Avinash on the definition of Enterprise Class analytics

Avinash Kaushik recently wrote an interesting response to my original post and request for comment from Google. I asked Avinash or Brett Crosby if they wanted to comment on the “Enterprise” nature of Google Analytics in one of the comments. Avinash didn’t want to answer that question, but he did offer some interesting thoughts on the definition of “Enterprise Class”.

In his post, Avinash says:

“Definition of an Enterprise Class vendor :

1) The Vendor has been around for more than 18 months, the longer the duration the better but beyond 18 months you the client decide what you are comfortable with.

2) The Vendor can scale its ASP infrastructure (or in house software solution) to A] capture the number of page views required by the client and B] process that data on a nightly basis.

{I am biased, I believe that most people don’t need real-time data even if they think they do. Here’s a check-list to figure out if you really need real time data: Is Real-Time Really Relevant?}

3) They have a support infrastructure to assist the client in need at reasonable price. If you are willing to pay for support, you should have to pay a reasonable price and expect solid support from the Vendor or their Partners.

That’s it. Nothing else matters. You need to know it has been around and that it’ll be there. No other golden rules.”

While I like Avinash a lot, I think this definition is way off the mark for what most companies are looking for when they say “We need an Enterprise-class analytics solution.” I think what Avinash has described is the core qualities of every analytics vendor in the market today (at least when he includes “or their Partners” in item #3 which picks up Google Analytics and a few others.)

The problem with this definition is that it does nothing to differentiate hundreds of vendors from one another. Kind of like a vendor constellation where every company is stacked on top of one another … So I propose that Avinash’s definition of “Enterprise Class” is incorrect, and instead point to the definition of “Enterprise software” found in the Wikipedia. Some relevant excerpts:

Enterprise software is software that solves an enterprise problem (rather than a departmental problem …
Enterprise software is often available as a suite of programs that have attached development tools to modify the common programs for the specific enterprise.
Mostly [enterprise software] development tools are complex programming tools that require specialist capabilities.

I personally can only think of one currently available web analytics solutions that would qualify as “Enterprise software” using the Wikipedia definition. Remember: Marketing is a department, not the entire business! This definition implies that the solution can be used throughout the organization to solve a variety of problems (for example, online and offline data analysis using the same suite of tools.)

Most interesting to this conversation are these criticisms of the use of the term enterprise (again from Wikipedia):

Often the term is used to mean virtually anything, by virtue of it having become the latest corporate-speak buzzword.
Some enterprise software vendors using the latter definition develop highly complex products that are often overkill for smaller organizations, and the application of these can be a very frustrating task.
Sometimes “enterprise” might be used sarcastically to mean overly complex software.

Ah ha, now the truth comes out! Perhaps the use of the term “enterprise” in web analytics is just a marketing ploy, designed to sound good but not really say anything at all about the sophistication of the solution.

Hmm …

Avinash goes on to offer a “not so humble rant” (his words):

“The reason most of corporate America is saddled with billions of dollars of sub optimal software is that companies judge tools/vendors on this vague quality called “enterprise class”, while completely ignoring what they actually need.”

While not necessarily untrue, Avinash incorrectly assumes that most companies know what they actually need. If most companies were able to “look deeply within themselves and figure out exactly what they need and then go get it” (his words) there would be no need for the vendor constellations that Forrester, Gartner, JupiterResearch and others produce every year. But these vendor assessments are reportedly among the most valuable of all analyst documents; for good or ill, they simplify the problem of differentiating hundreds of vendors who largely all say they do exactly the same thing.

So maybe, just maybe, corporate America has been fooled into buying solutions that are good but aren’t really “Enterprise class software” after all. But maybe they were only doing what they thought was best? As someone commented in Avinash’s post: “Nobody was ever fired for buying Enterprise class software …”

And maybe when the analyst firms call something “Enterprise software”, at least in the web analytics market, they’re simply looking for something that will let them write about a reasonable number of companies, not the 100+ vendors that Sebastian documents at web analytics book? Having written a constellation in the past, I can attest to the complexity involved in covering even a dozen vendors, much less 100!

