This document summarizes a presentation about scaling analytics in a maturing organization. It discusses focusing initially on data infrastructure, integrity, access and visualization. As the organization grows, processes need to change from everyone accessing data as needed to assigning roles like analysts and business users. Getting buy-in for changes requires pre-research and collaboration. Solutions should be shipped as minimum viable products and improved iteratively. Empowering others involves creating transparent processes and frameworks for teams to self-govern requests. The overall goal is to start with basic functionality and expand the system as the organization matures.
4. Alyson Murphy - Moz - @AlysonMurphy
Alyson Murphy is the in-house Senior Data Architect at Moz.
She works with stakeholders to build a data solution that
help Moz make data informed business decisions.
Sean Work - KISSmetrics - @seanvwork
Sean Work runs the blog at KISSmetrics.com. He’s been with
the team since 2010. He loves working from Southern
California where he can surf, snowboard and camp under the
bright starts with no blankies.
5. Table of Contents
1 Section One – Some Context
2 Section Two –Where to Focus
Data Infrastructure
Data Integrity
Data Access and Visualization
Infrastructure Change Process
Data Utilization Process
People in your Organization
3 Section Three – Lessons Learned
Ways to Get Buy-in
Ways to Ship Solutions
Ways to Leverage and Empower Others
8. Introduction
Some Context
I work as an in-house data analyst and architect at Moz. We sell inbound marketing so"ware to help
people do be!er marketing. Moz is B2B so there will be many B2B examples but all of what I am
talking about can be applied in a B2C context.
9. Some Context
We have been growing a lot
We had just broken 100 employees when I started less than two years ago and now we are around
150. Scaling so quickly is exciting but can present challenges as well.
10. And Changing a lot
We re-branded and changed our name to
align with the changes in the industry.
We released a new flagship product to
serve the shi"ing needs of the industry.
Some Context
11. Our data has been changing too
Some Context
We consolidated key data that was routinely used for analysis onto one reporting server. We then funneled
key pieces of data into our web analytics solution so that there were fewer places to have to look for data
when it was time to do an analysis.
Tool Usage & Subscription Info
Moz.com Production Data
Moz Analytics Campaign Data
Moz email
Moz Local data
Reporting Server
Web Analytics Tool
12. Some Context
Our Process has Been Changing as Well
We used to use email to get and prioritize projects. We shi"ed to Trello which allows us to have
templates to ensure request quality and to be transparent about when certain projects will be worked
on.
Email Trello
13. Some Context
We aren’t there yet though
We’ve learned a lot of lessons se!ing us up to scale. Some were things we just needed to struggle with,
but there were some tips I knew two years ago. Hopefully some of them will help you and your team.
25. Where to Focus
Components of a Data System
There are 6 main areas of focus for building a successful and scalable data system. All parts are required
with equal focus to make the system work.
Data
Infrastructure
Data
Integrity
Data
Access &
Visualization
Infrastructure
Change Process
People in
your Org
Data
Utilization
Process
26. Where to Focus
Data Infrastructure
Consolidating your data sources will make analysis easier and quicker which is important when you
start adding people to your team. You want your team to be as efficient as possible.
The data infrastructure is tied closely with Data Access and Visualization. The backend choices you
make will affect the front end options that you have.
Tool Usage & Subscription Info
Moz.com Production Data
Moz Analytics Campaign Data
Moz email
Moz Local data
Reporting Server
Web Analytics Tool
27. Where to Focus
Data Integrity
Data Infrastructure and Data Integrity are perhaps the most important places to start because decisions
in these areas waterfall into the other areas of your Data System.
Spot Check
at
Time of Analysis
All Systems
Comprehensively
Checked
And Maintained
28. Where to Focus
Data Access and Visualization
Data Access and Visualization is key as your company starts to grow. The goal is to make the data as
easy to access as possible for people who have the skills to fish for their own data. You want to be able to
have business users be able to answer simple questions on their own so that your time can be be!er
spent on more complex business questions.
Types of Business Problems you Need to Consider
• Simple business questions for researching upcoming projects
• Complex business questions for researching upcoming projects
• Health Dashboards for monitoring to ensure nothing is broken
• Dashboards and Reporting for product/website releases and achievement of Key Results.
29. Where to Focus
Infrastructure Change Process
When you are in startup mode, everyone might have access to do what they need to do quickly to
implement the changes they need to make. In a small organization this works out because everyone
knows what everyone else is working on. When you start to grow though, this becomes less sustainable
because not every does not know about everything that is going on so they will not understand the
complete ramifications of their actions. Therefore a point person will need to be introduced to sign off on
changes.
30. Where to Focus
Data Utilization Process
When you are in startup mode, everyone uses whatever data they can get their hands on. Everyone is an
analyst. They might aggregate information from a dashboard and an analysis that was used to answer a
different business question in an effort to be as lean as possible.
However when you scale, it becomes more efficient to have analyst do the analytics work and pass on
the results to the business person. This is because the work will be done more quickly and completely. It
is also the best use of business person’s time to allow the analyst to do the work so they can focus more
on strategizing and managing their area of the business.
