Statistics, R, and You: Advice for a New Analyst?

[sweetfancymoses]

Hey SEOMozers!

Two prongs to this question; I'll keep it succinct.

I've been working as an in-house SEO/SEM Analyst for about 5 months now.  While I'm generally savvy at telling the story behind the traffic/conversion data, and making forensic recommendations (I worked in SEO prior to this while in college), ideally I'd like to see my reports read less like these piddly Excel charts and percent change statistics.  Ideally they'd look more like Nate Silver's FiveThirtyEight blog for the New York Times, or OkCupid's periodic dispatches on OkTrends: visual, statistically-informed, and predictive, the kind of report that under other circumstances might plausibly generate backlinks.

Data analysts swear by R for statistical modeling, but is it useful for our Google Analytics data sets, holes and uncertainty and all?  Is the steep learning curve worth the effort?  Tutorials I've seen online assume a proficiency in programming or statistics that's beyond me, or they're written to support a textbook exercise.  Any recommendations for a book, online course, or general resource with more of a niche focus?

And a general question about stats too, since it's related: what level would you prescribe if I really wanted to kick this up a notch?  I studied a humanity in college and while it helps with the numerical storytelling, I wonder if the practical arcana of Bayesian Methods/abstract probability theorems have a place in Web Analytics.  Do they?  Are there options for us bushy-tailed young analysts to pick this up without resorting to B School?

Thanks in advance!

[DanielFreedman]

What Steve said.

Especially:

"Slipping in one or two suggestions about direction and budget in the verbal reports can be very beneficial."

I've found  verbal briefings every week or two are far more effective than written reports. Sometimes we take a quick dive into the raw data in a "guided tour" type of way. Often that goes away over time. But first you have to establish trust. That means getting some wins user your belt.  And it also includes  fessing up to failed initiatives.

I see a lot of people get hung up on metrics of questionable significance and obsess over reporting. They never ask a basic question: "Is any there evidence my reports are being read?"

Of course, meetings and calls are far more easily arranged with small business owners than in big, bureaucratic organization that lack a single decision-maker. But I'd move heaven and earth to try to get a monthly 20 minute meeting with someone as high up the food chain as possible. Pretty soon, he'll start inviting others and stuff will really happen.

[stevecounsell]

Hi,

I've just replaced an "In House" SEO with a consultancy model.  The process related to me from the in-house guy was very bureaucratic as you have said.

From a consultant perspective I need to prove what I do does have a difference to te bottom line and then this builds trust.

Work out what will make a difference to the bottom line and attack that first. In my situation, they'd been running AdWords unsuccessfully for a while so I managed to get a small budget (£100) to prove a point. We ran a campaign for 2 weeks and got a noticeable spike in traffic and business. This has built the trust in my approach and process enormously.

On the reporting side I only report once a month in writing with other ad-hoc verbal reports of how things are going. Slipping in one or two suggestions about direction and budget in the verbal reports can be very beneficial.

Report less - Do more.

When reporting try to focus on the "what you did, what outcome you expected and what you learned" This should then feed back into an SEO Plan and overall strategy. In terms of SEO I've found business leaders like strategy.

Hope that helps

Steve

[sweetfancymoses]

Love a good wacky, "out there" response to keep my intuitions in check!

I get what you're saying.  That's happening, but in-house SEO/SEM is a slow and measured process, subject to approvals that go far up the chain of command.  I figure if I can at least present my recommendations in a more salient way, they'll be able to defend themselves whether or not I'm in the room.

We also, as with most hotels, deal with affiliate sites.  The cost/benefit of involvement there has considerable room for interpretation, and while it isn't exactly my onus to understand it, I'd like to.  Basically looking to grow here.

[EGOL]

Right.... make a change on the site and show how it increased sales.

Show what you have done that has hit the bottom line.

[DanielFreedman]

Ready for a really wacky, "out there" response?

Stop writing so many reports and start trying to influence decision makers in other ways.