Old News Keeps Flowing ...
It's funny that it took Emptoris what seems like six months to make this announcement public. I first wrote about the rumored news earlier in the spring here and here. In one of those posts, here's what I had to say: "those in the Spend Management gossip circle know that that Mindflow has been shopped around since Bush's first term in office. And just like the Junior Senator from New York, the vendor has carpet-bagged its way around the Spend Management vendor world looking for votes in unfamiliar districts. But given all the chatter at the moment, it looks like this campaign is about to end, with a full endorsement from a leader in the market ... I've triangulated from multiple sources that the vendor who is about to pick up Mindflow is Emptoris."
I've got a few minutes scheduled to chat with Emptoris later today about the deal, and these are a few of the questions I plan to ask when we talk:
Why did you buy anothes optimization vendor given your reputation as the original leader in sourcing optimization?
Will the old image of Mindflow's slow solving time taint your reputation? With apologies to Michael Lamoureux, this was an issue that was discussed openly in the market for a while.
I can understand the up-sell approach and buying an installed base and maintenance revenue, but most of Mindflow's customers (many of whom at least considered price as one major consideration) would choke at the numbers you tend to put on the table for deals. How is this working out so far?
If readers would like to know anything else, post a comment or drop me a line, and I'll see if we can weave it into the conversation.
- Jason Busch
I've got a few minutes scheduled to chat with Emptoris later today about the deal, and these are a few of the questions I plan to ask when we talk:
Why did you buy anothes optimization vendor given your reputation as the original leader in sourcing optimization?
Will the old image of Mindflow's slow solving time taint your reputation? With apologies to Michael Lamoureux, this was an issue that was discussed openly in the market for a while.
I can understand the up-sell approach and buying an installed base and maintenance revenue, but most of Mindflow's customers (many of whom at least considered price as one major consideration) would choke at the numbers you tend to put on the table for deals. How is this working out so far?
If readers would like to know anything else, post a comment or drop me a line, and I'll see if we can weave it into the conversation.
- Jason Busch
Mindflow's solver was - and still is - slow! It is a fact of life that with regards to optimization, you have to understand the following - you can have any two of: deep, accurate, and fast. But just like a consultant will truthfully say you can only fix two when it comes to price (& resources), time, and functionality, you can not have all three simultaneously!
The reality is that the MindFlow model was, for years, much deeper than any other model available (sorry Emptoris, you may have been the "original leader", but as far as I can tell, only CombineNet has deeper models, and these models are primarily logistical), and, by default, configured for absolute maximum accuracy. This meant it was destined to be slow. It did not help matters any that the solver was usually run on a mid-end server (at best), that even most of MindFlow's own consultants did not know how to alter the configuration parameters to trade-off accuracy and a small piece of optimality for a faster solve time (and the reality is that even allowing for a[n additional] 0.1% inaccuracy can greatly speed up solve times in some models), that they would not invest in joint research with either the vendor or the local university to investigate potential methods to speed up solve time in general, and never did enough education on the upsides and downsides of optimization and how to find the proper balance.
Furthermore, the model was designed for strategic category sourcing and not "everything in the bucket sourcing". Because of this, and because of monetary considerations, MindFlow chose a solver that was optimized for small and mid-size models. However, some of MindFlow's customers had "categories" that were essentially "everything in the bucket" sized, which produced (very) large models, for which the solver (and the default platform it was run on) was not the best choice. [I have run side by side comparisons of Dash Optimization's XPress, ILog's CPlex, and Sunset Software's XA, three of the leading (mixed-integer) linear programming solvers on the market. They all have their strong points, and for certain model sizes they are all essentially equal, but once you get (very) large models, you find a clear winner in the speed category.]
I do not know about you, but when it comes to large dollar spends, I'll take deep and accurate over shallow and approximate any day - but I'd also make sure that (a) I was running on the best damn platform I could get my hands on (which included the best solver for the model at hand) and (b) that the accuracy was not set any higher then I needed for good results (for example, on a million dollar spend, I do not need 0.0001% accuracy confirmed, within 0.1% is plenty!).
A much faster and more detailed response than I expected, though I had a gut you would chime in before I finished off my second cup of coffee this morning. Thank you.
Yours in OR,
Jason
CombineNet's solver is deep and designed for 'everything in the bucket' for sourcing- nothing (outside of common client use and market perception) is particular to logistics. That is, the CombineNet solver accepts as input items/bids/etc - not lanes/carriers/etc.
One note on the delay of Emptoris' press release:
I believe the deal was complete months ago and was kept close to the chest due to the timing of SAP's Frictionless announcement.
Remember, Emptoris was rumored to have been an SAP acquisition candidate. Andrew Bartels (i'm told a close analyst to Emptoris) covered the SAP/Frictionless announcement and highlighted that SAP was still missing, among other things, bid sourcing optimization. I believe Emptoris waited for Bartels report to announce the Mindflow acquisition to make the acquisition appear like they are shoring up their bid sourcing optimization to make themselves more attractive or responding to SAP (and analysts).
