Confused about Web Analytics?
How about Web Metrics?
Have you been successfully scared?
Rhetorical questions. Don't answer. :)
Let's do this one step at a time, first let's demystify web metrics. We'll do web analytics another time.
The hardest thing you'll do in your life as a Web Gal / Guy / Marketer / Analyst / Researcher / Jack is identify what constitutes success for you when it comes to measuring Outcomes for your website.
Step two will be to identify the Critical Few metrics to measure those successful company, and hopefully customer, outcomes.
So what makes a great Web Metric? And what are the factors you need to keep in mind to ensure that your valiant efforts to measure business success in this first life will be successful? :)
This post is inspired from my segment of the WAA's well received webinar on the newly defined Web Metrics Standards last month.
The intelligent Ms. Angie Brown and the fabulous Mr. Jason Burby spoke about the tough work that went into creating the new standards. I spoke about Web Metrics Demystified .
In this post you'll learn how to find diamonds in the rough, how to know that a metric you have identified for your Management Dashboard is actually a good one, and you'll learn the process you can, and should use, to keep your web analytics metrics relevant.
Enjoy…..
Four Attributes of Great Metrics:
Metrics are a dime a dozen. Especially on the web. There are books and blogs full of 'em.
How do you know which one is your must have darling?
In my humble experience here are four attributes that all great, nay magnificent, metrics possess…
1. Uncomplex
Great metrics almost always are uncomplex.
Because we did not make much headway with recommended metrics foisted on us, our response has been to create complex metrics. Six things each with its own unique multiplier / variable predicting the position of the sun when visitors click on your site.
Here is the thing to think about: Decisions in companies are not made by one person. If you want action then the democracy needs to understand performance, the democracy need to make decisions.
Not you. Certainly not your consultant. Or best friend.
If you are the only person who understands the metric, the Key Performance Indicator, then you have just guaranteed that your company, big or small, will not take action. Because you know the metric, but not the business.
Don't sexify, uncomplexify. Solve for the masses making decisions. It is not as easy as you think, try.
2. Relevant
Is the metric you have chosen relevant to your business?
Since we have so many metrics we pick our favorites and then stick with them. The problem is that each business is unique, even businesses that seem like they might be in the same business.
In Web Analytics: An hour A Day I use the example of Best Buy and Circuit City. You might think that they should / would / could measure their website with similar Web Metrics. Nothing could be farther from the truth.
The only thing they have in common is the fact that they sell large screen TV's on their website. Everything else is different. Their business models, their priorities, how each tends to use the web in its multi-channel portfolio.
The metrics you would use for each company to measure success would be different.
It is ok to seek inspiration from your friends and competitors. In the end truly stress test that the metrics you identify are relevant when it comes to measuring the success objectives that are unique to you and your website.
Remember what works for Jason might not work for Shane. And those two are close! :)
3. Timely
A few years back I was interviewing at one of the biggest companies on the web. They had just closed their quarter, it had been tremendously profitable. I asked them what the reasons were for that great success. The rest of this is absolutely positively 100% true.
Them: "We have just kicked off the query against our data warehouse, it typically returns the results in three months."
Me: Stunned silence.
Granted they were a big business. But still.
I learned a very important lesson on that day. Be on time or die.
That big company's stock price is a fraction of its price at that time. While not all of it is related to their ability to measure, you can imagine how hard it is to be successful in your business (on the web for Pete's sake!) if it takes you three months to know what worked three months ago.
Great metrics can be provided in a timely fashion so that your business decision makers can….. make timely decisions.
I am not a big fan of real-time (see this post: Is Real-Time Really Relevant? ). But between real time and three months there is a sweet spot. Find out what your sweet spot is and then ensure your data can be collected, analyzed, and metrics provided with insights in that sweet spot.
Even the greatest metric in the world is useless if takes nine days to get (with insights!) when your world changes every three days (key word bids, affiliate bonuses, email promotions, web page updates, whatever).
Be timely.
Sacrifice complexity and perfection for timeliness.
4. "Instantly Useful"
I absolutely love this one. Smooch, smooch, kiss, kiss.
I credit my early experience with ClickTracks for that love. Dr. Stephen Turner and John Marshall had eliminated all the non value added stuff from the application so that no matter where I went, what report I opened, it was instantly useful.
It was a combination of the fact that there were fewer metrics but also the fact that they were presented in a way that made it easy to understand performance and get the first blush of insights.
