Sometimes it’s hard to know whether you should blame an author or his readers when a reasonable enough idea gets taken way too seriously.
I have in mind the now (in)famous assertion by Avinash Kaushik that companies should follow a 90/10 rule in their web analytics spending: 90% for people and 10% for tools.
The idea behind the assertion is essentially sound. Companies often make a fundamental mistake thinking that web analytics software can simply be grafted onto an organization, be used by designers and product managers, and somehow make a positive difference. There are so many fallacies in this type of thinking that it would be hard to know where to start – and I take the 90/10 rule to be a useful corrective to these bad ideas.
This is fine as far as it goes, but I have more than once been asked by entirely serious people if they should really be spending X on web analytics based on the cost of their tool Y. By the time I wipe the look of shocked – can you really be asking me that – disbelief off my face and try to pretend I think it’s a reasonable question the damage is usually long done.
As with any rule of thumb, strict adherence to it is logically absurd (you got Google Analytics for free so you can’t let anyone look at it or your ratio will be out of balance). But even taken as a general rule of thumb the 90/10 proposition is sadly inadequate.
It’s not as if such rules of thumb don’t usefully exist. In the restaurant industry, for example, there are some pretty general guidelines about what percentage of your cost should come from ingredients, staff, etc. And these guidelines can help restaurant managers get a good sense of whether or not they are running “lean” or “fat” in a particular area.
But there are, of course, different types of restaurant. And the rules of thumb have been adapted to these various types (fine dining, fast food, family, etc.). I don’t see anything like this in the 90/10 rule – and I fail to see why anyone should think a ratio appropriate to Intuit is also appropriate to the Ideal Cheese Company. Nor do I think there is anything like enough real-world experience to even begin to establish a reasonable ratio for any type of business.
I think I’m safe in saying that the variation in need/tools/people is far greater in the web analytics space than in restaurants of a specific dining class – far too great to believe that because company x has a ratio of 80/20 or 50/50 or 99/1 that they are necessarily running lean or fat.
And while I accept the basic underlying claim that you can’t do web analytics without dedicated staff, I’m more than skeptical about the often associated claim that the tool doesn’t much matter.
Your web analytics tool is a force multiplier – and a basic enabler without which any number of people will be totally unable to do the job. It’s like the gun a soldier carries into battle. It won’t matter how many troops you have or even how good they are, if their guns fire blanks, you’re going to lose.
- If you need your web analytics solution to produce over-time customer segments to do useful analysis then you need to spend whatever it takes to get a solution that produces over-time customer segments. It doesn’t matter if that’s half of what you are going to spend on analytics – without the capability the people are useless. There’s just no point in throwing people at a tool that won’t do the necessary job. And if you don’t think there are tools out there shooting blanks then you haven’t tried to fire all of them!
- So here are some very unglamorous rules of thumb about spending on web analytics – they aren’t “neat and tidy” like the 90-10 rule. But they have advantage of actually making sense.
- If you can’t afford an FTE for analytics and you want more than traffic reporting you should almost always outsource (this is self-serving but it’s also true).
- Buy a tool that does the job you need to get your specific analytics done. If you don’t know what that is, then get the help you need to find out (see note above). What it will cost you is far less than the time you’ll otherwise waste.
- If you are building your web analytics internally, hire people until the incremental return on their labor isn’t worth it. That’s when you’ll have spent enough (Okay – that’s true but not actually helpful – so here are few tidbits of actually helpful advice)
- Attach your FTE’s to your marketing and product channels – don’t isolate them in their own group. Analytics doesn’t need to departmentalized – it needs to be responsive to business owner needs. Find a way to cross-pollinate these people without undue organizational structure.
- Hire doers first and attach them to BU’s where they are managed by marketers and product managers. Hiring an analytics manager without any Indians is worthless. Hiring one manager and one Indian is a stretch.
- Don’t try to hire everyone at once. Let your analytics grow organically – by dispersing analytics throughout your organization you’ll increase your ability to absorb new hires.
- Don’t assume that just because you have good people that they know everything – chances are they’ve worked on your site and your tools and pretty much nothing else – so they may be very good at what they do but still have significant blind spots.
All of which amounts to just a small part of what might be said about structuring your analytics and allocating resources. And will, I hope, convince you that when you read something like the 90/10 rule you should NOT immediately start dividing your analytics budget by your tool cost. Oh – and don’t ask me stupid questions either.
Add to 
Bookmark murdok:
Gary Angel is the author of the “SEMAngel blog – Web Analytics and Search Engine Marketing practices and perspectives from a 10-year experienced guru.