Perhaps the best possible outcome from this conversation would be that all of the vendors stop calling themselves “Enterprise-class” altogether and instead work to differentiate themselves along other, more substantial lines? I can think of dozens of other good points of differentiation … cost, support strategy, approach to data integration, etc.

In the end Avinash comes closest to the truth when he reminds us that:

“Smart people with crappy tools can move mountains, without smart people even the most expensive and expansive tools can’t help a company move beyond measuring Visits.”

Which I would modify to say:

“Smart people leveraging good business process, despite crappy tools, can move mountains, without smart people and good process, even the most expensive and expansive tools can’t help a company move beyond measuring visits.”

Anyway, with all this in mind I suppose I don’t care if Google believes that GA is “high-end” or “Enterprise class” or whatever. I’m just happy that they’ve raised the bar on the online visualization interface and are continuing to drive interest in web analytics in general. Again, thanks Google (and thanks Avinash for picking up the conversation!)

As always, I welcome your comments and criticism.

Posted Thursday, May 31st, 2007 | 17 responses | Add a Comment | Share, Save or Email

Jacques Warren

To me “Enterprise class” would be an application with the potential of transforming the organization, because it touches all or most of its parts (here I agree with you). Such application could very well be an agent of change. I don’t see GA belonging to that category. Add to this that “enterprise class” in my mind is having full control over the whole process involving that application. Just try to have a look at your logs with GA…

Judah Phillips at Web Analytics Demystified » Blog Archive » Judah on Eric and Avinash on Enterprise Web Analytics…

[...] Eric, which if you’ve found me, you’ve probably already read. [...]

Marianina Chaplin

Hi Eric,
Good post. By intro, I started up a web analytics division for Base One in London and for clients such as Saab, London Stock Exchange and Powwownow. I know England nearly always lags behind the states on the web front – and there are some companies with superb in-house analytics teams and analytical frameworks – but on the whole that is not the case at all. I am persuading/training well known clients from scratch on the benefits of web analytics in driving their business etc. So a super duper “enterprise (for want of a better word)” solution that links to their oracle database etc, would be wasted on them (until the skilled people and analytical framework are in place). Which I guess was Avinash’s point. And for most companies’ needs, an ASP model (GA/indextools etc) will more than suffice IMVHO.
Marianina :)

eric

Marianina: Nice to meet you and thanks for the note about web analytics in Europe. You might want to have a look at the presentation I just published in my White Papers and Presentations section of this site that speaks to the differences between U.S. and European web analytics usage.

Regarding your point, this is certainly the debate, huh? I don’t doubt that many companies will benefit from GA and IndexTools and Clicktracks and the like, but one thing getting lost in this debate is that part of what the “enterprise” vendors provide that GA and IndexTools don’t is SUPPORT, and sometimes support is exactly what companies need to learn WA.

IMVHO “support” is not a Flash movie, a PDF, a Google group, or an occasional webinar — “support” is a person you can call, who will call you back, who will answer your questions, etc.

I mean, that’s how most of us learned web analytics. I suspect that even Guru Avinash had someone to answer his questions ;-)

Oh, I’m not saying that IndexTools and Clicktracks don’t provide support, but my understanding is that the support is primarily implementation support. And yes, I know GA provides support to very large companies … if I’m wrong about this, please correct me!

Regarding the partner-based approach so common in Europe and that adopted by Google: consider how few GAAC partners there are in Europe today — I don’t know the exact number but I suspect it’s not high. If GA really does have a million users around the world, how could a handful of companies possibly provide “support”? Weigh this against the relative rarity and potentially high cost of using external agencies for training and business-level analytics support (not high cost, but high compared to the $ZERO cost of GA.)

Please, please, please someone don’t write in and say that GA is “so easy to use companies don’t need support.” It’s not. One of Google’s sales people said that on a panel I was recently on and I was shocked … all of the data and anecdotal information suggests that the vast majority of people find these applications complex and difficult to use.

