(not provided) SWOT – its getting bigger, what you can do about it with Analytics
Here’s where SEO came from
I started in online marketing before search engines were big enough to be a viable marketing platform. Search marketing was very, very basic in 2001-2003. By comparison to the glitzy big data world of today, it was all really cave man stuff. There was no Analytics. There was no Webmaster Tools app. There were no Matt Cutts’ videos. There was no YouTube actually.
There were no funnels, except the one in the kitchen. The only time you worried about spam was when somebody old made you eat it. A hit was something you got in the face or on the streets of Dublin, a ninja was something Bruce Lee hit. In the face. A lot. A conversion was something people in the Bible did and a goal was when somebody sporty put a ball into a net of some description. It was really, really basic. It was so basic it wasn’t even a job. I used to tell people I was a car clamper, just because it sounded better than being an SEO.
We used server-side apps that recorded hits (server requests, and a dull metric), visits, IP addresses and used the ISP TLD as a form of crude geo-locating service. There was no AdWords. There were no search ads. There were 1000’s of search engines. There certainly was no Analytics.
A word about (not provided)
It’s no joke – (not provided) means making more work for webmasters and SEO’s – and in business, that means money. (not provided) is to privacy as a chocolate fire guard is to safety. It’s completely hypocritical – and deeply insulting – because AdWords has no such privacy. Users of other search engines don’t care. It’s not an issue. It is however, a great way for Google to fight a proxy privacy war instead of actually looking after your privacy with all the data THEY collect, not the data they deny.
The basic principle of search is thus: People search for stuff. If you want to debate SEO, then go ask people why they use search engines or read this blog post about What is SEO. The more you rank for a keyword used that is related to something you sell, then there is a high probability, that if everything works, people will buy it from you. When I say high probability or proportion, it may be around 1% or it may be 25% or it may be be 75%. It all depends on a multitude of factors.
This is SEO, this was SEO and this will be SEO: Pick the keyword ranges you want to rank for, and do everything possible to maintain it.
SEO plus Content Marketing – this is a big, big win.
Getting Intelligent about it
Getting more and more data to analyse is both good and bad. Firstly, we would love full loop sales circles. We want to be targeted. In many ways instead of learning about long tail from AdWords, SEO’s instead tried to learn about data analysis. This is pointless. If you sell phones and people search for phones, then know this: while 20% of people who search for phones might also buy a phone not everyone (i.e. 80%) of people are buying a phone (at this point/time/website)
KILLER TIP: I forecast that somebody could easily build a web app to take GWT data and Analytics Goals and give you middle level (one step up from exact) data about keyword conversions based on measuring pages that rank for a keyword and goals converted from that
- SEO will need more time and old style SEO’s
- Arguably, this could free up SEO’s time from Keyword obsessive clients. SEO shouldn’t – only blackhat linkbuilding could deliver – specific keyword changes and ranking. A good content plan with good SEO (aka Contextual Content Marketing) will deliver broad and narrow keyword ranking at the same times
- Better SEO – less tempation to do blackhat
- Freedom to focus on the Rank, the depth and the breadth
- Potential to realise that bad content, or high bounce content could be doing long term SEO damage
- Specific Keyword targeting – this will hit ecommerce sites the most. Web businesses, not SEO’s, will suffer here
- Too broad
- SEO’s having to explain (not provided) to customers
- Google’s GWT Tool isn’t good enough and doesn’t provide enough click data – less than 50% in some cases
- Better co-operation between SEO and PPC
- Remember that PPC reports into Analytics at a keyword level
- SEO’s can track specific keyword matching patterns and then check GWT for ranking and CTR
- Low Ranking keywords should be sent to PPC and high-ranking can be offered to the PPC manager to save or reduce spend (although this will never happen)
- Better SEO focus on broad and long tail
- Better focus on R&D budgets rather than on ranking for the one “big term”
- Find new keywords
- Get more leads
- SEO will get more expensive – the reports to combat (not provided) doesn’t protect anybodies privacy but it will take hours longer to produce
- More SEO is dead type stuff from your traditional marketing colleagues and peers
- More SEO is dead articles from PR and “journalists” showing up in your Google Alert feeds
- Coffee prices going up
SEO is the same thing today as it always was
Everyone says SEO is changing. Its been changing since day one. It changed when Analytics was born – and it had to. We were trying to build our own (crude) version but when Google launched Analytics, we saw it was much better and dumped it. We used keyword planing tools (not AdWords, because, frankly, it didn’t exist). I still use exactly the same software today as I did 10 years ago, its just massively bigger and massively more complex. We have a quad-processor Dell PowerEdge server to run it. It’s brilliant.
So when Google pulled the Keyword Tool and replaced it with the Keyword Planner, it was no loss.
We take our keywords and we map out a content plan. If we’re selling phones, we write up a space for a page for every type of phone sold and we group them by features. We build a sitemap and we call it a site plan. We look to who are competitors are and who our friends could be – and we connect with them. The siteplan is given to the web designer to fit into their aesthetic view of the world, and the sitemap becomes, well, a sitemap.
