Numerous studies have pointed out that while almost all Fortune 500 companies have great investments in "Web Analytics" they still struggle to make any meaningful business decisions. Most people complain that there are tera bytes of data and giga bytes of reports and mega bytes of Excel and PowerPoint files. Yet no actionable insights, no innate awareness of what is really going on through the clutter of site clickstream data.
Through my humble experience in this field I have developed a rule to fix this problem and achieve Magnificent Success. I call it the 10 / 90 rule. Here it what it says……..
- Our Goal: Highest value from Web Analytics implementation.
- Cost of analytics tool & vendor professional services: $ 10.
- Required investment in "intelligent resources/analysts": $ 90.
- Bottom-line for Magnificent Success: Its the people.
The rule works quite simply. If you are paying your web analytics vendor (Omniture, WebTrends, ClickTracks, CoreMetrics, HBX, etc) $25,000 for a annual contract you need to invest $225,000 in people to extract value from that data. If you are actually paying Omniture, WebTrends, HBX etc $225,000 each year then…. well you can do the math.
Most people reading this post probably think this is way overblown or silly or just plain stupid. I can understand that. Here are some of the reasons I have come to formulate this rule:
- If your website has more than 100 pages and you get more than 10k visitors a month you can imagine the complexity of the interactions that are happening with your website. Drop in marketing campaigns, a dynamic site, SEM, more pages, more traffic, promotions and offers and you have a very tough situation to understand.
- Most web analytics tools will spew out data like there is no tomorrow. We seem to be a rat race, one vendor says I can do 100 reports, the next says 250 and the one after that says I can measure the eye color of people who look at your web pages and on an on. Bottom line is that it will take a lot of intelligence to figure out what is real in all this data and what is fake and what, if anything in the canned reports, is meaningful in all this.
- It is a given that if you open most web analytics tools that they show the exact same metrics, almost all of them measured and computed differently! You are going to have to sort this out.
- Finally actionable Web Insights (or as I have now copywrited: KIA's, key insights analysis) does not come simply from ClickStream, you are going to have to have people who are smart and have business acumen who can tie clickstream behavior to other sources of data / information / company happenings.
A part time person, or your admin, providing access to your favourite expensive analytics tool can't help your management make actionable decisions.
So if you think your company is not following the 10 / 90 Rule then here is my humble recommendation for you to consider:
- Apply for a free Google Analytics account at GA Sign Up Page
- Once you get the code implement Google Analytics on your website in parallel with your favorite expensive analytics tool
- Get a comfort level for delta between the two sets of key numbers (you know visitors, conversions, page views etc etc) and create a multiplier (my tool shows visitors 10% higher and page views 10% lower than Google). You will use this multiplier in future to compare year over year trends if you want to.
- Cancel the contract with your favorite expensive analytics vendor and take that $50k or $100k or $200k and: 1) Hire a smart analyst for between $50k to whatever maybe your areas great salary 2) Put the rest of the money in your pocket.
- Your smart analyst will be able to extract just as much value from GA than your old tool, in fact my prediction is that it will be a lot more.
- As the level of savvy in your org grows, as the level of sophistication of supporting processes increased, perhaps in two years you might be ready to plunk down $200k on a web analytics tool and then be ready to extract a corresponding amount of value from it.
The cool thing about the recommendation above is that even if you get to Step 3 you can walk away, no harm no fuss and you would have learned something. But I hope that you will go through all the steps and provide folks like me with employment and add strategic value to your companies by providing actionable insights rather than reports.
Agree? Disagree? Feel like I am posting from la la land? Please share your feedback via comments.
Totally agree. There is so much data contained in GA reports alone, that any company serious about their business should assign/hire someone to analyze all this precious data.
It would be rude of me not to start with a bias confession: we sell and market a web analytics tool.
That aside, in many ways I agree with the notion of a 10/90 split (regardless of the tool), but I'm not convinced that "analyst" covers the real role of the 90 for most organisations.
"Actionable Insights" are great, but at the risk of stating the completely obvious, even more important is actually taking some action.
So, for every insight generated, you need human resource with the time to modify the campaign, change the buying process, fix the call to action, provide more customer support information, adjust the pricing — whatever. There's at least $90 there for every $10 spent on a tool.