Instantly useful is when you understand quickly what the metric is, and you can find the first blush of insights as soon as you look at it.
Here is a great example, the What's Changed report in ClickTracks:
Anyone can tell you what your keywords were this month, or last month. The ClickTracks reports shows you "what you should care about", keywords that rose in their importance this month and which ones reduced in importance.
All the complexity is "hidden", there is no crap, just stuff you should care about. In front of you.
Now does that not simply kick butt? [Click on the image above for a higher resolution version.]
It will take some nice analysis and time to understand all the nuances and unlock the mysteries and deep stuff (just like say for example with your wife / girl friend, less so with your husband / boy friend!).
But the first blush is there. As soon as you look at it.
[I think Google Analytics V2 also does this well through use of layout and color and summaries or things like Compare to Site Average, Compare to Past etc etc. But I admit I am greedy. Every time I look at a report, current or new, I ask for more instant usefulness! Phil and the team humors me by making stuff even better.]
In a data democracy metrics have to meet the bar of being instantly useful. And not just that, think of your boss, her boss, his boss. How little they know. If send them a metric and it is not instantly useful then it will be instantly ignored.
You want instantly useful, no explanations required, because that will give you the opening you need to show your "deep stuff", explain the nuance, highlight your analysis!
Smooch, smooch? :)
Example of a "Great Web Metric":
Let me give you a very simple example that I think will crystallize the methodology above.
I think Bounce Rate is a "great metric". Here is how it passes the required four attributes test:
Uncomplex? Single Page View Visits. Easy to understand, explain and propagate. Enough said. :)
Relevant? It identifies where you are wasting marketing/sales dollars and which pages stink when it comes to delivering on the "scent". Those two things apply to most web businesses. Bam!
Timely? Bounce rate is now standard in pretty much every web analytics tool, and available in every report. Every day. Nice!
Instantly Useful? You can just look at it and know what needs attention, what needs to stop. You see 25 – 30% for your site and instantly you know things are fine. You look at a page with 50% bounce rate and you know it needs attention. You see a campaign / keyword with 70% bounce rate and you know there is a fire.
Set aside half hour today or tomorrow or at the end of the week and apply the four attributes test to your own important web metrics. What do you see?
Three Important Final Lessons:
Here are three lessons that are directly from the front lines (sourced from painful battle scars!)…..
1. Perfection is…. the enemy of good enough.
Data quality on the web is not perfect, things change too fast, everyone wants a piece of data yesterday, your competitors are strong. Don't spend time getting things perfect when it comes to your metrics.
If you have 90% confidence in the data (how it is collected, processed, and presented) then make a decision. Don't wait for perfection.
Too many times we spend too much time being distracted by missing tags and the hoopla around deleted cookies and more. Follow best practices, then move on. Go for precision and not accuracy.
As Mr. Stuart Gold says: An educated mistake is better than no action at all.
2. Critical few baby, critical few!
I owe Steve Bennett the CEO of Intuit all the credit for this important lesson.
Steve is fond of pushing everyone to identify their "critical few". Priorities. Goals. Metrics.
My interpretation of Critical Few: When the proverbial crap hits the fan what is most important.
That statement has a phenomenal clarifying power.
If your business was on the line how would you know things are going well or badly? Cutting through all the clutter of data, what are the metrics that are your Critical Few?
Almost all of us have too many things we measure, too many things that distract us, take away our precious time / attention.
You probably have at most three Critical Few metrics that define your existence. Do you know what they are?
If you have 12 then you have too many.
3. Metrics life cycle process is your friend!
Metrics no matter how great have to stand the test of time. And business changes. Repeatedly.
I recommend this simple Metrics Lifecycle Process…..
The idea is quite simple really: Use the four attributes test to identify your critical few metrics, go measure them, then analyze the data you collect, take action. Here's the fork on the road. If you can't action anything then perhaps it is the wrong metric for your business. Eliminate it. If you can take action figure out how you can improve it further.
Execute the Metrics Life cycle Process in a timely manner, I recommend atleast once a quarter.
Some metrics will stay, those are your best friends, others will outlast their value, give them a warm hug and say bye bye.
There. Web Metrics Demystified!
Not that hard, right? Just a dash of thought, a drop of common sense and a pinch of passion.
Ok now its your turn.
Please share your perspectives, critique, bouquets and brickbats via comments. Thank you.
[Like this post? For more posts like this please click here, if it might be of interest please check out my book: Web Analytics: An Hour A Day.]
