Sunday, October 02, 2016

Dynamics 365 and the hopes for the Common Data Model

As many of you probably have heard Dynamics CRM will as of Nov 1 2016 be part of Dynamics 365 in a commendable push from Microsoft to bring the Dynamics products closer together. This is partly a branding thing as the products themselves, as Dynamics CRM, will still be the same product from a technical perspective, at least in the Enterprise Edition, (little is known of the Business Edition) and partly a technical concept as they are introducing something called the Common Data Model which is canonical data model (unified data model) to which all "Apps" are integrated to automatically. This is what I wanted to discuss some.

This all sounds great, and I will admit that I havn't had time to fiddle with it yet, but I have had the pleasure of working with quite a few integration projects between CRM and ERP. And that is not easy, even if you do have a ready made point-to-point integration. So I would just like to make a few points that I hope you do consider before switching it on and hoping it will solve all your issues.

1. Addresses in ERP and CRM are typically not the same. In ERP the addresses that are needed can typically be invoicing address and delivery address, while in CRM the most important addresses are visiting address and postal address. If you naivly presume these to be the same the effects can be dramatic and sometime even catastrophic. I friend of mine, Peter Björkmarker, told me a story of a company integrated just like this, and as CRM was set as the customer data master, it overwrote all invoice addresses in the ERP system with visiting addresses. Next month, all invoices which were sent out were automatically, without anyone noticing sent out to the wrong address, hence nobody paid them. The company got into an accute cashflow problem and almost filed for bankcrupcy. So this is no joke.

2. Ready built integration are usually on a technical level, but you expect it to work on a business level. Integration technology is usually about moving data, but just having the data in the other system doesn't always DO anything. An example is if you have a boolean field on the customer in the ERP where the financial people can block the customer from further business if they havn't payed their invoices. So you integrate this field to CRM and can now see it on the account form. But without any additional logic in CRM it will still be possible to create opportunities, quotes and orders. Maybe not what you would like.

3. Data structures are different. My colleague Rickard Norström, whose blog you can find in the list to the right, was part of a CRM project which integrated to Dyn AX. One of their issues was the AX address data structure. An address record in AX can be used by both an account and a contact, And I think even several accounts. When this address is changed, of course this is seen in all affected places. As this is very different from the customeraddress built in logic in CRM they had to create their own new address entity to solve this. Other typical areas where there are large differences are in the logic of setting prices on opps/quotes/order. As you can expect, a system like AX with MPC and many other deep links into costs can of course use that as a base for pricing, something that is very hard for CRM. It also has more complex or just different ways of handling pricelists. I was working with an iScala integration and iScala for instance can have a current price in a pricelist and a comming price with a specified date on which the new price will be enforced. However, no event in the system will trigger at that time. Customer specific pricelists are also something that occur, not advisable but existing especially for larger customer accounts.

4. Centralized integration architecture. The Common Data Model sounds great but it only handles two of the components in the Business system infrastructure. If you for instance are a Telco the amount of business systems will be a lot more, billing systems, provisioning system, logistic systems, product configurators, etc. Banks are also complex worlds. Many of these have tried to consolidate their integrations to integration hub technologies like WebSphere or BizTalk and if done properly they will of course have their company defined canonical data model. It would be interesting to see the story of how the Common Data Model works together with this. It probably can by shuffeling data using Logic Apps to and from the CDM, but in essence you will have two hubs to orcestrate. Another option is of course to use the CDM as the central hub for all information, as long as that is extendable and doable. So, from this perspective, the main issue is probably, if we have 8 systems connected to our existing integration hub, is it plausible to use the CDM or do we manually integrate anyway directly to each application?

To conclude, I think the CDM will be a good tool but I will keep my expectation to a reasonable level and I recommend you do this too. Do not think it will make your highly customized CRM and AX automatically integrate all data and make it work from a business perspective, that would simply be too increadible. If they manage that, I will buy the entire team building CDM a beer (or similar).

Gustaf Westerlund
MVP, Founder and CTO at CRM-konsulterna AB

No comments:

Post a Comment