A dedicated analyst can add amazing value from a strategic sense (Avinash — what you have achieved at Intuit is testament to that), but if ALL metrics must be channeled through an analyst this only reinforces a sometimes misplaced mystique about what the numbers mean and puts a block in the feedback loop (test, measure, evaluate… hang on … wait a minute…)
That sets the challenge for the tools to show far less information and make that information highly relevant to specific audiences so they can spot the insights relative to their particular function and take action.
Incidentally, I wonder if some of the high-end vendors are already approaching a 10/90 split between the base tool and consultancy services…
I heard you talk about this rule at E-Metrics and thought it was brilliant. Great to hear you discuss it more in depth here.
Do you think that $90 should also be applied to the people actually implementing the ideas? I seem to hear more that the insights are there, but there's no one to actually run the tests or make the changes because the rest of the staff (editors, IT, designers, etc.) are too busy doing their day to day jobs.
Excellent information and I agree with the problem of under-spending on analysts. I, too, heard you speak at Emetrics 2006 in Santa Barbara. Tools are important – without them the 90 would be useless, but all-to-often people do over-invest in the technology and not the people to make sense of it.
Even with a right balance between analytics and analysis you still have to take action, as other readers here have commented. Often, I see organizations get too ambitious with actionable insights and overwhelm the already overworked IT and creative staff.
Internal politics often present the greatest barriers to increased performance, not tools, analytics, or actionable insights.
My recommendation to the smart analyst is to bite off small pieces of the problem. Yes, it will take longer, but in the end you will get things accomplished. Start with benchmarks, make a report that can be easily shared throughout the organization. Host a meeting to talk about it, then pick one area to take action on that will show some real results and give the credit to the people who actually made the changes (the IT, developers, designers, etc…). That will get them excited, get management excited, and break down barriers to getting more accomplished. Do this and you'll soon win buy-in from others in the organization for a faster pace of change.
Well, I work for IBM and while I agree that people are more important than the web analytics package – IBM has millions of pages – and can't see Google Analytics cutting it.
As you might know, Coremetrics just bought IBM Surfaid and frankly, that was a good thing. While there's mountains of data we have to ream though the problem is more profound than that.
In large organizations the distribution of data repositories has to be done with great care. It's possible to create a comprehensive view of the organization OR…create hive view of the organization. For analytics – it's a nightmare trying to pull data as in the Hive view, there are dozens of data repositories many with operlapping data but different numbers and different url tagging rules.
I can not over state how much of a nightmare this is. It's not so much the amount of data that's collected, it's the inability of the web tools to properly catagorize the data.
So the 90/10 rule could work for a smaller organization but absolutely would not work at any large corporation.
Webmetricsguru: Thanks for your comment.
Perhaps I failed in clearly delivering the core message: Web Analytics and web interactions (between our customers and our businesses) are both inherently complex and to get maximum value, independent of tool, we need to invest at a massively higher level in brains. Dumping the in house tool for GA is a suggested solution to find money for brains. That’s it. You might be right about how much load GA can handle (though Brett might disagree).
Speaking of IBM, I might have a contrary view. Given the size and scale you describe I think the 10/90 rule applies even more to IBM than others. In fact it might be more like 5/95. You keep the tool you have need even more brains (nice bright shiny brains) to make any sense (recommend actionable insights) to your business leaders.
Trust me I don’t underestimate your challenge. Hopefully you can take a copy of my blog post to your boss and demand the raise you have always deserved. :)
Caleb Whitmore: Couple quick thoughts……
You are right about this but my observation is that true “actionable insights” are really hard and they are truly insights, and not something the “Top Exit Pages” reports shows then it will throttle work because of the delightful challenge the word of real Analytics is.
Amen!!! :) I would also add the entrenched “traditional web analytics” / reporting mindsets to politics as a massive challenge that should not be underestimated.
If anyone is reading this comment and you have not read Caleb’s comment above please scroll above and read them, pure gold there.
Chad Parizman: I appreciate your kind words about the emetrics presentation. Thank you.
Hmmm… this is an interesting one. In my mind I was simply reallocating the $$$$ that were going to a “tool” to be redirected to “brains”. Even exchange. My assumption was that the people who exist to “IT” the site or design or do creative are still there and now thanks to Web Insights actually have great work to do.