Via 
hi avinash,
Its amazing how you come up with a GEM every time. This one too goes far beyond what traditional way of analysing web data used to be.
Bounce rate really justifies itself as a great web metric for any and everything on the web. It is directly proportional to customer engagement on the page and thats what makes it a comparatively better web metric
I am thinking of a few others at this point of time : Possibly, Average no. of pages visited & New v/s returning visitors come to my mind immediately and they seem to satisfy the 4 attributes mentioned above.
Thanks Avinash
Keep demystifying Web Analytics ! ! !
As always, a post worth reading. As someone who has been on the inside of very large corporations and is now looking in from the outside, I've found that the hardest part is implementing the Metrics Lifecycle Process.
Killing a metric is one of the hardest things to do, even when it's the right thing to do. Reasons why this happens:
* The metric was only developed at great effort and high expense (GEHE)
* Once the metric ceases being collected, it can only be restarted at GEHE
* The person up the food chain–who last really paid attention to that metric was when it was useful, might come looking for it if something goes awry, resulting in GEHE on your part to educate them as to why it's no longer relevant
Yeah, there's a lot of concern there about sunk costs, but that's a bigger problem.
For those of us working in small companies without resources, putting old metrics out to pasture isn't that hard. For those in larger organizations, the key to avoiding problems after eliminating a metric is to get buy in a cycle (maybe two) before eliminating the metric from all the stakeholders. That eliminates some future GEHE and lets you keep moving forward.
Avinash,
The one trick I see here is convincing those higher up in the food chain what a valuable metric is. The metric of "visits" doesn't hold up as a Great Web Metric because it's not always that relevant and really not instantly useful. It is used as a goal, however, because it's easier for some (execs) to understand.
We end up tracking both kinds of metrics – those that hold up to the test and tell us what to take action on, and those that are requested by the execs. The art seems to be gently introducing the new metrics into conversation so that we're all (eventually) on the same page.
BTW, nice observation on Shane and Jason! ;)
Avinash,
Great post. Identifying the "right" KPI's seems to be a struggle within most large corporations from what I can tell. Within mine, the two people with offices right next to each other don't agree on the most important action on our website. Next comes the opinions about relevant website content from the exec's on the other end of the building which is solely based upon how they handled print advertising 20 years ago. This leads to the HIPPO making a decision that makes absolutely no sense to anyone. Therefore, it's nice for me to come back and read your posts for a quick dose of sanity.
Great post Avinash — this is becoming a cliche:-). As I was reading through, I was thinking about corollaries to your points but then saw that you have covered it all in your final three lessons. To me that is what makes it fun – you always make a full circle with your discussions.
Have to agree that revisiting a metric periodically and recommending a change is one of the big challenges that I have faced. The general mindset is that once the 'right' metric is selected, it is set in stone for eternity.
On point#3, the 'Timely' aspect nicely ties in with your other post 'Accuracy vs Precision'. The ability to be precise and have a faster go-to-market cycle is immensely more worth than the ability to calculate a metric to the nth degree of accuracy.
On #4, in addition to analysis and time, you forgot to mention intuition (especially in judging moods) as one of the skills necessary to understand your wife / girlfriend :-).
Thanks for a good read.
-Ned
Avinash,
I love the blog.
Not related to this specific post, but I've been beating my head on a problem, and I realized you might have the answer.
For my e-commerce sites I can get the margin of error for my conversion rate at a given confidence level. With that data, I know if my A/B test results have diverged enough to give me a winner.
What's the margin of error formula for Per Visit Value or Average Order Size? Per Visit Value rather than conversion rate is actually my critical metric.
If you don't have it, any pointers?
Either way, kudos on the great blog.
Good stuff Avinash, as always.
I could not agree more about bounce rate just like Ankur wrote in his comment. I wrote a long post about it myself.
… you write so so well. i wish i could buy your ability.
Rishi
Hi Avinash,
My only concern here is of a semantic nature. Why use the term "Uncomplex" when "Simple" is, well, simpler? If the goal is to simplify the metrics, using language to match should be part of that process.
Many thanks for this and the many other fine posts.
Dave
[…] Follow the money trail. Every site has a money trail. And along the way, people perform certain actions, “success events.” These success events contribute to the bottom line of your company. What are these? Signing up for your newsletter, reviewing your product, buying a product/service, site visitor engagement measures etc. To read more on this step, head over to Avinash’s blog. He wrote a great post about metrics today. […]
Dave : Good point. I certainly was recommending simplifying the metrics, decision making democracy was rarely fostered by a metric by throwing in seven different dimensions and adding / multiplying / crushing them together.