So maybe inexpensive tools are the right approach, but the major failing of my friend’s 10/90 rule is the rarity of people able to fill that “90″ slot. Ultimately the tools don’t matter, it’s the process through which the tools are used. Learning is a process, hiring is a process, reporting is a process, etc.

It’s great to hear you’re out there fighting the good fight. But there needs to be more of you, huh?

Thanks so much for writing and I look forward to your response!

Bhupendra

I dont think I agree with you guys fully, both u and Avinash.

I agree is the need of smart people. I have seen many smart people move mountains. Actually, the main thing is information is knowledge not the goal. We need to use the information. I have been using Google Analytics and I am very impressed with the type and quality of information it gives for free. Regarding calling it an Enterprise solution is a different fight.

I would call an software enterprise solution is it has the following qualities:

1. provides all information regarding web usage
2. have information usage direction for making tactical and strategic changes for betterment
3. good presales and postsales support
4. have considerably good number of clients lasting atleast 12 months

GA is quite weak in 2 and I have no information about 4.

The main thing I would like to critisize you is, you have been very critical of the use of the word “Enterprise”. If you really have some reservations against some companies or product then be particular and attack. You can’t generalise and attack one and all. I am actually very confused about what you want to deliver to the readers through this post.

If you want to convey that use of enterprise is not justified in web analytics, then I would counter you by saying “Enterprise” is just a word to define quality in terms of length and breadth of the service that software provides. You might have taken it in wrong way. When SAP says its ERP product as Enterprise then it guarantees all features added w.r.t. its domain. I cant ask them to provide Decision Management and Analytics solution within their solution nor you can ask them Web Analytics. It is an Enterprise Solution for manufacturing companies which solves all core operations. The high tech areas are left for specific domain experts.

I am sure I didnt understand what issue you are actually rising here. If I went wrong, please correct me.

Bhupendra

eric

Brupendra: I wouldn’t say my post is specifically an attack on anyone, certainly not the vendors, but more a call to recognize that the words “Enterprise Software” actually mean something to large companies and that this definition has been cleverly co-opted to make some very good software products appear less appropriate in many situations than they actually may be.

Not to be critical, but I do not believe that the four point definition you provide above has anything to do with “Enterprise” softare. Your definition, like Avinash’s, speaks to the quality of the company, not the depth or breadth of the application’s ability throughout the organization.

“Enterprise” is just a word, you’re right, but as I said in my post, it is one that provides little substantial value to the overall conversation. So perhaps it is time to stop using the word?

I’m sure many people would love to argue with your statement “GA is quite weak in [having] information usage direction for making tactical and strategic changes for betterment” but I firmly believe that it’s not so much the tool you use as how you use it.

Thanks for writing!

Marko Muellner

Eric,

Interesting post. While I agree with your main assertion that to be “Enterprise Software” you must serve more than one (or two) departments in the enterprise, I think you missed diving deeper into the marketing value of “Enterprise Class” positioning. Regardless of whether or not the software is acutally Enterprise Class today, or will be tomorrow, there are key benefits to positioning the company as Enterprise Class. Of the hundreds of analytics solutions, maybe 5 maintain large enterprise customers. By definition, are these not Enterprise Class solutions? Maybe I’m splitting hairs on Enterprise Software vs. Enterprise Class Software, but the word and the positioning seem to be both true and powerful. As always, thanks again for challenging us.

eric

Marko: Trust me, I know ** why ** the top vendors position themselves as “Enterprise Class” … I’m just saying that the basis for such an assertion is dubious at times. As Avinash points out, some pretty small companies serve some pretty big ones in this sector, which begs the question “Can a 10 person shop, or a free tool, or a tool that doesn’t provide customer support, truly be “Enterprise Class”?”

As an example: Is Google Analytics “Enterprise Class” or not? On one hand they largely only provide support through partners, have no APIs, no non-web reporting infrastructure, no ability to import metadata or lookup tables, no datawarehouse functionality, or any advanced visitor segmentation capabilities.