The point you bring up is excellent. If a company does not have supporting Processes and People then actionable insights will stay on the powerpoint slide or excel file and won’t add any value. I would do two things, if there are No supporting resources then fund that first, if there are Some supporting resources than that recommendation to put money into your pocket can go to do incremental add. Does this make sense?
Andrew Hood: We won’t hold it against you that you sell a analytics tool !!
On the topic of having someone to take action, please see my comment to Chad above.
The perspective on the analyst/’s is that it is not just someone who cranks out reports but someone who actually is plugged into the business and strategy and aware of multiple things to start driving real actionable bottom-line impacting value by herself/himself first. But what I have observed happens is that once people taste that proverbial “blood” they start wanting to do more themselves and there my recommendation is to have a Web Analytics tool that balances simplicity with encouraging analysis. ClickTracks is such a tool IMHO (see Disclaimers & Disclosures). There are others as well.
Summary: get a good gal/guy to show the massive value, then start giving the simple yet analytical tool to your core users that allows them to do analysis (not just reporting like most packages out there) and make decisions, your gal/guy moves up the food chain of analysis (for example migrating from simple analysis to complex multivariate or statistics). Would you agree with this? Too naïve?
Thanks Avinash – I've just added a new slide to my ever-growing PowerPoint stockpile:
Data: petabytes
Reports: terabytes
Excel files: gigabytes
PowerPoint files: megabytes
One business decision based on actual data: Priceless
That just made my day.
Smartest thing I've heard in a long time.
I absolutely agree. Analytics is always a stumbling block to web managers, as they believe reports will provide a "golden key" to understanding what people are doing on the site. What they fail to realize is that we have to extract data from a machine – a machine that does not "see" the website as we do.
The key is in asking questions to extract data – something only a human can do.
What are your thoughts on outsourcing the $90 part ?
ie, instead of going out looking for a web analyst, you look for a person with good knowledge of "people on the web looking for, or with an interest in, a particular product" This knowlwdge would be gained from SEO research, practical experience with similar websites, etc
FYI, going back a few years ….
“ The human skills needed to bring true value to an enterprise from Web analytics are scarce, but they are more important than the technology involved. “
Frank Buytendijk & Astrid Van Dorst, Gartner Group, April 2001
Mike: I was not aware of the Gartner quote, thanks for sharing.
I have to admit I am biased. I work for a large company and that clouds my point of view:
Initially it is quite ok to go out and outsource (get someone from your vendor, get a consultant) to bring the expertise to the table and help. But long term this is not a good strategy but the vendor professional services people or consultants will never be in a position to get your business and truly understand it. That will limit the value they can add.
Let me say that that is not becuase they are not competent, they are hugely. But only someone in the company truly plugged into the day to day and strategy and decisions can get the reality and then use the tools to ask the right kind of questions.
MI completely agree with the 10/90 rule. A high percentage of the people part will come from outsourcing still, but that's OK for the SMB market. GA is the product, but for those companies that don't trust Google to collect their data, or simply want to buy a lisenced product for comfort feeling, Google has to hurry bringing Urchin 6 (the licensed GA version) to market
Erik, Thanks for visiting my blog and sharing your comment.
My core recommendation is not so much that you should go out and buy Google but that you should spend a disproportionately high number on brains rather than the tool. GA is just a example I used, folks have many choices in the industry of tools that are quite cost efficient which would allow you to spend $90 on brains.
I found your your article very very interesting to read. I totally agree with you. I do SEO for my company and I refuse to spend the $$$ on tools that I can do the work without it. At the end of the day i am there to cut costs and not make up different ways of spending it.
:) cool article..
Avinash: Thanks for your article. I only recently discovered your site, which in turn led me to the other analytics sites. I have been crowing the importance of web analytics since I was the webmaster of one of the very first search engines on the Internet.
But damned if I could ever get anyone to listen. No one I worked for (contract/ salary) ever cared or even understood the value of web analytics. Except one person, who recently asked me to write an eBook on Google Analytics. Except that ROI Revolution's site (and free webinar) made me realize how much there is to know about that tool.
Even today, I skirt around the fact that one of my blogs is about web analytics. I dance around the topics because I'm worried that people will be scared off. It hinders my writing, and I end up throwing away article ideas.
