Stickiness of SaaS/On Demand

Maybe it is just me, but I’ve never understood the idea that SaaS/On Demand applications could be simultaneously sticky and consumable.  It seems to me that processes/process steps that can be consumed — purchased and used only as needed — won’t likely have all that much “stickiness” — hard to replace/move away from. 

This issue came up again in the commentary around NetSuite’s Q4 release.  Sramana Mitra notes (here) that N has stopped giving visibility into its churn metrics (noted here), which metrics should give some insight into stickiness.  She then discounts any concern about stickiness by analysts (noted here), by posing the question:

Why would a company, once in, want to switch its ERP system?

I doubt if that’s the real churn issue here, switching.  It seems more likely that SMEs would simply stop using NetSuite if they found little value.   Also, I’m not sure that NetSuite becomes the ERP system, at least not at first.  Implementing quickly has its advantages, but deep integration into a firm’s way of working is not necessarily one of them. 

Benioff: SAP and Innovation…not!

Just ran into a new SAP blog at (here), written by Demir Barlas.   The first post I read highlighted Marc Benioff at ranting about how:

…SAP has “not seen innovation in the last 10 years.” Those comments were made in a recent interview with ZDNet, in which Benioff twice claimed that SAP has not been responsible for creating anything “unique to the industry or value-added technology.”

Huh…this from a guy who followed the time-tested “innovation” of porting of an application from one OS or architecture to another (Multics calculators -> VisiCalc -> Lotus -> Excel = ACT -> Siebel ->  Demir sets the record straight, however.

The logical response to that assertion is that, whatever the case with SAP may be, is not an innovator either. Siebel tried hosted CRM before, and the AppExchange is manifestly indebted to the iTunes and eBay B2C models.

When is going to scale?

Another impressive revenue quarter for CRM, but when are we going to see more bottom line from that vaunted SaaS scale?  George Gutkowski at Seeking Alpha has some comments (here).  

I suppose that can plausibly say that they’re not seeing scale because of the platform play, but wasn’t supposed to leverage what they already had in place? And reducing transparency on subscriber growth ain’t a great sign either…

SFDC outage — Can one really “test” packaged software

Apparently SDFC went down for a few hours on 11 February…’s release testing process is broken. Testing should have caught these problems before they were released into the wild.

OK, testing “should have…” is a truism.  But don’t we forget what inherently complex machines packaged enterprise software are?  Once configured and filled with data, aren’t they virtual (and often real-time) models of flesh-and-blood business.  These aren’t trivial exercises in coding.

My position is that any packaged software with complex features and many customers is almost impossible to test exhaustively before release.  Thousands of transactions, tens of thousands of tables (some with many million records), millons of lines of code is difficult enough.  SAP has 40K ERP customers with wildly different configurations and with users can modify, enhance, and otherwise interact with the application.

How does one build a test system, never mind test cases, to test that in a non-geologic timescale?  The quality must be built in from the start — an inspection approach to quality is doomed to fail.

SaaS/On Premise Coexistence

Jeff Nolan gets beyond the gnostic, either/or thinking that pervades the valley when he reiterates that on premise ain’t going away even in the Enterprise 2.0 world.

Being web-based behind a firewall or on the public internet are not that dissimilar. There are enterprise 2.0 apps that depend on a network effect or a market that is strongly aligned to on-demand hosted, but the fact remains that many companies and many markets still want on-premise capabilities. This is one point that Oracle and SAP both nailed, rather than chasing the multi-tenant hosted trend they listened to their customers, who by and large simply weren’t asking for it.

I’m trying to work through the heuristic firms will use to decide what goes SaaS, what stays SaaS, what goes On Premise, etc.

  • Will SaaS really have a big role in value chain segments that are critical to a company’s success?
  • Will enabling processes like talent management continue to see SaaS boom, or will they get sucked back OP?
  • Will SOA be how we get the best of both worlds, where OP apps can call services that provide fresh, complex content (e.g., GRC apps where the risk data stays OP, but regulation-driven content like checklists, codes, etc. can get pulled in on demand)?

On Demand/SaaS and New Year’s

New Year’s always brings back memories of applying the old “LCPs” (Legal Change Packages) to R/3. LCPs are required to ensure HR systems reflect the latest legislation or administrative rulings for the various jurisdictions that the HR system covered. Not exactly a value-added activity, as needed as it was.

This type of content update is exactly where SaaS shines. The single code base justification for SaaS gets a lot of play, but I’ve always liked the single content base angle better. This angle is relevant in HCM and GRC solutions (e.g., COSO, COBIT updates), other regulatory environments (banking), and enabling methods and tools (e.g., implementation methodologies created, maintained, and distributed via wiki).

Pareto, Saas, and the “Real World” (Part 1)

All ERP customers lament the size and complexity of their implementations.  My experience has been that they just don’t get what they’re trying to do — create a real-time model of their businesses.  Most IT folks have been used to only automating bits and pieces, without looking at how they optimized the whole.

The seduction of SaaS is that you’ll be able to get “good enough” or “roughly similar” functionality as needed, at dramatically lower cost.  The core processes will work just fine for most everyone.  Per the consultant’s mantra to calm the customer: “We want the 80:20 solution, not the perfect one, right?”  Vinnie Mirchandani has hit this theme again and again (here, here, and here)

I don’t quite buy it.  Someone, somewhere is going to have to interface with the real world.  So many of the implementation cost drivers are that last 20% of customization — RICEF+Workflow objects — which provide the bridge from bits to bricks.

  • Reports for the pointy-haired among us and the authorities
  • Interfaces to other systems — e.g., barcode readers, handhelds, RFID, etc.
  • Conversions for systems that go away
  • Enhancements that add customer-specific logic
  • Forms for Customs, Shippers, IRS, Inland Revenue, Zoll, Douane, etc.
  • Workflow for approvals, alerts, etc.

This 20% that gets talked away during process design time gets talked right back in once the business gets its hands on the solution.

%d bloggers like this: