It is commonly stated that tier 1 ERP systems are very advantageous, particularly for large buyers. One of the reasons that are often presented is that using both tier 2 ERP systems as well as other applications from the same software vendors saves buyers money. And in fact, that while it is well recognized that the major vendors provide inferior functionality to best of breed vendors (although the term “inferior” is rarely used), the trade-off of accepting this inferior functionality is well worth it when one considers the substantial cost savings that are assumed to be available from buying most of one’s software from one vendor. This is the standard logic presented by both large software vendors and large consulting companies that are mostly their partners.

Most often, the single software vendor is either SAP or Oracle. Still, other software vendors such as Infor, which have grown through acquisitions, will make essentially the same argument to their potential buyers. Software vendors will often develop because they anticipate that the profit margins available for them if they can offer “full suites”, is greater than if they provide more narrow and targeted solutions. As is explained in the book, Gartner and the Magic Quadrant: A Guide for Buyers, Vendors and Investors, IT analysts also tend to promote large software vendors as more stable and preferred vendors for their readers. In fact, because of how the Magic Quadrant is configured, the simple act of one software vendor acquiring another software vendor will move the acquired vendor’s products up in the ranking. This is because so much of Gartner’s Magic Quadrant criteria are simply a proxy for the size of the software vendor.

The (Lack of) Evidence for the Single Vendor Hypothesis

While the logic or reasoning presented above has been highly influential and purchasing decisions for decades. In our research, we could find no entity that ever demonstrated, or even attempted to quantify the total cost of ownership for following a single vendor approach versus a best of breed approach. Considering how commonly the logic of the single software vendor purchasing strategy is invoked, we found this curious. Before reviewing our quantification of this issue, several characteristics of the single vendor argument should be understood.

  1. Financial Bias: Most of the entities that propose the single vendor argument have a financial bias. They are either large vendors with many applications. Therefore the advice they provide to purchase as much software as possible from a single vendor (that vendor curiously enough always being them) is self-serving. The other major proponents of this purchasing strategy are large consulting companies. However, they have built their IT consulting practices around specific large vendors, and have resources trained in the applications of the significant vendors. SAP and Oracle trade away a great deal of their consulting business to the major consulting companies in return for recommendations to purchase their software to the clients of the major consulting companies.
  2. No Attempt to Provide Evidence: The single vendor argument neither has any evidence to support it, nor do any studies exist to support it. Furthermore, it is not designed to be either proven or disproven. It is instead part of a marketing program to increase sales of the software vendor and the consulting companies, which propose the hypothesis.
  3. The Research Problem: No entity that would be traditionally relied upon to perform research on this topic would be able to complete the research without financial bias influencing their results, as the large software vendors in some way compensate them. For instance, even IT publications receive a disproportionate percentage of their advertising from either large software vendors or large consulting companies – both of which would push against any research which showed that the single-vendor approach was less practical/more costly (etc.) than a more diverse approach. Publishing such research would be an excellent way to have one’s advertising revenues cut. Entities do not publish research, which contradicts their financial model. The one entity that could research this topic is the academic system, and there is no record of this type of research every being performed by an educational institution or published in any academic journal. This should not be surprising. Academics do not investigate every issue that is of interest in the industry.

This Analysis and Assumptions

This analysis is based upon the individual application TCO Calculators that have as single detailed analysis that breaks down the TCO into software, hardware, implementation and maintenance costs. This 100% Oracle Versus 100% Best of Breed Calculator brings the analysis up a level of abstraction and only lists the overall TCO of each application – but lists the TCO for multiple applications, and then compares competing solution architectures. In our application, TCO Calculators, there are inputs, which change the predicted TCO, including the sophistication level of the implementation and the degree of customization, among other factors. To produce apples to apple comparison as well as to simplify the analysis, all applications are listed as their purest state.

  1. The uniqueness of Analysis: This is the only analysis of this kind that we are aware of exists. We believe we are the single source that compares the costs of the single-vendor strategy versus the best of breed strategy in a published form.
  2. The Underlying Research: The TCO analyses are based upon rigorous and detailed analysis into each application. This is a bottom-up analysis where first the TCO is estimated for each TCO component, then added per cost category, and then finally aggregated to account for the overall TCO. For those interested in the detail below that shown in this analysis for any specific application, we recommend our application-specific TCO calculators. Those offerings have adjustments that can allow them to be matched to the actual implementation environment.
  3. Estimated Integration Costs: We do not break out integration costs separately. A primary reason for this is that when dealing with software vendors, they also do not break out integration costs – but instead, they quote the overall implementation effort. Therefore integration costs are included in the total TCO costs.

The Comparisons

Our first comparison shows all Oracle applications versus all best of breed applications. The first comparison, or Alternate One, which follows, has no ERP system but instead uses a best of breed application in each category. In our 100% best of breed application grouping, Rootstock would perform the non-financial ERP functions, and Intacct would provide the financial functionality. Arena Solutions provides a bill of materials and PLM functionality, which is much greater than any similar functionality inside of an ERP system.

Alternate One - 100% Oracle VS 100% Best of Breed (Small to Medium)

This compares going 100% with Oracle versus best of breed in an integrated solution.
CategoryApplicationTCOUser #
ERPOracle JD Edwards EnterpriseOne$ 23,400,000200
Production PlanningSAP PP/DS$ 3,059,02510
Supply PlanningSAP SNP$ 2,454,90012
Demand PlanningSAP DP$ 2,794,90012
Business IntelligenceSAP BI/BW$ 5,945,48040
CRMSAP CRM$ 5,954,48050
Bill of Materials/PLMOracle JD Edwards EnterpriseOne*Costs Accounted for in ERP Covered by ERP
Inventory ManagementOracle JD Edwards EnterpriseOneCosts Accounted for in ERPCovered by ERP
Financial & AccountingOracle JD Edwards EnterpriseOneCosts Accounted for in ERPCovered by ERP
Total$ 40,736,878324

This analysis is for a smallish environment with a total of roughly 300 users. These are generally good value best of breed applications. As one can see the cost savings based on TCO is quite significant. Now we will show the same applications for a larger environment.

*Bill of materials/PLM functionality exists in SAP ERP, but only the bare minimum necessary to support accounting and MRP. Therefore it is listed above as partially covered by the ERP system — however, we do not recommend attempting to manage the bill of materials in any ERP system. 

Costs of Service Level Scenarios - Cost of Lost Sales

Cost of service levels by scenario.
Scenario NameCostMarginICCCLS at 85%CLS at 85%CLS at 90%CLS at 95%CLS at 99%
High Profit Items - Low Carry Cost$50$7015%$28,000$21,000$14,000$7000$1400
High Profit Items - High Carry Cost$50$5012.5%$20,000$15,000$10,000$5000$1000
Low Profit Items - Low Carry Cost$50$2025%$8000$6000$4000$2000$400
Low Profit Items - High Carry Cost$50$1520%$6000$4500$3000$1500$300

This analysis is for a small to a medium-sized environment with a total of roughly 300 users.

We have selected some of the better value best of breed applications. As one can see the cost savings based on TCO is quite significant.

Now we will show the same applications for a broader environment.

Costs of Service Level Scenarios - Inventory Costs

Cost of service level scenarios, profit versus carrying cost.
Scenario NameCostMarginICCCOI at 85%COI at 85%COI at 90%COI at 95%COI at 99%
High Profit Items - Low Carry Cost$50$7015%$1133$1403$1733$2220$3143
High Profit Items - High Carry Cost$50$5012.5%$944$1169$1444$1850$2619
Low Profit Items - Low Carry Cost$50$2025%$1888$2338$2888$3700$5238
Low Profit Items - High Carry Cost$50$1520%$1510$1870$2310$2960$4190

This analysis is for a larger environment, with a total of roughly 1300 users.

Costs of Service Level Scenarios - Net Costs

Profit versus carrying cost scenarios.
Scenario NameCostMarginICCCOI at 85%COI at 85%COI at 90%COI at 95%COI at 99%
High Profit Items - Low Carry Cost$50$7015%$26,868$19,598$12,268$4780-$1734
High Profit Items - High Carry Cost$50$5012.5%$19,056$13,831$8556$3150-$1619
Low Profit Items - Low Carry Cost$50$2025%$6113$3663$1113-$1700-$4838
Low Profit Items - High Carry Cost$50$1520%$4490$2630$690-$1460-$3890

There is some user aggregation on the Oracle side because multiple user licenses fall under the same overall ERP application. Again the cost savings based on TCO is quite significant, although it is less of a cost-saving than applying the best of breed strategy in the smaller environment.


According to our research following a single vendor, the strategy is not only more expensive than following a best of breed strategy but is significantly more costly. It also is important to be emphasized that this analysis included all integration costs. However, even if they had been left out, it is impossible that the costs of the best of breed strategy would have come anywhere near the costs of the single vendor strategy.

Other critical points – in addition to the cost analysis presented above is that the best of breed strategy results in the following attributes:

  1. 1.     Much Higher Functionality of Best of Breed: All of the significant vendors score significantly below the best of breed vendors in application functionality. This is explained in our Software Selection Packages in detail. The more software that is purchased from one vendor, the more uncompetitive the overall grouping of the software will be. A significant reason for this is that firstly, not software vendor can be best in every software category. If a software vendor develops its applications internally; this is because a software vendor that is good at developing ERP software is never going to be the best at developing other categories of enterprise software. In the case where the software vendor acquires applications, most acquired applications tend to languish with little development and tend to become less and less relevant as the time from the acquisition passes. This is explained in our Fake Innovators article.
  2. 2.    Significantly Lower Implementation Risk: Targeted best of breed solutions have a lower risk than software purchased from a single vendor. It’s not hard to see why. Best of breed applications are a better natural fit for the implementation environment and require far less customization. The traditional way to manage risk is to listen to advice from major consulting companies and to purchase software from major vendors. This does reduce the perception of risk, or the political implications of a failed implementation as the executive can say “We bought from SAP and used Accenture – what else can one do?” However, this strategy increases the probability that anyone implementation will fail, increasing the needs of the executive decision-maker to have the political cover of having purchased from a major “brand.”

Our analysis leads to the conclusion that the official logic of purchasing from a single software vendor is flawed, and that it is flawed in every dimension – cost being just one dimension that it turns out to not be true.

Our TCO studies show, and it is no great secret, that vendors with the broadest product lines have the highest TCOs. Furthermore, full product lines do not merely sit on one platform or use one database. The applications have adapters that connect the applications, in the same way, that best of breed vendors have adapters. Of course, the major software vendors have advantages in integration as they own all the applications when a buyer follows a single vendor purchasing strategy. But these hypothetical benefits are not anywhere near as significant as generally thought, and cannot come close to making up for the higher TCOs of large software vendors. It should not be surprising that when one purchases software from vendors that have the highest TCOs per application, that a buyer’s overall enterprise-wide TCO will also be higher.

This is not to say no applications should be purchased from large software vendors – only that the proposal that buyers should buy their applications from a single vendor to reduce costs and to reduce implementation risk – is not supported by research. Our research shows precisely the opposite, that buyers should pick and choose the best software from the best application provider.

We have analyzed this issue from multiple dimensions, and it is difficult to see how the single vendor strategy is even a serious proposal. Our best guess is that it only began as marketing hyperbole proposed by large vendors, and developed a life of its own after this point – and is repeated without anyone noticing that it only would make sense if integration costs were genuinely enormous. And apparently without anyone noticing that the logic is presented without ever evidence provided to support the claim.