The Evolution of Search, from Google

Insanely cool video from Google showing the evolution of search alongside the evolution of the web.

Yes!!!

Ever wondered how you get those maddeningly targeted ads?

You know how you get those super-targeted ads that sell you stuff based on very specific personal information? Ever wondered how that works.

Basically, it's this.

1. You visit a site where you enter some personal information or where some information about you is inferred. (Your location can be inferred from your IP address).

2. A cookie like this one is placed on your computer. See how it knows my zip code and age?

Yes, I am 37 and I live in 97211

3. Next time you visit the original site or a partner site, it looks at your cookie and serves up ads that are relevant to you.

Simple and deadly.

Natch'

Ever wondered what makes a yummy cookie?

Here's a simple and tasty breakdown of what is in a Google Analytics cookie.

Cookie ingredients.

Yum!

Graphic courtesy of Google Conversion University.

Avinash: 'Data Quality Sucks, Let's Just Get Over It'...Heck yeah!

In the immortal words of Avinash, web analytics ninja, 

'Data quality on the internet absolutely sucks. And there is nothing you can do about it. At least for now.'

There are so many reasons for data quality to be far from optimal and Avinash does a great job of listing them and explaining why they don't really matter as much as technical analysts and some clients believe.

But he also correctly (IOHO) states that you should be happy with 95%.

This means that sometimes data will not tally exactly between what traffic your media company says they drove and the actual traffic you see on your site. It also means that your analyst will sometimes have to re-state results when better data becomes available. 

It also means that you will be caused to look goofy in front of the detail-obsessed VP or junior producer. (Why is it that the people at the extreme poles of the food chain get caught up on the most pointless things? Maybe middling managers are too busy to worry about it).

Here's our advice for dealing with all of this:

1. Take a chill pill. Stop, breathe, think about how unimportant a +/-10% variation is in the grand scheme of things. Remember that...
2. We are looking at trends. If one measurement tool is consistently lower, even by an order of magnitude of 50%, that's fine, as long as it's consistently lower. The most important thing is how the results vary in relation to the same measurement with the same tool last week/quarter/year. Accuracy is much more important than being over-precise.
3. Focus on insights. 'How can we trust this recommendation when the data quality is untrustworthy?' Our response is that you're focusing not on the trees, but on the tiny bugs that live in the leaves. Think bigger - what is the overall picture? In most cases, we find that data tends to back up your hunch anyway so you should only be concerned if the data is telling you something completely incredible, as in unbelievable. If it looks and smells right, it probably is. Think big.

Big picture

Here's one piece of advice that will help you and your insighteer immensely. If you do have concerns, raise them privately with your analyst, not with everybody and their Mom and their Mom's SVP cc'd on the email. It's probably a simple thing to answer and it may be that you misunderstood something. Every time you raise a concern about what you think is data integrity, you make life harder for everybody....including yourself when you need that critical rush piece of information. Don't get the reputation as the consumer who is so focused on the details that they miss the big points.

Waste of time - zero insight

We are definitely not saying that you should not be as rigorous or analytical as you possibly can but if you just follow one piece of advice, it should be this: STAY OUT OF THE RABBIT HOLES. It's so easy to spin and spin on getting that extra 2% or 3% of of precision. Some analysts love that sh!t. 

But they're usually the ones who produce those crazy dense 'dashboards' with zero insights.

Think accurate, not necessarily precise to the nth degree.

Suh-weet Data Visualization

In our last blog post, we talked about the importance of excellent data visualization.

Here's a pretty delicious example of a dense graphic we produced.

(Dense in a good way).

It shows relative search traffic of various shoe brands.

The thing that is most excellent about the graphic is the amount of information it contains, on so many variables.

The Five Big Fat Lies of Web Analytics (5/5) - Looks don't matter!

He has a lot to say.

But he's not too easy on the eyes.

Could those sentiments apply to your latest report or dashboard?

Ugh.

The truth is that most reports are ugly enough to make your eyes bleed. That might be perfect for Halloween but not so cute when you're trying to extract meaning from a bunch of tables and graphs. 

This is because many reports are produced by report-monkeys or are automated. The goal may be to show you how much data they have access to. 

Dense, cluttered, unnecessarily granular. 

Yuck.

Of course, the underlying data need to be remorselessly reliable but that's what appendices are for. Marketers need digestible insights they can act on - we call them action nuggets - not data tables and pie charts. Definitely not the traditional pie charts, even if they are exploding and especially even if they are 3-D. Tufte would turn in his grave. Especially if he were dead.

Information design is as important as the data you're trying to digest, if the recipient is to have any chance of applying the insights.

There, we said it.

State of the art APIs won't give you insight.

Great judgement and analysis from humans who understand your business and can express themselves visually will.

Want more Lies?
Check it:
1. It's all about web traffic
2. Apples vs. Oranges
3. You need....
4. Dashboard's live; our work is done

We eat what we preach. Online ads for site traffic maintenance.

One of the things we advocate to our clients is to maintain at least a basic level of advertising presence even if there's not that much going on or not too much to speak about. Definitely not at the same intensity level as during a big campaign but just something to tell the market that the lights are still on.

It helps to break through the clutter and may even help your SEO.

We put our money where our mouth is.

Having noticed that, outside of larger promotional moments such as articles being posted, traffic to actionnugget.com dropped hard, we set up a modest, targeted ad campaign using Facebook.

Action Nugget Facebook ad

The immediate result has been that traffic between big news days doesn't drop as fast. 

You can see that after 10/25, we don't get those traffic doldrums.

Traffic to actionnugget.com

Obvious maybe but many brands go completely dark between campaigns and have to start all over again next time there's something big to talk about.

Keep the lights on, even if they're turned down low and cosy.