Want cheaper analytics? Snub SaaS: Ye olde ELAs might do the job

Pricing and licensing "are central to all data and analytics initiatives" and vendors are using organisations' enthusiasm for the field to coin it at your expense. But controlling costs is more than possible, Gartner principal analyst Etisham Zaidi said yesterday at the firm's Asia-Pacific Data and Analytics Summit in Sydney.

In a talk titled "Modern Data and Analytics Initiatives Require New Pricing Models and Negotiation Strategies" Zaidi outlined vendor tactics like "land and expand" whereby an initial engagement on a freemium price plan can quickly spread and cost big bucks. Vendors will also cash in on the fact that a new analytics initiative creates a need for tools that organisations may not already possess. Zaidi mentioned metadata management as such a requirement that creeps up on organisations excited by the prospect of better analytics but inexperienced in its operation.

Zaidi also said that vendors of SaaS analytics tools are only too happy to let you keep buying more and more seats as your needs expand - even when doing so sees the price exceed the cost of boring old three-year enterprise licence agreements. Departmental buying of analytics can have the same effect: teams that buy the same software without central control can mean an organisation collectively overpays for software that could be had for less on a central licence.

Another thing to watch is complexity being used to drive up support costs. The large number of dependencies in Hadoop, for example, means vendors of distributions have been able to ratchet up their support prices because users can't keep up with all the updating that needs to be done to keep the tool current and secure. He advised that smart organisations will negotiate flat support rates, or rates that grow only at the rate of inflation, rather than let vendors set higher price jumps.

Tiered support is another way to save some dough. Zaidi said vendors generally try to have customers sign up for the same level of support for all users. But developers don't typically work weekends, making it sensible to buy some support for their nine-to-five existence and weightier other contracts for other users.

Paying for storage and compute discretely is another tactic the analyst recommended.

He also recommended checking the capabilities of the software you already own and its potential role in an analytics project. Conventional DBMS, he said, often possesses more-than-decent unstructured-data-handling capabilities, making purchases of new software unnecessary.

Zaidi's added that buyers should do their most serious negotiations towards the end of vendors' financial years, when sales teams are under pressure to hit targets. When negotiating at such times, Zaidi advised, know your preferred vendor's key competitors and the products they offer. Mentioning those rival products will quickly produce ten to fifteen per cent discounts.

While he also advised "vendorising" requests for proposals so that vendors you fancy can respond easily, he also said buyers must never let a vendor think they're the only one left in the race for a deal. ®