As sites got bigger and bigger, and SEO moved mainstream, we found ourselves working on sites with 1k monthly visits which then became tens of thousands, and then 100’s and then millions. And long before we got to 1,000 different search words, I gave up caring. Its just too much data to obsess about and doing so doesn’t change it.
Keyword obsession is like watching pots boil. You may want to, you may feel compelled to watch it. But it will boil whether you do or not. And ask a few choice ex clients of mine who chose Keyword obsession as a new hobby – it doesn’t work for me.
What should you do?
Look at patterns and trends. PPC managers are not all numbers experts. Analysing what people buy on what words is important to PPC because you have to reduce waste on “commercially unviable keyword targeting practices.” Please take it from someone who last year managed a very generous amount of PPC spend – while we talk targeting, and measuring – there is no 100% in PPC (we did brag once about 78% but that’s different :)) – and what worked last year doesn’t work this year and an awful amount of analysis quite often ends up going back to broad.
Take a look at PPC – PPC managers don’t just bid on keywords like this: [iPhone store]. They also buy “iPhone comparison” this might indicate intent to buy but not by any means an intent to buy from you.
Analysing past buying habits, keywords should give you a pattern, not a roadmap. Use your intuition and grow your gut feeling. Test and experiment. Analysis of past data is as useful in gambling as it is in selling.
Use GWT like a PRO!
I have a good habit of correctly predicting SEO trends and when asked about Page Speed, I said that Google Webmaster Tools’ Search Queries report had launched that same day and was a much, much bigger event, I was pretty bang on
The search query report is an SEO agency report in one. Even if its missing data and looks like FUD. Here’s the 10 minute hack:
- The Filters tab lets you see all Keywords by country
- You can star your branded keywords to separate brand from Generic
- Total up the branded search clicks, subtract from total and that should give you total generic
- Plot the trends monthly, to build up a generic vs branded monthly
- Branded search is not the problem for SEM. Everyone still believes in Mad Men Era marketing – that the brand is everything.That’s why they don’t get search. All they need to know is that if branded search is down, they can blame
- The PR people
- The events people
- The product engineers
- You’re just here for Generic
- If generic is growing higher than branded, chances are, that you’re doing a great job. End.
In my case example, The Auld Sod Irish Gifts website (a hobby-horse of mine), this is our branded search, which is quite narrow (handy!)
So in the first post, total traffic seemed to be down by about 20%. In this case it looks like the brand is down a whole lot! So time, to take the sales and events team outside and shoot them. But lets not be too hasty, because unfortunately, now isn’t a great time to sell Irish Soil – its a Thanksgiving, Christmas and St Patricks Day product, with little or few sales in between. So we might just have to plot this over time.
The best way to do this is to view it Analytics:
Analytics is easier, because you put in search phrases – so you can exclude your branded terms or include them, making reporting much easier!
Getting over (Not Provided) and conversion tracking
Well, GWT gives you a lot more great data that 99% of SEO’s and Webmasters miss – it also tells you which pages rank for a keyword:
If you take conversions and work from pages, you could deduce, that because this page ranks for “Irish Gifts,” then conversions from this page that AREN’T from other sources (i.e. were Organic search landing pages), then a report like this in Analytics will do the job
Create a segment with Organic Search Traffic and set either Conversions >1 or e-commerce value >€0
Notes: In a Direct traffic world, where search traffic is reported as Direct (worse than (not Provided) – you might want to drop Source = Organic – and just pick landing pages regardless of source!
Now go to Content > Site Content > Landing Pages. Make sure just this new segment is enable:
What this tells me is this:
- Pages that people landed on from a search
- I presume I don’t know the keyword (but you can add it in just in case it came from somewhere nice like Bing, Yahoo, or Blekko —> I recommend giving these a shot!)
- I can then check the keywords these pages rank for
- I can then divide my total sales by pages based on the traffic they got
- So if a page got 10 conversions and it had 100 visitors and it ranked for “Irish Gifts” and “Unique Irish Gifts” – then I can apportion that bucket of keywords to these conversions
Doing a new Post (not provided) Report
- Take from Google GWT or reporting software
- Total Sales / leads / conversions (where applicable)
- Total Traffic
- Executive Summary
- Did Traffic go up?
- Did SERPs Go up
- What experiments did you run?
- What did you target
- Did you hit the targets
- did the traffic go up or is still growing
- Did sales also go up?
- what is the % change by traffic source
- Is branded search going up or down (this is almost always internally controllable)
- Search Traffic
- Not Paid
- Branded (estimate)
- Referring traffic
- Top, dropped and new
- Direct Traffic
- The new (not provided) !
Why I think this doesn’t matter
Yes its important to know that “Irish Gifts” got you 100 sales. But that could be out of 10,000 visits. The thing is, the keyword is not the person. Also, knowing what to rank for doesn’t immediately result in automatically being able to rank for it (something that 99% of business owners would do well to know too)!
As I mentioned in PPC – we often go targeted. But going targeted may mean higher PPC (e.g. the CPC difference on “Job in Ireland” versus the CPC on “jobs for doctors” is 1000%) – it also means excluding long tail (accidentally AND intently) and it also means losing serendipity.