Since the beginning of Gartner’s SWAMI Software Asset Management conferences over 20 years ago a large number of speakers, consultants, industry “experts”, and research analysts have given hundreds of presentations on how to negotiate a good deal. The measurement of a good software deal has consistently been measured by (1) how much you can get the software vendor to give in on their T&Cs, and (2) how big of a discount is negotiated. The emphasis is on the discount.
The pursuit of a good discount is often the primary measurement for a successful negotiation. That reasoning is flawed because the best deals are often bad for your health. Like in many industries, if you want the lowest per minute rate on your mobile plan, you need to purchase the most minutes. If you want the best deals at the department store, you need to purchase more items. The more items purchased, the better the discount. Buy one item, get the second for 50% discount, or buy two items, get the third free or buy 12 items for the price of 10. But what if you only need one item?
In an extensive pricing analysis at a major global financial service company, it was determined that it received a 95% discount from a software vendor that rarely gave discounts greater than 40%. However, they paid $70 million above list price if they had licensed their software in an efficient data center environment. Their problem wasn’t the discount, but the cost. Having focused on their discount, they had ignored the cost of the ELA, which caused them to inadvertently overpay.
Based upon 20 years of researching software ELA pricing and costs from hundreds of data centers worldwide, data show three primary causes that differentiate best in class data centers from the average data center. The single biggest differentiator is product count and mix. Data centers that are able to effectively manage the product count and product mix consistently pay substantially less. Knowing how many products are needed (not wanted) first requires understanding the functionality required. Purchasing products without knowing the functional requirements will inevitably lead to purchasing more than necessary and often purchasing from a competitive vendor at a higher cost. Typically, when measuring the variance between average software costs from best in class, product count/mix accounts for about 50%-60% of the variability.
The next largest contributor to variances in software costs is product licensing. How and where a data center licenses software represents 25-35% of the variability of software costs. For example, were 500 enterprise editions in an environment only 25% virtualized the most effective solution or should there have been 125 enterprise editions + 150 standard editions in an 100% virtualized environment? Knowing what type of license in what environment with what type of hardware architecture will always result in significantly lower costs than what the best negotiators were able to negotiate from a proposal given by the software vendor.
Finally, the common denominator focused on the most by consultants, research analysts and data centers are pricing, and they yield the lowest variability in a data center’s software costs. Research across all industries shows that to get the best discounts, you are required to purchase the greatest volume. That is why whenever I hear people tell me about the great discount they negotiated, I invariably ask how much waste did they have to buy to get the discount? So the next time you negotiate a software agreement, remember that when a vendor offers a great discount, it’s because they realize you may be purchasing more than you need and that is why the best deals may be bad for your health.