The fact is, that Gartner Group quote is exactly right. Unfortunately, there are not many people who even understand the potential value, let alone are able to extract value.
I'm glad to now see you and all the other analytics bloggers.
at my company we're literally at 10/90. The staff of our two FTEs is paid for in one month of the vendor's costs. our workaround to this is creatively using designated superusers, wiki's and other forms of distributed responsibility to manage. it works ok and it forces the analytics closest to the individual business units. but the planning, support, training, pressures on the central pair are intense.
To me, you're talking more about a philosohpy, rather than a rule. It's important to have a good analyst(s), it's essential, but you can't just convert that into the 10/90 rule for magnificent web analytics success. Web analytics does not end at analysis: making the most out of analysis is the hardest part, maybe. Do you have resources to actually implement the insights from the analyst? Do the decision makers "believe" in those insights, and even if they do so, is it going to be the rationale behind their decisions, or there might be political issues in between?
Of course, you're likely to be much more successful in an organization that is willing to spend 90% of the resources into human rather than technological capital… but still it's just the beggining of the story… now the fun part begins :)
Thanks for your blog, by the way… I am a begginer myself in the world of web analytics blogging, writing modestly from Spain.
Dear Avinash,
First of all forgive me for my english, from Girona (Spain)
I am in the analytics business from 1999 first as user (webtrends, hitbox,…) after as a consultant in Spain (hbx, sitecatalyst,…) and now as a manufacturer (benchmarking online service http://www.netsuus.com).
In my opinion, I see analytics beeing used for everybody in the online team instead of concentrate the efforts in one specialist. I see analytics as a marketing technique used for every Internet profesional (design, advertising, sales, IT,…) as the email or the browser instead of going throught one person in the company. Then:
a) What do you think about this vision?
b) Do the 10/90 rule works with this vision?
Thanks in advance,
Jaume Clotet
Jaume : I agree with you. I call it Data Democracy, and I touch on it at the start of this post:
https://www.kaushik.net/avinash/2007/06/six-data-visualizations-that-rock.html
In order for any company to gain maximum benefit from the data, they will have to create a environment (through training and empowerment) were maximum number of people (decision makers) have access to the data and can use it.
For a small company having one expert works because there are fewer layers, but that starts to lose its attraction even for medium sized companies.
With regards to question two, yes the 10/90 works with data democracy as well (in fact it is almost mandatory with your vision). That's because you can't buy a million dollar tool and expect it to magically create a democracy. You'll have to hire some folks (in different teams) or expand job responsibilities to dedicate time to use data and have budgets for training and …. and … and … more things.
If I were to summarize: The goal is to empower as many people in the company to use data and make local decisions for themselves, you'll need a tool that allows that to happen, but you'll to invest in resources and skills that can understand the data and make decisions.
Hope this helps.
-Avinash.
PS: Your English is great, you should not worry about it. :)
Totally agree. I was going to employ a company to get me further up the search engines, but decided to go on a course to learn abit about it before doing so. I spent £300 on the course. The company who were promising page 1 rankings on Google wanted £600 per month with a guarantee of our website being on the first page within 3 months. Within 1 month I am on page 1 for certain keywords and for the others on page 5. Which I think is pretty impressive. If more companies take the time to understand the web and seo they will save money. I did.
This is a great rule. In the end of the day, it is all about the analysis and for that we need a pair or pairs of hands and thinking. It is also question on putting the hours in and really doing the analysis part not just taking a climpse on reports.
Simply could not agree more!
This is a great post. There is just too much information and possibilities for change that it actually blogs my mind.
This information is pointless however without the input of a professional that understands the web, website, marketing and Analytics in order to give the right information.
Spend the time on the people, that really makes sense.
PS, I read your book, what a great read, recommend it to anyone who wants to truly know about their website.
All to often you see companies demanding data, data, data on a constant basis, the problem arises with the conflicts it creates. Human nature will always throw substantial anomolies, even in the largest of datasets. Keep it simple. Tracking every last window licker through your website will not make you a millionaire next week as a result of any changes you make. If your site is full of nothing but window lickers, change your site!
We used to use Webtrends. $20k a year and almost a fulltime job just keeping it going. Now we're just jusing GA! Love it!