31. Where to Focus
People in your organization
As your business grows, there will be more of a need to level-up people in your organization so that
everyone can access the data and draw sound conclusions.
Level of
Complexity
Strategic
Tactical
Analysts Business Users Developers
Research potential
projects through
analysis.
Leverage complex
analyses to make good
business decisions.
Leverage the work of the
analysts in a Production
environment.
Builds and QA’s tools so
others can access data.
Pull basic requests
correctly.
Implement and maintain
the technical-side of the
Web Analytics Solution.
Wrangles tools to access
data. Ask Analyst to pull data. Structure Databases for
Production Use.
38. Lessons Learned
Data
Infrastructure
Data
Integrity
Data
Access &
Visualization
Components of a Data System
Infrastructure
Change Process
People in
your Org
Data
Utilization
Process
39. Lessons Learned
Data
Infrastructure
Data
Integrity
Data
Access &
Visualization
Components of a Data System
Infrastructure
Change Process
People in
your Org
Data
Utilization
Process
40. Lessons Learned
Data
Infrastructure
Data
Integrity
Data
Access &
Visualization
Components of a Data System
Infrastructure
Change Process
People in
your Org
Data
Utilization
Process
Buying
Decisions
42. Lessons Learned
Buying Decisions
Below is the typical process one might go through to pitch why a new tool should be bought.
1 Define problem
2 Lay out possible solutions
3 Research solutions
4 Make decision on solution
HAVE MEETING
HERE!
43. Lessons Learned
Buying Decisions
For the group, a 22 slide deck and 30 minutes was all it took to make the decision to buy a new web
analytics tool.
This is because a lot of research went into those slides. We made a well informed decision quickly.
44. Lessons Learned
Data
Infrastructure
Data
Integrity
Data
Access &
Visualization
Components of a Data System
Infrastructure
Change Process
People in
your Org
Data
Utilization
Process
Process
Changes
46. Lessons Learned
Process and Content Changes
Below is the typical process one might go through to pitch why a new process should be changed or why
we should look at KPI’s differently going forward.
1 Define problem
2 Lay out possible solutions
3 Research solutions
4 Make decision on solution
HAVE MEETING
HERE!
HAVE MEETING
HERE!
47. Lessons Learned
Process and Content Changes
In order to re-evaluate the KPI’s we look at, we had a collaborative meeting where each of the groups
came up with a dashboard. We then looked for areas where we needed to create alignment. A"er that, we
started building.
48. Lessons Learned
Data
Infrastructure
Data
Integrity
Data
Access &
Visualization
Components of a Data System
Process
Changes
Infrastructure
Change Process
People in
your Org
Data
Utilization
Process
50. Lessons Learned
Leveling Up
1 Be intentional about the resources you give people.
2 Provide in-person training for people who should
use data everyday.
3 Provide optional training for all.
52. Lessons Learned
Minimum Viable Product (MVP) vs. All-in-one
Do you want to ship as li!le as possible as soon as possible and learn and add versus shipping a totally
finished product all at once.
Phase and Benefit Minimum Viable Product All-in-one
Planning Phase
Must plan theoretical phases of
implementation, thought that may
all change as you go.
Must do more detailed planning
around what you are shipping since
you are shipping more stuff.
Implementation Phase
You may ship a dashboard in 5
phases with a step to re-evaluate
the rest of the plan between each
phase.
You ship it all at once. There will still
likely be optimization but it will be
small tweaks as opposed to huge
feature changes.
Benefit Allows you to ship sooner so people
have something to work with.
Allows for a smoother rollout to a
larger group because they will not
be lost in where in the
implementation phase the tool is.
A hybrid exists where you have only key stakeholders involved in the MVP-style iteration then do an All-in-
one rollout to the rest of the interested parties.
54. Lessons Learned
Ship your Minimum Viable Product and Learn
Plan, then ship an MVP in an environment that allows for quick creation, update manually for a few
updates where you learn what needs to change, change it, then automate in a more permanent solution.
Plan 1. Ship MVP and Learn 2. Automate 3.
56. Lessons Learned
Create a Transparent Process
Put all requests in a public place.
Have teams self-prioritize the list to save you time so all you have to do is work on the projects on the
top of the list.
List I used to Prioritize Lists Teams Prioritize
57. Lessons Learned
Enable Teams to Use a Framework
You may not have time run all of the processes that an analyst would in a larger organization. Create a
framework that the teams use to self-govern. This may not be a long-term solution but it will help free
up time to allow you to optimize other areas of your Data Analytics System.
58. Table of Contents
1 Section One – Some Context
2 Section Two –Where to Focus
Data Infrastructure
Data Integrity
Data Access and Visualization
Infrastructure Change Process
Data Utilization Process
People in your Organization
3 Section Three – Lessons Learned
Ways to Get Buy-in
Ways to Ship Solutions
Ways to Leverage and Empower Others