Clients Application Usage vs Clients Bandwidth Usage

[Yaakov Neuman at NetOp AI]

I am reaching out regarding a discrepancy I've observed between two API endpoints:

1. getNetworkClientsApplicationUsage: https://developer.cisco.com/meraki/api-v1/get-network-clients-application-usage/

2. getNetworkClientsBandwidthUsageHistory: https://developer.cisco.com/meraki/api-v1/get-network-clients-bandwidth-usage-history/

I am using the same t0 and t1 parameters (covering a 2-hour window) for both endpoints. When I compare the summarized bandwidth usage history against the total application usage for each client, I see a significant gap. In most cases, the sum of application usage is approximately 30% higher than the reported client usage.

Could you please explain why these figures would differ so significantly? Below are a few examples of the data I am seeing:

Client usage    : k97268c : {'received': 8334203, 'sent': 171219}
Application sum : k97268c : {'received': 10899393, 'sent': 224441}
================================================================================

Client usage    : kb2be72 : {'received': 347, 'sent': 394}
Application sum : kb2be72 : {'received': 336, 'sent': 383}
================================================================================

Client usage    : kc56021 : {'received': 685063, 'sent': 30746743}
Application sum : kc56021 : {'received': 865825, 'sent': 38833513}
================================================================================

Client usage    : ke36f13 : {'received': 13874177, 'sent': 345362}
Application sum : ke36f13 : {'received': 18269226, 'sent': 453685}
================================================================================

Client usage    : k0792bc : {'received': 7571593, 'sent': 146228}
Application sum : k0792bc : {'received': 10415457, 'sent': 200971}

[pieterh]

be aware that you are comparing counters of different magnitude/units
Get Network Clients Application Usage           - in kilobytes
Get Network Clients Bandwidth Usage History - in megabits per second   NB! for all clients !

[Yaakov Neuman at NetOp AI]

Thanks. I missed the units.

But still have a question about the usage history.

The duration of intervals are not equal. The diff between the samples are 6m, 7m or 1m.

{'ts': '2026-05-19T07:33:00.000000Z', 'received': 2, 'sent': 2}
{'ts': '2026-05-19T07:39:00.000000Z', 'received': 75, 'sent': 23}
{'ts': '2026-05-19T07:45:00.000000Z', 'received': 2, 'sent': 2}
{'ts': '2026-05-19T07:52:00.000000Z', 'received': 0, 'sent': 0}
{'ts': '2026-05-19T07:53:00.000000Z', 'received': 2, 'sent': 2}
{'ts': '2026-05-19T07:59:00.000000Z', 'received': 6, 'sent': 8}

What is the meaning of that?

Is there way to calculate the amount of actual data the client used?

[aleabrahao]

You’re basically comparing:

Client bandwidth usage = how much data actually passed through the network
Application usage = how much data Meraki thinks belongs to each app

Application usage is an approximation for visibility, not an exact accounting of bytes, so it will usually be higher than the real client usage.

Traffic Analysis and Classification - Cisco Meraki Documentation