Anshu responds to the grouse that ERP systems are:
1. Prohibitively expensive
2. Hard to install and implement
3. Even more expensive from a support and maintenance costs angle
4. Vendors do not pass the 'amortized' RnD costs to customers
5. Vendors do not pass the cost savings leveraged due to development and support from offshore locations in low cost labor markets to the customers
6. The costs for implementation have not gone down even after more than 100,000 implementations and past learnings are not leveraged to reduce newer costs (see in relation to 2 above)
Guys like Vinnie have been cribbing about the high costs for ERP systems since time immemorial. Seems even the Utopian promise of Saas vendors like Salesforce.com (who offer a fixed, predictable monthly maintenance charge per user with no license costs) is able to quench their thirst to reduce the costs further. Anshu makes the point that if you feel the costs are high for an ERP and costs have been amortized, why dont you do one of the three - build one of your own, keep quiet and think why, or be a reseller to sell the best.
One way to look at this is what are the choices the customer has when she looks at an ERP (assuming the case of being a laggard oops.. a late adopter/Gartner type C and thus expects end of life (?) pricing and costs:
1. Do nothing
2. Build her own
3. Go for SaaS and have a fixed, predictable opex with little capex.
4. Go to the blood thirsty breed of today's vendors and live with the capex and opex.
Out of these, if we rule out 1, the only viable options are 2 and 3 which can classify as economic substitutes to 4. What will happen to SaaS is anybody's guess, so 3 is a do not know yet (comes with its own set of risks and bells and whistles). On 2, I have the following premise:
- It is much much harder to implement than 4 even if you rule out the costs
- The costs will in all probability be higher than what you end up incurring if you go with an ERP vendor.
- Its like reinventing the wheel for much of the already avaliable robust functionality like order processing, payroll, payables and receivables which makes it more bug prone and riskier, even for a type C profile.
- The IT major who implements it is now your new blood thirsty dracula - she has a open field to herself when it comes to costs, maintenance and fixes. Reminds me of the IBM Mainframe situation.
Does anyone have any other option? If not, why do you crib about the costs?