But I was trying to stay away was Simple. There is a semantic implication there that was wanted to avoid. :)
But you certainly got the point and that makes me happy.
Aaron : I don't know of a handy source, I would recommend using data from your site, perhaps via a couple tools (GA / Omniture etc) or couple data sources (logs / tags) to compute it.
Given all the differences between sites and platforms I am a smidgen doubtful that there is a global margin of error number that will apply to everyone.
Mark : GEHE, what a wonderful acronym!
I can completely relate to your experience, I have seen that time and again. But I have learned that keeping the faith is important.
This goes with your last recommendation: Way upfront share the picture above (metrics life cycle process) which will set clear expectations that some metrics will die if they are proven not deliver actionable insights.
Everyone : I have four speeches in four cities this week, in four days. Monday Las Vegas, Tuesday Chicago, Wednesday New York, Thursday Miami! [And I still wrote a post – I am hooked, what can I say!! You keep encouraging me with your comments and conversation. ]
If there is a delay in moderating comments or replying to your email then I apologize.
Thanks,
-Avinash.
Hi Avinash,
A very well written post. Working for a IT Outsourcing company, we have this problem where clients go into an analysis-paralysis mode and asking us to implement metrics that are so hard to define that it defies logic!
I think I'll shown them this post and see how they react! :-)
Regards,
Akshay
If you're ever interested on expanding the details of the metrics lifecycle, I recommend looking at http://www.crisp-dm.org/ which for years has been a cross industry approach for data mining.
You may find it shares similar approaches to metrics and analysis combined with insight from many industries over the last two decades.
Avinash:
Nice post – we definitely need simple, meaningful, and well understood metrics for the web. We are working on a key metric of online engagement over here at and I will be sharing this post with our team.
Great work.
TO'B
Avinash, great stuff as always.
When I saw the ClickTracks screen shot, it reminded me of my wish that Google Analytics offered the same kind of keyword trend report. How much more productive it is to let the software illuminate the meaningful and actionable trends, as opposed to a static matrix that's simply a snapshot in time.
–dave
Avinash, as always, another great post!
What I've run into the most is when, in an effort to aim at "Relevant," the discussion stomps on "Uncomplexity" (or simplicity!) like a cockroach skittering hastily across the kitchen floor.
Rather than stopping at, "Given our business's strategy, is this metric relevant?" the question becomes, "Given that, at the end of the day, we are concerned with revenue and profitability, how can we ensure that all of our metrics track directly back to one of these two corporate KPIs." The result is metrics that either never see the light of day, or metrics that include so many leaps of logic in their calculations that nobody pays attention to them.
Based on several recent experiences, I blogged a highly tongue-in-cheek post on a "simple framework for establishing corporate metrics:" http://tinyurl.com/2eca7x. It's broader than web metrics, but most of your posts can be applied much more broadly! It dives a bit deeper into how NOT to search for relevancy in metrics.
Great post Avinash. It's always a pleasure reading you!
It is crucial to set metrics and polish them with your process. I'm wondering what should we do with goals? You can have the best metrics in the world but if your goals are not well define, there are useless. So, how do you set goal? you can use litteraure or historic data, but as you said, each webiste is unique and evolve, so what can we do? Should we look only on variation between two periods? What is a significant variation? is it 15% or 50% ?
The themes you have used in this post seems genereric and aplicable for many responsibilities; which I think is very good. Be simple, relevant and on time.
I'm going to archive this post for reading again. Thanks
Jean-Sebastien: I'll cheat in my answer to you. A little while back Robbin asked me a question about how to create goals for conversion rates for a company.
My summary for you:
See if you can find any external benchmarks that you can use to see how others are doing and use that as a critical input. If you can't find external benchmarks then use your own historical performance and trends there as a guide, just remember to add improvements expected due to investments in people / tools / processes.
Now here is the cheat answer in reply to Robbin's question about how to create goals specifically for conversion rate….
From Three Interviews, Six Interesting Web Analytics Questions.
Hope this helps!
-Avinash.
Hey Avinash,
Another awesome post….really useful insight and your book is fantastic,especially with your constant updates over here.
cheers,
deric
Hi Avinash,
Its a great post. Really covers broad criteria and relavant examples. thanks for information.