On the other hand, I can see them on 5% of the Fortune 1,000 sites (Chevron, GE, McKesson, Dell, and DuPont).

If we didn’t have to debate whether they were “Enterprise Class” or not, we could simply say, “Wow, Google Analytics is on 5% of the Fortune 1,000, 10% of the Top 100 domains, and 9% of the Top 1,000 domains (both by Unique Visitors) … how cool is that?”

Yeah, we’re splitting hairs to be sure, but I would much rather see an assessment of web analytics vendors derived from a clear set of criteria deemed critical to businesses at different stages in their analytical evolution than an arbitrary list of “Enterprise Class” vendors.

I know that Megan Burns is writing the Forrester Wave for web analytics vendors right now. Hopefully Megan/FORR will weigh in with something similar to the assessment I’ve described.

Thank you for challenging me!

Marianina Chaplin

Hi Eric,
Good to meet you too:) Thanks for your response and thoughts. Having looked at your Europe presentation (thanks for pointing it out – interesting), there is another segment to the analytics pie – the difference in analytics development in B2C versus B2B. I’m not saying large brand name retailers over here in the UK all have good analytics solutions in place, but many of them do. Whereas, B2B/professional services (which are where my clients are all coming from, with the exception of Saab) still have some way to go:)
I am not a cheerleader for GA – no way. And in fact I know that using third party shopping carts via google analytics for multi-domain sites – are really so so not scalable. Was working with the programmers today and because we have drop-menu choice built in javascript taking you from one domain to another, to add the utm links to track each time you link from one domain to another etc etc. And the written documentation/support is far less than we would like:) But for plenty of clients site (primarily ecommerce/one domain), GA really does give them really good insights into the online marketing performance which can only be a good thing.

I do see this as a battle – because I am still persuading clients to move away from “show me the stats!!” – but it is happening and I’m not going to stop trying. And along the way, I have been and will continue to keep adapting and using the right solution for that specific client’s needs – be that Netinsight/VS/Clicktracks/GA or whatever that might be.

May the challenge continue!
Marianina

Marianina Chaplin

whoops I mean’t to say non-ecommerce/one domains sites, GA great for.

Bhupendra

I could finally get into the root of this discussion and I really appreciate Eric for starting this great discussion.

This time I have written a blog entry on this subject and I would like to link this discussion to my blog.

http://analyticsbhups.blogspot.com/2007/06/on-enterprise-web-analytics-tools.html

I could actually find much relevance on the points raised by all three gurus of Web analytics (Eric, Judah and Avinash), and I ended up taking a mid-way. Let us take the discussion forward and give a meaningful conclusion that paves way for better and enlarged Web Analytics area.

Cheers,
Bhupendra

eric

Marianina: I hear you there. B2B often gets under-examined which is a shame since it’s hugely interesting. One of the reasons I wrote all those posts about measuring visitor engagement was because it is so valuable in the B2B model. If you haven’t read them, have a look starting at:

http://blog.webanalyticsdemystified.com/weblog/category/engagement/page/3/

Bhupendra: Thanks again for your comments and for the blog post!

wanalytics » Blog Archive » Web Analytics Enterprise Lösung in München gesichtet

[...] Zuletzt wurde vehemt bestritten, dass Google Analytics eine Enterprise Lösung sei, was allerdings auch niemand behauptet hatte. Verschiedene Definitionen wurden zu Rate gezogen, und es wurde über die Zahl der tatsächlich existierenden Web Analytics Enterprise Lösungen spekuliert.Eine sehr robuste und überlebensfähige Spezies dieser Gattung konnte heute in München besichtigt werden. Meiner Meinung nach die einzige dieser Spezies, die derzeit in freier Wildbahn zu bewundern ist. Ex-Visual Sciences, Ehemals WebSideStory nun wieder Visual Sciences CEO Jim MacIntyre war persönlich aus den Staaten angereist, um die Lösung, nach der kürzlich das Gesamtunternehmen benannt worden war, in Europa aus der Taufe zu heben und gleichzeitig eine für HBX Kunden erfreuliche Mitteilung zu machen: Über HBX Visual Workstation können auch sie vom Quantensprung der Analyselösung profitieren.. [...]

