Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
moesif
What Does API Monitoring Mean for
Product Managers?
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Who am I?
● Co-founder and CEO of Moesif,
an API analytics platform
● Focus on API strategy, platform
growth, and analytics
● I love IPAs!
derric@moesif.com
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Lots of infrastructure level metrics for
engineering goals
Uptime
and SLA
Resource
Usage
Average
Latency
Request
Per Minute
Errors
Per Minute
1 2 3 4 5
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
However, product metrics should be aligned
to one of three product/business goals:
Adoption
(Evangelism)
Engagement Retention
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Similarly, many top of funnel metrics, but are
these the best for product success
Page views Sign ups/
Sign ins
Requests Per
Minute
Average
Latency
Visits to a landing page is
not engagement
Can be faked by unqualified
leads signing up
Could be artificially high
from simple health probes
A consistently slow
API can be fine
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Adoption: Map out your funnel to find
barriers to adoption
The developer funnel tracks a user’s journey from initial sign up
to shipping a working app using your API
Pre-Integration
Test/Sandbox
Production
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
What to measure?
Unlike the mobile app acquisition funnel, an API user could stay
in a single stage for days. You should track both:
Conversion rate
for each step
Time to reach
next step
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Track both conversion rate and
time to next step
Sign Up First API Call First Working App
30%
2 days
30%
6 hours
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Pre-integration stage: Get them to sign up
Content, conferences,
and 1:1 meetings
Stellar and up to
date docs
Paid ads on developer
focused networks
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Sandbox stage: Get them to their first hello
world
TTFHW (Time to First Hello World)
is the time a new developer takes to:
Sign up
Create a simple test app
Make first transaction and
verify API fits their needs
1
2
3
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Lack of docs/simple onboarding slows
TTFHW
Ambiguous or lengthy
onboarding
No framework for their
environment
Lack of documentation
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Production stage: Get them to ship a
working app
TTFWA (Time to First Working App)
or TTFPA (Time to First Paid App)
is the time a new developer takes
to ship a fully working solution
to their users.
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Internal stakeholders slow TTFPA
Legal and
compliance risk
Project
priorities
Functional &
performance
testing
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Let’s create example funnel goals for Algolia
Pre-integration stage:
Developers who signed-up and viewed onboarding docs or their API key
Sandbox stage:
Developers who completed at least:
One POST /v1/indexes operation AND
One POST /v1/indexes/{indexName}/query operation
WHERE response.headers.Content-Length > 0 .
Production stage:
Developers who completed at least:
1,000 POST /v1/indexes/{indexName}/query operations
WHERE response.headers.Content-Length > 0 .
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
API Engagement: Weekly active tokens can
be a good north star metric
The more accurate version is Weekly Active API Users.
This is because each user can create multiple keys, such as for dev and prod.
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Once you define a north star metric,
understand what’s driving it
Which SDKs have
higher or lower
integration rates?
Which marketing
channels drive the most
integrations?
2%30%20%2%
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
With proper attribution, you can measure the
effectiveness of acquisition channels
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
It’s also important to look at API volume such
as which partners are most active
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
For most APIs, inconsistent performance is
worse for users than a high average latency
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
API Retention: What causes customers to
become loyal?
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Programing languages
or SDK
API versions Features used
Digging into your API Retention
Find areas of improvement by grouping retention curves by:
V 3.281.32
V 2.812.93
V 1.472.48
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved
Finding SDK issues via retention curves
Confidential and Proprietary. © 2020 Moesif, Inc. All Rights Reserved23
Derric Gilling

What Does API Monitoring Mean for Product Managers?

  • 1.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved moesif What Does API Monitoring Mean for Product Managers?
  • 2.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Who am I? ● Co-founder and CEO of Moesif, an API analytics platform ● Focus on API strategy, platform growth, and analytics ● I love IPAs! derric@moesif.com
  • 3.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Lots of infrastructure level metrics for engineering goals Uptime and SLA Resource Usage Average Latency Request Per Minute Errors Per Minute 1 2 3 4 5
  • 4.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved However, product metrics should be aligned to one of three product/business goals: Adoption (Evangelism) Engagement Retention
  • 5.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Similarly, many top of funnel metrics, but are these the best for product success Page views Sign ups/ Sign ins Requests Per Minute Average Latency Visits to a landing page is not engagement Can be faked by unqualified leads signing up Could be artificially high from simple health probes A consistently slow API can be fine
  • 6.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Adoption: Map out your funnel to find barriers to adoption The developer funnel tracks a user’s journey from initial sign up to shipping a working app using your API Pre-Integration Test/Sandbox Production
  • 7.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved What to measure? Unlike the mobile app acquisition funnel, an API user could stay in a single stage for days. You should track both: Conversion rate for each step Time to reach next step
  • 8.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Track both conversion rate and time to next step Sign Up First API Call First Working App 30% 2 days 30% 6 hours
  • 9.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Pre-integration stage: Get them to sign up Content, conferences, and 1:1 meetings Stellar and up to date docs Paid ads on developer focused networks
  • 10.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Sandbox stage: Get them to their first hello world TTFHW (Time to First Hello World) is the time a new developer takes to: Sign up Create a simple test app Make first transaction and verify API fits their needs 1 2 3
  • 11.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Lack of docs/simple onboarding slows TTFHW Ambiguous or lengthy onboarding No framework for their environment Lack of documentation
  • 12.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Production stage: Get them to ship a working app TTFWA (Time to First Working App) or TTFPA (Time to First Paid App) is the time a new developer takes to ship a fully working solution to their users.
  • 13.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Internal stakeholders slow TTFPA Legal and compliance risk Project priorities Functional & performance testing
  • 14.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Let’s create example funnel goals for Algolia Pre-integration stage: Developers who signed-up and viewed onboarding docs or their API key Sandbox stage: Developers who completed at least: One POST /v1/indexes operation AND One POST /v1/indexes/{indexName}/query operation WHERE response.headers.Content-Length > 0 . Production stage: Developers who completed at least: 1,000 POST /v1/indexes/{indexName}/query operations WHERE response.headers.Content-Length > 0 .
  • 15.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved API Engagement: Weekly active tokens can be a good north star metric The more accurate version is Weekly Active API Users. This is because each user can create multiple keys, such as for dev and prod.
  • 16.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Once you define a north star metric, understand what’s driving it Which SDKs have higher or lower integration rates? Which marketing channels drive the most integrations? 2%30%20%2%
  • 17.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved With proper attribution, you can measure the effectiveness of acquisition channels
  • 18.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved It’s also important to look at API volume such as which partners are most active
  • 19.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved For most APIs, inconsistent performance is worse for users than a high average latency
  • 20.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved API Retention: What causes customers to become loyal?
  • 21.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Programing languages or SDK API versions Features used Digging into your API Retention Find areas of improvement by grouping retention curves by: V 3.281.32 V 2.812.93 V 1.472.48
  • 22.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved Finding SDK issues via retention curves
  • 23.
    Confidential and Proprietary.© 2020 Moesif, Inc. All Rights Reserved23 Derric Gilling