Why investing in a software to increase transparency, quality and efficiency through automation?
The reason is obvious, but still we are asked how the costs can be justified, so we provide a value assessment, where we calculate the savings through automation.
Sometimes it is also not about saving money (external/internal resources), but simply adding a capability which is otherwise not performed and the risk of missing out (when outages/delays/downtimes occur) is even higher…
However, there is also non-quantifiable value:
- Outside: The perception of the integration team is more professional (by other teams in IT and business)
- Inside: The self-awareness of the integration team is increased (they receive proper tooling and there is no more wasting of time with stupid admin activities)
Regarding quantifiable value and cost justification through automation we use the following parameters (of quality assuring interface related activities):
- Inventory: Transparency
- Less effort to maintain an up-to-date inventory/catalog of all your integrations
- Less effort to determine where specific technologies/users/routing rules are used in integrations
- Faster changes
- Less effort for enterprise architects to maintain the EAM-Tool (e.g. LeanIX, LUY)
- Less effort to create (standardized) technical documentations
- Less effort to align with other (IT-) departments
- Checks: Monitoring & Reporting
- Less effort to check messages, queues and channels (errors)
- Less effort to check certificates (expiration)
- Less effort due to incidents because of missing/slow messages
- Less effort to allocate costs to interface consumers (departments/units/countries/brands)
- Less effort to check if interfaces are deployed across the landscape with the right version
- Less effort to identify which interfaces can be decommissioned
- Less effort to execute quality gates
- Less effort to identify slow/instable/high-volume integrations
Again: If we do not perform the activities above, we take the risk of downtimes and outages and our business processes are suffering (delays, revenue loss, fees, external costs)…
Value Assessment Approach
- Achieve consensus about the need of each activity (e.g. cost allocation might not be relevant for your business)
- Determine which of the activities above are still done manually
- Estimate the manual effort of each activity (hours per week)
- Multiply the costs with an hourly rate (internally with at least 50 EUR/USD)
- If the total costs are higher than the subscription, then you have a valid business case!
Example (with low figures)
- Inventory: Transparency
- Keeping a basic inventory up-to-date requires in average about 1/2 hour per week per integration layer component (adding new interfaces, updating, removing deactivated ones).
- Searching the inventory (for specific technologies, security, routings) takes hours if it not available in a more detailed inventory. It helps to speed up all kinds of initiatives and interface-related changes and transformations (e.g. release-changes, carve-out/merger, move-to-cloud/datacenter migrations). To support impact analyses we also calculate with an average of 1 hour per week. This clearly depends on the interface landscape and can be much more.
- Putting together test lists (“all interfaces related to warehouse management/EDI/E-Commerce”) are use-cases of simple catalogs and not considered separately
- Updating the EAM (based on a basic inventory to also keep this in sync) requires another 1/2 hour per week.
- Generating technical documentations manually typically requires about 3 hours in average. In addition we assume that about 20% of existing interfaces are changing once per year with additional 1 hour. So generating the documentation for a landscape with 1000 interfaces requires 60 h per week (one-time) and another 4 h per week for changes. In a mixed calculation we use 10 h per week.
- The fact that you are 1) enforcing, 2) standardizing and 3) always max. 1 week old are just soft factors and not considered in the calculation.
- Checks: Monitoring & Reporting
- Manual interface landscape checks (periodic checks twice a day to see if all is ok wrt messages, channels, queues) take 15 minutes each, so 2,5 hours per week in average (this is an absolute minimum).
- Checking for expiring certificates can be done once per week with 0,25 hours (in reality probably once per month and then it takes one hour).
- Working on incidents which are based on missing or slow interfaces where we get triggered by other departments (“where is file xyz?”, “why is the portal performance so slow?”) takes another 2 h per week in average.
- Generating a cost allocation based on traffic or message volume takes 1 day per month, thus 2 h per week.
- A manual weekly check to verify if all interfaces are deployed in the correct version across the landscape is not realistic to handle. However, even if only a subset is checked we can assume at least 2 h per week.
- Determining a decommissioning list for each integration layer would assume that you have a (basic) inventory and check manually if there was traffic for the last month(s). This can be done on a monthly basis and would require 4 hours, this 1 hour per week.
- Quality gates can be performed very individually, but even if you only enforce basic aspects like naming conventions, descriptions it takes another person (not the developer) about 15 minutes per interface. When we use the same assumptions like for the technical documentation (initial build and changes) we can consider 1,5 hours per week for a 1000 interface landscape.
- The hunt for exceptional integrations (slow, instable, high-volume) is often ignored, but required to have all interfaces in a good shape and avoid downtimes. Analyzing them manually as a weekly activity would require at least 4 hours, just to know for a given subset (your main business processes) if there many retries/delays or unexpected traffic.
When we sum up ALL aspects, we have a total of 27,25 hours per week, split by areas as follows:
- Inventory: Transparency: 2h
- Documentation: 10 h
- Checks: Monitoring & Reporting: 15,25
When you sum up the hours with an hourly cost rate of 50 EUR/USD, you have a total of 71.500 EUR/USD per year.
Conclusion: The yearly costs for an interface management solution should not be higher than that. Of course it must fit to your needs (make or buy as a usual question in software) and you have to adapt it to your own complexity (# of integrations, # of integration layer components).