.

When we were at school we learnt about iterative approaches to solve equations. The principle was simple - you took an equation, plugged a random couple of values into it and saw whether it was close to being true or not. When it inevitably wasn't, you then iteratively changed some of the values until it was close to being true.

The problem with iterative approaches to equations like this is that for particularly complex equations, there is a possibility that you will iterate away from the best answer to a local high, this is a classic problem in the postman equation.

With the postman problem, you choose a way that you think is going to be the best way to get to all the houses (say starting with the two outliers and working inwards) and then iterate until you end up with the shortest route (or of course the seven-bridge problem at Konigsberg that Euler tried to solve 400 years ago).

Strategy and Optimisation effectively are an extension of the postman problem. What we're looking at here is trying to find a way at the beginning that will allow your optimisation work to let you win. That's an oversimplification of the process. Of course you could come up with a strategy that gets you more in a winning position to start with than a losing one could ever get you even with the best optimisation.

How do you do it? Well you do research. You do as much research as your budgets will allow you to do. You find out what your customers and prospective customers want. You segment them, find out what they want, how they are going to react to the way that you present it to them and you find out what they are doing that would make them want to use your product or service.

Then you build your product, service, marketing or target audience (or any combination of those) around them. This is the point where you build into your business objectives what it is that you want to be doing and your Optimisation expert can take over.

What you then do is carry on updating your strategy on a frequent basis as firstly your audience changes over time and secondly your competitors change over time. This is a landscaping process that means that you will always be up to date and will always be winning, rather than optimising a strategy that is no longer leading to optimal results.

And of course I'm not just saying that because I work for an organisation that does strategy projects (although if you do want to get in touch, etc, etc).

Monday, September 09, 2013

The six presentation you won't see at Measure Camp 3

Measure Camp is coming this Saturday! For those of you who are blissfully unaware, Measure Camp is an 'unconference' held here in London for Web Analytics people. The unconference bit refers to the fact that we have no preplanned speakers and no preplanned presentations. Somewhere between 140 and 180 people are going to turn up on a Saturday at the Mozilla headquarters not quite knowing what to expect, but thinking that they might end up running a session themselves. At the beginning of the day there is a mad scramble as everyone who wants to do a presentation sticks it up on the board, then the rest of the people decide what session they want to go to. Some sessions have 80 or more people. Some have two or three.

What am I looking forward to on Saturday? Well if you think you are about to find out, you are sorely disappointed. Sorry. This is a top six Measure Camp sessions we won't see this weekend, for better or for worse.

1. How to send your boss data

What your boss really needs is someone to send him or her more data for them to try and work out what to do with. This session will teach you how to download data out of your tool of choice and turn it into a spreadsheet full of rows of numbers that your boss can ignore.

Yes that is right. In this session you will learn the beauty of a 'dashboard' with visits, time on site, bounce rate and a nice graph. Your boss is a busy man or woman. Why waste his time with things like 'analysis' or 'recommendations' when you can send him or her a graph that tells a thousand words.

Why not turn up to this session and learn why it is pointless understanding what your business does, why it operates and what your colleagues do with their day. Your colleagues know all these things so they'll be able to work it out from the reams of data that you send them. Why bother duplicating the effort of knowing all that stuff.

2. Why you should change your site to make your analytics tool work properly

There is nothing worse than setting up an analytics tool perfectly, only to discover that the site changes over time and suddenly your set up doesn't work any more. How inconsiderate to change the site in a way that breaks your analytics!

That's right, you will learn how to tell IT teams that they are only allowed to use standard set ups. Anything that is non-standard will need a new standard otherwise it won't be allowed. Nothing should be manipulated in a way that will allow it be multi-purpose if it means that some bit of content gets misclassified.

Marketing teams are often building campaigns that do odd things. How dare they? Campaigns are only allowed to go to set pages and the names should fall into the standard format. Anything that doesn't fit into that will have to be changed so that it can be reported accurately.

3. Segmentation is for losers

What your analytics system needs is more standard reports. This session celebrates not just the standard metrics of visits and page views, but also the standard dimensions of pages, campaigns and traffic sources.

They say that 99% of your analytics system is unused. This session shows that this figure isn't high enough. Now that 'hits' is no longer a popular phrase in the business, we describe the best ways of showing page views using different names such as 'instances' or occurrences. Yes that is right, you can report on your highly complex site with nothing more than site visits and average time on site.

Why bother with segmentation when you can set up complex reports in advance based on ever increasingly complex and difficult to maintain tagging requirements. Instead of wasting your time hiring analysts whose job it is to look at the data, why not hire them to make sure your tagging is set up properly.

4. Visitor Level Data is possible

Cookies are just the start for this session or for any session in the online world. But they aren't the end in this case. Yes we look at how you can tie together increasingly complex user interactions across a multiple of different devices in different locations.

So what if when the user finds out they run away from your brand screaming. So what if the incremental value that your business people get from the additional data isn't significant. so what if you have to come up with increasingly legally dubious ways of making sure users can't escape from their cookies. So what if we force sites to insist on login so that we can track the users, to their detriment. So what if we ignore all the 'anonymous' use, including from before the user logs in.