I am looking for a part 2 of this post that delves into looking at a specific tool say google analytics and explore and look into what that information tells you. Then another post about how exactly you derive your conclusions from the reports and go about fixing some typical common problems.
Am I asking for too much.
I guess so. Like you I am also greedy.
If you re not planning to post it then I ll digg for more information and see what I can post on my blog.
cheers
vineet nair
http://internetbusinessjunkie.blogspot.com
Excellent posting Avinash. It's always surprising how much effort and information you are able to put into your postings.
I wanted to ask you what do you forecast for the web analytics/metrics world for 2008? more of the same or something exciting on the horizon?
Do you think that Game theory can help to understand visitor bahavior?
Best Regards and keep it up….
PeterPhil : I have tried to stay away from predictions this year – most I have seen are too obvious or seem too far fetched (hence might come true!!!). But since you ask here is a quick stab…..My overall feeling is that the segment itself will continue to explode. The more I meet people, at companies big and small, the more I realize how much more data needs to be used. The good thing is that people are finally starting to see its importance. So great for Vendors, Practitioners and Consultants. It has been hard to make money on Web Analytics, this will be less of a issue for the next couple of years (for all three groups).
The other thing I see accelerating is the concept of Web Analytics 2.0 (http://www.webanalytics20.com). The essence of it is that we are going to move more aggressively beyond simply clickstream data (which is like 99% of what web analytics is today). We will realize how the other components of that model proposed are important and finally data can start to drive action rather than languishing.
The problem with Web Analytics 1.0 is that it is a exercise in data torture and reporting with long lags in taking action (if any). Data torture needs to get automated and expanded, decision making needs to get automated (think MVT, BT etc), people need to be left for smart hard things (vs what happens today!).
So two things for you to think about. Two trends that will accelerate in 2008.
On the second point of your comment, it is interesting to think of applying Game Theory to understand visitor behavior. I think the core problem is that when you say Visitor you are probably thinking of a Person, the reality is that in web analytics Visitors are "cookie_id's". When you come to grips with the difference you realize that there is a lot less to learn about "Visitors" then might be apparent on the surface!
Here is a post that might be interesting in that context (though predictive analytics does not exactly have much in common with game theory):
https://www.kaushik.net/avinash/2007/09/data-mining-and-predictive-analytics-on-web-data-works-nyet.html
Thanks again,
Avinash.
Thanks Avinash for the extra info.
What I find utterly interesting, yet complicated to contemplate at the moment is how companies will progress in blurring the divide between offline and online data. I also read your post on non-line marketing (courtesy of David Hughes). There seems to be a big push in the advertising and marketing worlds to make this divide seamless, pooling these vasts amount of data and technologies together so that the distinction becomes even more difficult. The potential seems fantastic, yet where I struggle to make ends meet is in the reconcillation. This is why I thought your posting on predictive analytics was quite interesting.
As for 2008, I've only just made a New Years Resolution (yes a little late I know) and that is to learn all I can about this sector, I've been a generalist for quite some time, now I've found a subject that really stimulates the grey matter and presents constant challenges. To help this I've decided to buy myself a late Christmas present in the form of your book.
Thanks again
Phil (not Peter)
perennial thinker at http://think-train.com
I'd just like to pick up on one of Phil (not Peter)'s points about reconciliation between on and off-line customer journeys.
Firstly, I love a word coined back in 2003 by Jack Aaronson (http://www.clickz.com/showPage.html?page=3622794) which is "Channibalism" :
"A basic fear felt by many executives and managers that one channel, or line of business, will steal business away from another. There are two facets to this fear: one tied to fears of consumer confusion and loyalty and one tied to managerial salary structures."
If we take that last point about salary structure as painfully valid then we have reached the core of the multi-channel tracking problem: If I'm an on-line marketing manager in silo-marketing-cover-your-own-butt-Inc why would I want to drive a prospect to a store or call center if I don't get my commission.
I was speaking to a colleague who recently joined an organisation that only made 5% of its sales on-line, despite being a really efficient and capable site with a target audience of IT professional. The reason – nobody got commission for on-line sales, but the personal rewards of driving people to the call center or the key account team were high.
So is this where we really stand with integrated marketing, that the company conspires to do what it best for the individual stakeholders?