wanalytics » Blog Archive » Web Analytics Enterprise Lösung in München gesichtet

[...] Zuletzt wurde vehemt bestritten, dass Google Analytics eine Enterprise Lösung sei, was allerdings auch niemand behauptet hatte. Verschiedene Definitionen wurden zu Rate gezogen, und es wurde über die Zahl der tatsächlich existierenden Web Analytics Enterprise Lösungen spekuliert.Eine sehr robuste und überlebensfähige Spezies dieser Gattung konnte gestern zum Visual Sciences Launch Summit 2007 in München besichtigt werden. Meiner Meinung nach die einzige dieser Spezies, die derzeit in freier Wildbahn zu bewundern ist. Ex-Visual Sciences, Ehemals WebSideStory nun wieder Visual Sciences CEO Jim MacIntyre war persönlich aus den Staaten angereist, um die Lösung, nach der kürzlich das Gesamtunternehmen benannt worden war, in Europa aus der Taufe zu heben und gleichzeitig eine für HBX Kunden erfreuliche Mitteilung zu machen: Über HBX Visual Workstation können auch sie vom Quantensprung der Analyselösung profitieren.. [...]

Steve Jackson

I read this late as I’ve been on holiday for a while however this post resonates with me because of the biggest client I currently work with.

The client I mean is truly global – I won’t mention the name but it is one of the biggest brands in the world, constantly in the top 6 globally. You have internal politics, you have hundreds of individual contries and their local vendors to work with, you have multiple client contacts to keep happy, multiple content delivery systems to work with, multiple campaigns to organise reporting for and many, many, many challenges and hurdles to face with data integrity and data presentation. This is before you can even start analysis.

You know what I use primarily to compile the data and put it into a presentable form? Excel and Powerpoint.

Yes, the only true enterprise solution used by everyone in my clients organisation is MS Office.

I have multiple data streams to cope with. Google analytics, HBX, Instadia (now Omniture) Google Adwords, MSN, Yahoo, Dart,
Various (and I mean hundreds) of individual banner campaigns and other offline data to cope with.

Currently there are two I would say enterprise analytics tools with powerful web analytics built in. SAS and Visual Sciences.

Actually getting my clients corporate wide adoption of either of these systems would in the current environment take years.

All of the other tools I have seen (now that Omniture has effectively removed the potential challenge to VS and SAS by acquiring and destroying the Instadia system – though even they used excel as an output source for custom reports) have flaws in compiling all data sources and making the reports completely customizable. Therefore I am still (9 years into my web analytics experience) using excel and powerpoint to do enterprise level reporting and analysis.

Nearly all of the web analytic tools provide some level of intergration with excel. This to me speaks volumes about “Enterprise class analytics”.

Unless you have a corporate wide data collection/compilation and processing solution which allows you to completely customize reports *AND* you have a corporate wide process which means the whole enterprise uses the platform, then you don’t have an enterprise class data analytics solution.

SAS and Visual sciences are two that can do this, though the level of training and support required for both systems would mean they would be wasted on most of my clients and the one that could use a tool like this would need a massive policy shift.

So I’d largely agree with Eric that the term Enterprise needs removing from nearly all of the web analytics tools currently available.

I also agree with Marianina that most companies (even enterprises) are going to need to take things in smaller bites to begin with, until that is they have the understanding of what they can get from their euros/dollars.

This is why web analytics tools are successful at the moment and why most vendors are moving toward the ASP integration model so that at least data can be easier to compile.

I don’t expect the landscape to change that much in the next couple of years either, though it is hard to say what might happen in the mid/long term.

eric

Steve: Welcome back! Without asking about your comment that Omniture is “destroying the Instadia system” I wonder if you have any experience with Unica? Obviously I was considering Visual Sciences when I referenced “one currently available web analytics solutions that would qualify as “Enterprise software” using the Wikipedia definition” but a friend made me consider that Unica may be the second such system (I don’t know much about SAS but I don’t consider SAS a “web analytics” system as much a true Enterprise-class analytics platform — using all the terms correctly!)

