Chargeback and showback have different requirements and purposes and should be treated as completely separate processes, even when some elements are common to both processes. This blog concentrates on the chargeback process, and explores the options available for chargeback within AWS cloud services, and how Strategic Blue’s customers can construct effective cost allocation strategies.

Introduction

Chargeback is the process of identifying, allocating and charging the cost to your end customers, or cloud cost owners. 

 Costs can be broken down into two components, Direct cost which is clearly identified as belonging to a single cost owner and a Shared cost which may be distributed across multiple cost owners.

 Both components must be clearly identified, agreed and auditable. This provides clarity for the cost owners and avoids unexpected charges.

Chargeback Allocation Strategy

Principles

Some principles that should be applied:

1. Start simply, and allow your allocation strategies to evolve as your cloud capabilities mature. 
2. Ensure that the effort required to produce your cost allocations is a small proportion of the costs being allocated. Unfortunately, we have seen many businesses spend a disproportionate amount of time and effort to identify, categorize, fully tag and allocate a relatively small amount of cloud spend. 
3. Identify your stakeholder groups and understand how they use the data. Ensure that the frequency and granularity of data match the requirements.

Technical requirements

Whatever the business needs and showback charges are, we recommend using the amortized cost throughout the cost allocation process, rather than the invoiced costs, as this provides a couple of benefits:

1. It avoids the ups and downs when upfront payments are made, and avoids the large cost variations within the month due to how costs are charged at the beginning of the month. 
2. It ensures that commitment costs are more accurately attributed. Commitment benefits and commitment costs often occur in separate locations, but using the amortized view automatically attributes the commitment cost and the benefits to the appropriate account and resource.

Using the amortized on-demand equivalent costs can be useful for engineering teams looking solely at managing and controlling usage levels.

Direct Cost Allocation

There are several ways to allocate costs. Allocation by Account is the easiest approach, as this provides a natural cost boundary and costs are automatically applied to a single Account. However, this needs to be considered when planning your initial deployment, and unfortunately is often an after-thought.

Allocation using Tags should be considered when using Infrastructure as code (IaC), with Tag Editor to manage the key-value pairs. Together, IaC and Tag editor help avoid errors in both the key and the value to provide a more consistent approach. Tags are case-sensitive, which can create issues when tags are manually applied. E.g. CostOwner is different from costowner or Costowner, and ABC-mgmt is different from ABC-management or abc-mgmt.

Uncontrolled and manually applied tags can require a very high effort to apply thoroughly and consistently (breaking principle 2 above). If you use Tags you will need to monitor any untagged resources. Further information is available here. 

Allocation by Service, particularly for engineering teams, is straightforward. However, some services (such as EC2) are a mix of compute, storage and networking, so a finer- grained breakdown by usage type may be required to identify critical costs.

Allocation using AWS cost categories should be considered for complex configurations. Cost Categories allow you to create hierarchical rules for cost allocation and categories using several dimensions, such as account, tag, service, region or charge type.

Shared Cost Allocation

Once direct costs have been identified you may need to allocate shared costs. The shared costs may include both cloud costs and people or license costs. 

 There are multiple allocation strategies to choose from, depending on your business needs:  

Fixed - a number, which can be either $ or a percentage of the total shared costs. There are many possibilities, but the common options are listed. The fixed fee will generally be reviewed on an annual basis (or after major changes)

• Ratio of owners (e.g. 4 cost owners pay 25%)
• Agreed ratio, based on best guess and seen as ‘fair’ (e.g. 20%, 20%, 30%, 30%)
• Fee plus residual (e.g. $1000 for owner A, $2000 for owner B, and owner C picks up whatever is left)
• A complex (manual) analysis and calculated split based on previous years’ usage to be used for the next 12 months

Direct costs ratio - a split of the shared cloud costs based on the direct cloud costs. This works well if the shared costs are small compared to the direct costs. 

Technical ratio - a split based on a technical measure, either between directly identifiable resources or on some shared platform metric, e.g. no. of Vm hours, amount of storage, transactions count or web hits.

Business ratio  - a split based on the agreed cost owners’ business metric(s), such as the number of sales, sales revenue or profit in the period. This only works if all owners have the same ratio and it is seen as proportionate to the costs.

User ratio - a split of shared resources based on a user count. This only works if each owner has an identifiable user base. E.g. number of active users or registrations per month.

Stakeholder Engagement

It has been shown that it is important to ‘sell’ the process and get acceptance from your customers or stakeholders. Aim for both acceptance (perceived fairness) and effectiveness (relatable to the business drivers). When the shared costs increase compared to the directly attributable costs the acceptance can become more difficult and the metrics used are more important. 

Some key points to manage with stakeholders are: 

• Accuracy (measurability) - the shared costs should be divided based on the consumed usage as much as can be possibly measured, but you must also consider the cost of the measurement.
• Transparency and understandability - there should be a simple understanding of what is being charged for and its predictability for forecasting. This means that the process and formulae on how charges are split and allocated should be available and repeatable by the end user. This will tend to push towards a simplistic model.
• Controllability - how much influence can the owner bring to control and optimize the costs they are being charged.

Other factors to consider include: 

• Predictability - avoiding charge shock through early warning and alerting. 
• Accountability  - can the owner verify the number, i.e. can they determine the accuracy or fairness? 
  Cost efficiency - Is the shared cost optimized in the same way that direct costs may have been? This may put the onus on the central team to ensure cost-efficiency increases.
• Benchmark/comparability - does the shared costs “feel right”? This is usually compared to other direct costs and the owners’ experience of cloud costs in general.

However, it is also important to understand and manage the “cost of costing”, i.e. the cost of measuring and allocation, which must be a small proportion of the total costs.

Strategic Blue

As part of our rate optimization service, we support four different visualization methods to help our customers understand and allocate their costs accurately, efficiently and effectively.  Start your journey with Strategic Blue here.