What we end up with is a very powerful bit of data. Yes that's right, exact user interaction data that you can use to market to them. You can even print a picture from the users facebook profile and stick it on a balloon that you hang in your office to show that, despite the fact that you are treating them as a data point, they are real people.

5. I know what my customers want, so why waste money asking them

Years ago the head of a website told me that he, or his sales team, personally knew everyone who might possibly want to sign up to his subscription website, so why would he possibly need us to reach new audiences through SEO?

This wasn't the same person, mind you, as the one who having been given a load of data on the performance of his online quote form and having had a lot of usability done by watching users go through the form decided that five minutes of looking through it himself was far more valuable. What do these people know - he'd worked in IT for ten years and knew everything about forms and so pushed for his changes to be made.

And they were both right. What a waste of money user research and usability is. You are far better off just coming up with ideas off the top of your head and then putting them live. If the users don't like them, then they'll learn to like them after a while.

6. Analytics should be an IT project

Once you've got your analytics tool live, then it produces data on its own, right? So why not make the implementation an IT project that can be delivered on time and to budget. That way you can get on with spending your budget on advertising.

Everyone knows spending money on analytics is basically just taking away money that your are going to be able to spend to drive more people to the site. Instead of spending on hiring people to do the measurement, we can put it into a new banner advert. Instead of hiring people to make recommendations, we can put a new PPC advert live. 

In fact, instead of investing in a tool, why don't we hire Kate Moss to be in our new TV advert? That has got to be more valuable.

Thursday, August 22, 2013

Conversion Funnel Analysis in WebTrends

Conversion funnel analysis is so 2008. But I'm in a retro mood, so we're going to revisit it again, this time updating the series to show how you do conversion funnel analysis in WebTrends. It is quite worrying actually that 2008 was when I wrote my original post on Conversion Funnel Analysis: When, How and What. This was the definitive guide on how to do conversion funnel analysis back in the day.

Of course you all lapped it up making it the most viewed post in the first five years of this blog, which is why there were follow ups for how to do funnel analysis in Omniture SiteCatalyst and how to do funnel analysis in Google Analytics.

Webtrends funnel analysis tool is called Scenario Analysis (there is a great three part series on it by Sandra Elliot on the WebTrends blog - albeit on version 9, not the latest version).

The fact that version 9's scenario analysis takes up three blog posts to describe how it works suggests that it was overly complicated, so I am going to focus on what it looks like in version 10. But before we get there, lets revisit the point of conversion funnel analysis.

Why do Conversion Funnel Analysis

Conversion funnel analysis is what you use to work out why you aren't getting enough people to your conversion point. If you can identify where people drop out of the process you can make changes there to see if you can get more people to the end.

And if you think that there aren't changes that you can make to get more people to the conversion point, you'll know how many additional visits you need to pour into the top to get those additional conversions.


Depending on your funnel type, will depend on where you should be focussing your efforts of making change. Whilst we like to visualise them as types of glasses that you serve alcohol in, the realisation of what causes each, will allow you to make a change to get more people to the end.

Scenario Analysis in WebTrends

In WebTrends they have a visualisation of the funnel analysis that allows you to work out where people drop off.


Here we have a four step process that shows where people are dropping out of our site through to a quiz download.

What's the important information in this picture?

13.9% represents the number of visits at the last step divided by the number of visits to the first step. I worded that very carefully, you might have noticed - I didn't say 13.9% of the visits in the first step converted. What is the difference you might say?

The difference is that some of the 191 at the final step might not have hit step one. Here we have an intrinsic failing of the conversion funnel - frequently there is no 'path' that the user has to take through the site to get to the end. And even in cases where there is a definite path, there are frequently branches that go off in one direction or another that can distract the user.

Anyway, going back to the report that you were looking at. There is a nice line at the top of the report that allows you to see for a step how many entered the process at that step. This will allow you to work out some of the reasons why you are getting more visits in a later step than an earlier one (in version 9 you can see where they came from).

There is also a drop off percentage for each individual step. In the example that you can see above it does actually work quite well as a path, but the major drop off is whether people download the final summary or not (in actual fact, the download is a conversion point, but we're actually also interested in step 3 as well, because this is the end of the process).


What should you do with these reports


These reports are there to give you insight on ways that you could optimise your site. They suggest where the drop off points are, but they won't give you the answers. The only way of getting the answers is to ask you users or by testing.

The other advantage of this report is that you can test things by making changes and see what the impact is on the drop off points.

The other thing that you need to think about from this report is whether you are getting enough visits in at the top in the first place. This is where you can analyse whether your marketing is working as it should be (although I think this could be improved by WebTrends).


How could WebTrends improve this report

I think this report could do with something additional (I've already put my wishlist for WebTrends 12 - somehow I managed to miss out a version). That additional thing is that I need to be able to segment this report, by campaign, by keywords, by whatever. 

