Cost Visibility Quick Start

this quick start walks you through a complete flow for understanding usage based costs in amberflo by the end, you will have a meter capturing usage a cost rate defining how usage translates into cost a workload representing internal infrastructure spend data visible in dashboards the goal is to get a simple, working cost visibility loop in place what you’ll build to keep this concrete, assume a simple use case you want to understand the cost of an internal service, such as a text processing pipeline , based on usage you will track usage with a meter define cost rates for that usage attribute usage to a workload view resulting spend in dashboards step 1 create a meter meters define what usage you are tracking navigate go to cost → configure → meters click create meter (top right) configure the meter label use a descriptive name (for example api requests or token usage) api name auto generated from the labeluse the default unless needed otherwise no spaces allowed letters, numbers, hyphens, underscores max 50 characters description (optional) meter type select count add dimensions (optional but recommended) dimensions describe your usage in more detail and are critical for cost modeling examples model operation type region if you are tracking token usage, typical dimensions would be model token type (input, output) by default, workload is included you can leave it as is create the meter click create step 2 define cost rates now you will define how usage translates into cost navigate in the meters list, select your meter go to the rate tab click create meter rate (far right) configure the rate label example internal cost rate description (optional) choose a rate model you have two main options per unit every unit of usage has the same cost example every request = $0 01 this is simple but often limited per unit with dimensions (recommended) costs vary based on dimension values this is the standard approach when costs differ by factors like model, operation type, or region build the rate matrix each column represents a dimension each row represents a unique combination of dimension values each row has its own per unit cost example if your dimensions are model token type then each row might look like model = x, token type = input → cost a model = x, token type = output → cost b model = y, token type = input → cost c configure the matrix remove any dimensions you don’t need (for example, remove workload if not used in pricing) use add row to define combinations use add column only if needed for each row fill in dimension values enter the corresponding cost per unit notes this can also be configured via api or csv import you do not need to complete every possible combination to get started click save once done you now have a cost model for your usage step 3 create a workload workloads represent internal systems you want to track costs for examples text processing pipeline recommendation engine background job system api service navigate go to the workloads page click create workload configure name example text processing pipeline workload id any unique identifier click save you now have an internal entity to attribute costs to step 4 send usage (attributed to a workload) now you will send usage tied to your workload navigate go to cost → meters find your meter click event upload submit usage fill in customer leave blank workload select the workload you created value enter a number (for example 1000) dimensions provide values that match your rate matrix(for example model = x, token type = input) click ingest meter you have now sent usage tied to your workload step 5 view spend navigate go to cost → spend (top item) what you’ll see spend month to date spend quarter to date spend year to date since this is your first data all three values should be the same under overview you will see top spend by workload your workload should appear here top trending workloads your workload should also appear here top spend by customers likely empty unless usage was also tied to a customer what you just did you completed a full cost visibility workflow defined what to measure (meter) defined how cost is calculated (rate) defined where cost is attributed (workload) sent usage data viewed resulting spend this is the core loop for understanding usage based costs in amberflo what’s next this was a minimal setup from here, you can expand into more detailed dimension models multiple workloads and hierarchical attribution automated ingestion pipelines combining cost data with customer billing or move back to the customer billing quick start to connect cost and revenue into a single system