This post is really useful for us, a start-up online biz. We'll think of web analytics a great tool to convert better values to end-users.
Thanks.
Excellent view using analytics as a services as Just software cannot seal the deal… You need the acumen of a full person involved with business and market.
[…]
According to the Web Analytics 90/10 rule put forth by highly regarded web analytics strategist, educator and author Avinash Kaushik, 10% of web analytics expenditures ought to be spent on the technology. The remaining 90% are best spent on expert human analysts.
[…]
This is brilliant. We just fell in to the same place when thinking about how to manage our soon-to-be "complex problem to solve with signficant incoming analytics". This just validated what we were beginning to realize.
Excellent view using analytics as a services as Just software cannot seal the deal. You need the acumen of a full person involved with business and market.
We used to use Webtrends. $20k a year and almost a fulltime job just keeping it going. Now we're just jusing GA! Love it!
This is brilliant. We just fell in to the same place when thinking about how to manage our soon-to-be "complex problem to solve with signficant incoming analytics". This just validated what we were beginning to realize.
Hi Avinash,
What is the performance impact of having 2 parallel analytics solution? There will definitely be a hit due to multiple JS files. In your experience have you see anyone do it this way?
Raghu: The most common implementations of web analytics javascript tags are as close to the /body tag as possible.
In that case having two or ten tags won't have any impact on the customer experience as the page and all the content would have rendered before the analytics tags.
With regards to what other issues might arise, please see this post:
~ 10 Fundamental Web Analytics Truths: Embrace 'Em & Win Big
Refer to item #1: If you have more than one clickstream tool, you are going to fail.
-Avinash.
I'm totally agree. But, in real world, I'm still have difficulty in finding "the valuable insights".
Would you like to describe one by one insights that i can get in every metrics?
Mega: Sure. Here's some guidance that can jump-start your efforts.
If you don't know where you are going any road will take you there. Your first action should be to create your Web Analytics Measurement Model. It will deliver simplicity and focus.
Then go read this blog post:
~ Beginner's Guide To Web Data Analysis: Ten Steps To Love & Success
It teaches you exactly how to do analysis that helps you find valuable insights.
You can buy a copy of Web Analytics 2.0 if you want to go even deeper / get even better.
All the best!
Avinash.
Thanks for all the wonderful posts, Avinash!
The information is concise, insightful, and gives hope for those seeking an understanding of new media and it's uses.
-Kyle
This is a great rule.
In the end of the day, it is all about the analysis and for that we need a pair or pairs of hands and thinking. It is also question on putting the hours in and really doing the analysis part not just taking a climpse on reports.
Thanks so much for making this concept clear to me: "It’s the people, not the software, that determine social monitoring success."
Sometimes we tend to think the opposite, that it is the software that you use what will make you succeed.
I guess it all depends on the set up and what you are trying to get out of the tools you are using. Some people just like to see that their website has not broken and is leaking traffic etc due to a technical failure, while others need to seriously look at the consumer journey and learn more about their customers.
Both of these require different levels of investment on the tech and personnel side of analytics.
I also worked at IBM and there's one thing I have learned it's this. Outsource everything until it's no longer profitable. Then adapt and change to new circumstances.
Pulling from your article, it seems you think along the lines of "get the right person for the right job" and not only do profits increase but operational & marketing efficiency also increase.
I'm 100% in agreement.
Roles and positions within businesses are fundamentally evolving as we speak. Today's modern businesses willing to evolve are screaming out for jack-of-all digital marketers.
Someone that can do the job of 6 people. And, due to no cross functional lag time, the jack of all marketer get's things done in 10% the time.
Avinash – Would you alter this ratio for 2019 at all? Machines are smarter than they were in 2006. I still value people above all, but wonder if the ratio is more 80-20 or 70-30 today?
There are two developments we have to consider now, 13 years later.
There are more free tools to allow you to do almost all of what you need in digital analytics, except perhaps the 2% of the answers (which do hold a lot of potential). The secret to success is still, humans.
The other development shaking up our industry is Machine Learning. Again, the hardware you need is incredible affordable (or even cheaper in the Cloud) and the software you need to use is open source. The secret to success is humans.
If anything with time, the rule is more like 5/95 rather than 10/90. :)