Interestingly, I have been doing some work with a few organisation (mail order, travel, financial services) where they find the conversion rates AND average order values are higher if they push people to a call center at specific stages in the process. Why? Maybe people are better at persuading prospects to upgrade to "champagne and flowers on arrival" than a radio button.
So there we have it…we can dynamically create unique phone numbers for "generic" versus "brand" PPC ad's so we can track off-line conversion, and we can bribe people to type in vanity url's from magazine ads to get an extra 5% discount, but if we're not getting a bigger piece of cheese at the end of the year why don't we just push 'em through the web funnel and life will be good.
In Web Analytics Avinash makes the point that we can trade value for insight to track off-line outcomes, but maybe nobody does if it affects their bonus?
Ho hum.
It is kind off-topic but maybe you have time to answer – Google Analytics
I am trying to track conversions between two websites. One site (A) sends referrals (content site) to another site (B – e-commerce platform) which processes the actual transaction.I tried to setup up google analytics gooals (site A) to track conversions between sites A and B but it seems (maybe I am wrong)as though analytics can only track conversions on one domain.
So, I have a question and I hope you'll be able to help me. I've been searching through your blog, through Google Analytics help, have asked twitter, have searched on Google and I can't seem to find anything on this.
I'm looking to figure out what metrics would be good metrics to track for a website that is primarily concerned with people clicking on their AdSense ads and leaving the site. Traditional metrics I would track for content-oriented sites don't really fit the bill here but I can't figure out anything beyond Revenue Per Visit and Revenue Per Visitor. There's no way this site is able to track when people click on ads, as far as I know, through GA. Your blog obviously focuses on e-commerce sites or content sites but nothing mentions AdSense-driven sites.
Thanks!
Emma: First I wanted to mention that the below metrics will be available for you in GA soon:
It is in Private Closed Beta right now. But fingers crossed it will be in public beta soon and then you can use it.
It gives you a robust set of metrics.
But other than above you would want to ask yourself the same questions: What's your outcome, what are your goals.
I am assuming you want lots of people to come to your site, you want them to come repeatedly (so that), and have them click on your adsense links.
In that case I would also track, over time, the main traffic sources (search, direct, referring sources) and keep track of the top ones over time. I would also focus on metrics like Visitor Loyalty (to know if the same people visit my website again and again, this is a good thing) and I will also track what other "micro conversions" people are doing on my site (if they are not clicking on ads!). See this post:
Excellent Analytics Tip #13: Measure Macro AND Micro Conversions
Hope this helps a little bit.
-Avinash.
I am taking the Masters Web Analytics course on Market Motive and this blog post was a required reading. I am thankful for the information.
The BEST way I show my clients reports are by showing ranking reports for their main keywords. My clients aren't computer geeks. They are busy running their businesses and need something simple.
I personally use SEOmoz Keyword Ranking Reports (PDF) and show my clients where they rank and where their competitors rank.
I agree that web metrics don't have to be complicated. I much prefer a simple approach myself. The complex stuff has a way of confusing clients and making them feel unsure of what they are really getting.
Consider " actionable" as a qualifier. " Can I take action against the data being reported?"
If the data you report does not change/confirm current resource investment then is likely adds little value.
Dear Avinash,
No doubt your blog article is good. But isn't it not relevant for a website which has 100s' of products that would require multiple web metrics to achieve at least a "good enough" result.
You are right in saying don't make your web metrics be too complex –
"Too many cooks spoil the broth" :-) .
I feel because there are no enough hands to work on huge data (travel industry can be one example) or lack of configuring or personalization of any analytics tool. I don't blame web analytics company.
I wish; just the bounce rate could be one answer to all the problems of metrics but you know that's not the case.
Greetings of the season!
Ryan: Here is how I've framed this for my Web Analytics Master Certification students at http://www.marketmotive.com :
Every business, no matter how complex, can only have a few KPIs it is solving for. Maybe five or six. No more. As you dig deeper (custom reporting, advanced segments) to identify why performance of the critical few KPIs are performing well or badly, you'll use different diagnostic metrics. These metrics are subservient to the KPIs and only used for the purpose of the prior statement. That keeps reporting focused and efficient (rather than data pukey as people who believe in having tons of metrics).
And any student that picks Bounce Rate as a KPI will fail the course – it is a decent diagnostic metric, it will Never be a KPI.
Net, net: Figure out the critical few KPIs (no matter how many products you sell), focus, in your analysis use metrics that have a direct line of sight to those KPIs, win.
Thanks so much for raising this important issue.
-Avinash.