A lot of speculation here. Jason was going in the right direction when he mentioned building on strong organic growth with a strategic acquisition to further expand the installed base. And, to your question, Jason, yes, Emptoris has seen some real traction with the Mindflow customer base—they’ve kicked the tires for sure (with Mindflow) and see the value in the larger solution.
As for all the who’s faster, better, stronger, at the end of the day, it's really about how easy it is to use the optimization (so it can be leveraged by hundreds of procurement professionals—not just a few consultants) and how flexibly it is (to address all your commodities) so you can have a level of impact on the business that’s substantial (noticeable on the bottom-line). As a proof point, I’ll defer to the “who’s who” of Operational Research and Management Sciences, the Franz Edelman Award judges, who awarded Motorola the prestigious top prize for using Emptoris’ optimization to save $600M. That sounds faster, better, stronger to me.
Applause to Ed Macri for honing in on the real question when evaluating optimization solutions - "What do I need and what is going to get me results?"
Although MindFlow had more depth with regards to sourcing models than many optimization solutions on the market, the reality is that the full model was, and is, overkill for many (strategic) sourcing scenarios and, thus, many customers. And when you are talking about optimization, too much (forced) capability is often just as bad as too little capability - as you have to spend more time building a bigger model that will take much longer to solve, and if you are having to estimate data that you do not have, the solution might actually be worse! The best optimization solution is the one that fits your needs - and if this is a lighter solution (which is often the case), then that solution will, as Ed points out, be a "faster, better, stronger" solution.
Furthermore, as I have hinted at in a few of my posts at SourcingInnovation and eSourcing Forum, decision optimization is at its most effective when it is part of a larger solution, and to be precise, when it is part of a suite that covers the sourcing cycle. And, as Ed points out, when it's easy to use. (Unfortunately, unless you were a real power user, which is not true for the majority of procurement and sourcing professionals, MindFlow was not an intuitive solution.)
On a different note, Paul points out that he believes Emptoris waited to announce the acquisition to make themselves appear like they are shoring up their bid sourcing optimization. I'm interested in what Emptoris is going to do with the MindFlow product "suite" going forward.
The idea that the Emptoris acquisiton of Mindflow was a market-share play doesn't make any sense for a variety of reasons, not the least of which is the fact that Mindflow's primary delivery model was behind-the-firewall -satisfying companies who simply refused the ASP delivery model due to data security concerns.
As many of you know, Mindflow was partnered with CombineNet -prior to its acquisition. At that time, Jay Reddy, founder of Mindflow, not only openly acknowledged CombineNet's optimization capabilities were/are without peer, but Mindflow's optimization capabilities exceeded those of Emptoris' -which should be obvious when looking at the recent transaction. It would lead a logical person to conclude that the acquisition was in large part a move by Emptoris to catch up.
The Emptoris reference to the award is 'one from the archives' - going back to 2004- the recent awards featured CombineNet as a finalist in 2006:
http://www.informs.org/article.php?id=453&p=
While the rest of market relies on growth through acquisition, making moves that often force them to acquire redundant functionality, CombineNet has remained true to its core. Our clients continue to see steady growth and improvement in usability, more strategic category support and optimization-enabled, self-service features that drive uniquely competitive value. Our clients are driving optimization-enabled requirements. They are not asking us to add or replace functionality that overlaps with their existing ERP systems -or sourcing applications.
Most of CombineNet's R&D money is spent on improving the user experience and making optimization more relevant -not just in a sourcing context, but in a supply planning context. We like where we are.
Paul
Although it pains me a bit to say this, I do believe that at this point in time, and for the last couple of years, CombineNet's optimization capabilities are, and were, without peer in the realm of customized algorithms, but, as I pointed out above, it's not just algorithms. You can have the best solver in the world (a claim I'm sure Ilog would challenge if CombineNet chose to make it, but their solvers have their weaknesses), but if the model, and the modeling capabilities of the tool, are not appropriate to the problem, or not useable by the end user, it does not matter how good the solver is.
As I indicated in my first comment, I believe that CombineNet has logistics down pat. Since that is where they achieved early traction and how they won a lot of their customers, that is where their capability evolved rapidly. However, as for sourcing, I'm not entirely sure that they are "without peer". Why? I've read many of their academic publications, or should I say, the publications of their researchers, and although it's quite clear that these are top notch optimization experts, it's one thing to be an expert in optimization and algorithm construction and another thing to be a domain expert in sourcing.
So, bring on the slug-fest and let's find out once and for all what is speculation and market perception and what is truth.
Thank you for your kind words.
Please contact me and I will personally walk you through our application.
412 535 6067
I look forward to your feedback.
Paul Martyn
Paul, I'd love a walkthrough of your current capabilities. And, in addition to a SpendMatters sponsored bake-off, I'd also love to post some deep dives on both CombineNet's and Emptoris' capabilities over on SourcingInnovation.