Without the ability to segment, I currently make lots of custom reports with these 'steps' set up as measures, because that gives me the segmentation I need. It's annoying though, because you can end up with hundreds of custom reports that do very similar things. So come on WebTrends - hurry up and sort your segmentation out!

Build Custom Reports

My advice is to build custom reports with your KPIs for the segments that you've identified in your measurement framework. Once you've created the custom reports then you can use REST to take the reports out of WebTrends and into Excel.

Once in Excel you can then visualise the data how you wish. Although it feels like a bit of an indictment of the tool that you might want to take the data out at the first opportunity, many other tools make lots of effort on data visualisation and then the precise one that you want to use isn't available.

Wednesday, July 31, 2013

The Great FireWall of Cameron is bad for non-parents, parents and children

Net neutrality is an important topic - far more important than a simple Analytics blog could ever hope to cover - and covers many different areas. From ensuring that network operators don't filter or throttle particular connection types for the benefit of their partners to making sure that Governments don't legislate against sites who comply with the law but don't fit with their political leanings.

Enter David Cameron. In what the Daily Mail proclaimed victory for them, David Cameron has vowed to ban porn. More specifically every person with an internet connection in the UK will be asked whether they want to have 'family friendly filters' put in place and all new connections will be given it by default unless they choose to opt out.

We'll ignore for the moment that David Cameron appears to be talking about something completely different to the ISPs (as this leaked letter demonstrates) and that this isn't actually legislation, it is Cameron abusing his power as leader of the country to request something that would get through the Commons or the Lords if it was turned into real legislation. Let's look at why it is a bad idea in detail.

This is bad for adults without children

This isn't a porn ban.

I happen to have my mobile with O2, who have implemented an over 18 filter already, unless you choose to opt out (which I haven't done yet, mainly because O2 seem to make it more difficult than a massive tick box on my login screen).

O2's filter has been plagued by problems where they've filtered out things that they shouldn't do. Specifically they've filtered out the BNP's website. However unsavoury you may think the BNP are, they are still a legitimate politically party with an MEP, two councillors and a European Election coming up. How am I meant to know if I want to vote for them (I don't) if I can't see their website.

The great firewall of Cameron isn't a block on porn, it is the potential for a block on political discourse. If you don't have children and are wondering whether you should opt in to the ban, you absolutely should not.

For net neutrality this is awful. It has the potential for ISPs to start blocking sites that they claim are for adults, but in reality offer rival services (look at the fact that Torrent Freak often falls foul of these filters, despite being a legitimate news site).

Moreover this is particularly bad for business. Imagine if you site was given a false positive as a porn site and it suddenly got dropped from a load of ISPs. What would it take to do it? Will my blog be blocked merely because I am talking about porn? Will a site with an image with the wrong url be banned? What happens if I have a business where one of my major traffic referrers gets hit by the ban. Are the Government or ISP going to reimburse for lost revenues?

Moreover, how will something like an adult site be distinguished from page 3 (Cameron has said he wouldn't look to ban this) or even the Daily Mail website itself, who as many gleefully pointed out has its own right column of shame?

This is bad for parents

I'm sure most people are aware of the continuing ban on websites such as the Pirate Bay, as issued by the UK high court. This giant game of whack a mole being played out by the big music and film associations as they try and close down file sharing sites, which are instantly replaced by newer sites.

The problem with a 'porn ban' is that is essentially the same whack a mole process, but with nowhere near the self interest that the entertainment industry is offering.

Either the list of sites that are filtered is going to be out of date almost as quickly as it is implemented if it is created manually. Or the list is going to be automated based on keywords in which case it will be relatively easy to break unless it is constantly updated (l33t sp34k was entirely created because of usenet forums banning particular words).

So the system doesn't work, but parents think it does. They spend less attention to their children whilst they are on the internet under the false apprehension that their kids are 'safe'.

So if I was a parent (and I'm not, so I suppose you have to take this with a pinch of salt) I would not be implementing this ban for my kids. I think it would give me a false sense of security. Instead I would opt for a more specific programme on my computer and mobile connections that I could turn on and off when the children are using it. More importantly I will be making sure that my children have some level of supervision whilst they are on a computer (until they are old enough to make their own, informed choices).

This is bad for Children

Ok, well I might be exaggerating this point slightly. The ban mainly has no effect on children because the pages that are banned won't be appearing on their screens anyway. If a child does 'stumble' across a port website - which sounds hideously unlikely to me - it will either be filtered because its on the list (possibly enhancing their curiosity to find out what is behind it) or they will see it because it hasn't hit the filter yet (which is far more likely).

More importantly, we're bringing children up to believe that internet access is something that can be controlled by government and it isn't.

The internet is a wonderfull thing, but it is just a tool. It is a tool that allows us to create, share and learn. Instead of blocking porn for kids, we should be explaining to them what it is and why it isn't representative of real life. Prohibition has never worked and we aren't learning the lessons of the past.

 
Blog Directory - Blogged