I also am very interested in your comments about “taking years to implement” and requiring “a massive policy shift” since at least one vendor in the WA marketplace has pinned their future to being “Enterprise” class and being deployed broadly throughout the organization across multiple channels. Do you have any feelings about the likelihood of short- and long-term issues this vendor may face (Hint: They’re your vendor!)

Anyway, thanks for the (belated) comments and I very much look forward to hanging out with you in Helsinki and Stockholm in September!

Steve Jackson

Hi Eric; I’ll address the Omniture point first (and this could be a post in itself). Instadia had a very useful system, one I considered a real contender to be the No.1 web analytics system in Europe because of the system structure and data collection method it had.

You could use it in a very similar way to Visual Sciences. Practically all the work with Instadia was done on the vendor side backend (filtering, report set-up, segmentation etc). Once your (very simple) tags were implemented on the site, you pretty much had it all. All you needed to do was then use the Instadia ASP system to find and get what you wanted, you didn’t need to go back and add tags to pages if you wanted to do something new. It is/was a web analytics system designed as web analytics systems should all have been designed from the first time someone thought of using Javascript and an ASP server.

Now when Omniture bought Instadia I was immediately worried because my first reaction was that they will absorb the system so that everyone moves to Site Catalyst. While Instadia is still available to existing clients for the next year the idea is that everyone will move over to SC eventually.

When I questioned Omniture directly the answer was that the technology is to be discontinued. This is a real shame. Omniture’s technology is simply not as good as Instadia’s was from the respect that much of the data collection relies on the Omniture tags. It means that tagging for specialised and custom reporting becomes an issue. While this is not that important for smaller companies, often, when you’re talking about large enterprises this is a huge issue and a big cost.

In my opinion, if Omniture had any sense they would develop their next release based on Instadia’s data collection method and backend, not discontinue the system. They could then use Site Catalyst as a superior front end for Instadia but still allow people like us to develop very powerful filtering and segmentation into the SC suite. You’re very welcome to pass this onto anyone you know at Omniture to get their comments as I’m sure they will defend their stance, however they will probably take the view that the SC system is currently far more advanced than Instadia.

It is much stronger from the eye candy point of view and the integration with other systems SC has is impressive, BUT, this integration would be even more useful with Instadia’s backend and data collection method.

I think they have invested too much into SC to change it and their policy therefore is to continue with what is not an ideal way to collect data.

I won’t mention names but when I raised this with an ex-Instadia guy now working for Omniture, he was equally as frustrated as I was.

Omniture purely and simply have bought Instadia for the market share and the very good people they have gotten as a result of the acquisition in Europe.

I am not criticising Omniture for the acquisition, i think it was the smartest move they could’ve made. I do however feel they are wasting the potential of the Instadia tool.

One of the big problems with web analytics tools today is the amount of deployment work required. It sometimes takes months. With Instadia it typically took a day to tag everything.

We did a test to get some very advanced KPI’s reported in both Instadia and another system (lets call it system b) with a similar data collection method to Omniture. Instadia far outperformed it. It took 6 weeks just to implement the KPI’s on system b. It took 1 day with Instadia to track the entire site correctly and get the custom KPI’s.

Anyway rant over! ;o)

Regards Unica. I have not enough experience to comment. I’ve never seen or heard them at any trade events which I would expect to see a web analytics vendor therefore have not investigated them too deeply though they do look like a contender due to their support structure.

Regards my taking years to implement comment. That is misleading. What I really meant is it might take years to get a policy change that would make such a system effective. Implentation wouldn’t take long at all, but getting people to use it multi channel and effectively would.

I tried for 9 months with one company to get them to move to a system I thought perfect for their needs and had backing with many of my clients in that company but still face too many internal barriers. I will discuss that more with you in Helsinki/Stockholm (or offline) which I am also looking forward too.

Add a Comment