The reality is that there are not a lot of vendors (left) with optimization, many users do not really understand the upsides and downsides of optimization, and, even worse, most users do not really have a good handle on what your solutions have to offer (which makes Ed's comment that there is "A lot of speculation here." a sad reality).
Now, I know a few people have read my first comment to imply that I thought MindFlow was "better" when I said that it was, for years, "much deeper than any other model available", but as I have clarified in my later comments, deeper is not always better. Sometimes you need light, sometimes you need fast, and sometimes you just need raw solving power. The keys are flexibility and applicability to your problem domain - which, in this case, also requires a deep knowledge of sourcing. (And, to this end, to be fair - as I complemented CombineNet in my last comment, I'll applaud Emptoris for their deep knowledge of sourcing and history of always making the right acquisitions which have bolstered their application pool to the point where there are now few competitors who can even think of challenging them on spend and contract management.)
So, for those of you keeping score, in addition to taking a jab here and there, I've now complimented MindFlow, CombineNet, and Emptoris. Personally, I think CombineNet and Emptoris both have some great solutions. In my view, what is up in the air is how flexible their respective solutions are and how much of the current sourcing marketplace they are appropriate for.
So, again, I'd love to see the bake-off, demonstrations of, and more materials on, current solutions, lots of posts, and some deep discussions on optimization in general. Where optimization is concerned, I think the best is yet to come. Let's start educating the marketplace!
< Sourcing Innovation Blog: http://blog.sourcinginnovation.com >
Appreciate all suggestions on the same.
CombineNet and SAP are two entirely different solutions. The latter is a traditional ERP with some sourcing capability and the former is an optimization platform. SAP does not have optimization capabilities.
Whether CombineNet is economically viable depends upon the expected return of investment. How much do you spend each year on direct materials and reasonably complex service agreements? Based upon analyst and vendor benchmarks, how much does the average firm save on these categories when applying an advanced optimization solution? How conservative are you - and what percentage of these savings are you willing to bet on? How much will the CombineNet solution run you for the categories in which you expect to see savings significant enough to warrant the optimization solution.
The calculations are:
Expected Savings = Sum (Spend * Average Savings Percentage * Expected Realization
Deployment Cost = Sum (Extra Man Days of Effort * Daily Cost)
Return = Expected Savings - Solution Cost - Deployment Cost
If Return >= 3-5X Solution Cost, it's worth it.
I'm assuming you already have SAP. If you don't, and you need an ERP then, unless you are very large and very deep pocketed, I'd recommend also looking at some of the newer open-source ERPs like Compiere and Open Bravo. These solutions are often more than enough if you are a smaller or mid-size company. (If you are focussed on distribution, try 3rd Wave as well.)
Also, if you haven't, I'd also recommend comparing CombineNet to the other optimization solutions, like Emptoris and Iasta, before making a final selection to evaluate what you are getting for your dollar. CombineNet can carry a significant price tag - which is well worth it if the expected return is there, but not if the expected return is minimal. It is true that there are some models that are pretty much exclusive to their solution, but there are others that will solve just as well (even if they take a little longer) in competitor's solutions, which have a lower cost. It's all about ROI. If it's there, you go for it. If it's not, you don't. If it's in the middle, you look for another solution that nudges it up to the point where you go for it.
For more on decision optimization, and CombineNet, see my blog if you haven't already:
http://blog.sourcinginnovation.com/categories/Deci...%20Optimization.aspx
U mentioned that SAP does not have optimization capabilities. I wanted to clarify if the Frictionless acquisition has provided them with optimization capabilities or have they buried those skills?
Also are you aware if Ariba's optimization toolset is developed in-house or is it based on ILog like other solutions from Procuri, Iasta and Emptoris?
Regards,
Gagan
( 2 ) If SAP has optimization capabilities, it's not in the strategic sourcing arena. (And if they are developing, or planning to develop, any, they're doing a great job of keeping it hush.)
( 3 ) Ariba is about as close to nothing as you can get. I haven't heard them advertise optimization in about four years. I don't know if what little they had was developed in house or based on a commercial solver like iLog, but even Emptoris had them beat back in the day.
Michael
Michael makes some excellent points.
We are available to answer your CombineNet specific questions and have ample SAP client references (testimonials to how easy we integrate).
www.combinenet.com (+1 412 471 8200)
Paul Martyn
Chief Marketing Officer
CombineNet
"Optimize Supplier Awards:
Identify the lowest total-cost and best-value award strategies
by using rules-based functions to optimize your award
decisions. Easily analyze optimization possibilities by comparing
options using reports and charts."
i.e. Tell it how to divide the award with a rule, and it will create an award scenario - and since it's what you want, it must be the optimal solution and, just to be sure, you can tell it what you don't want, and compare what you want with what you don't want side-by-side in a report!
Wow. It makes the argument that:
* Wealthy people tend to be successful.
* Wealthy people typically drive luxury automobiles.
* Therefore, wealthy people are successful because they drive luxury automobiles.
look good in comparison.