Today I’m attending the Analytics and AI summit put on by the HighEdWeb association. It is an online summit featuring various speakers across the fields. This is just a place for my notes.

Leverage AI and Analytics to Supercharge Your Content

Brian Piper — University of Rochester
https://brianwpiper.com/getmyslides/

We all need to create incredible content, put it out on the right channels, and monitor the data to discover the insights to make it as effective as possible.

Notes:

• Recommends sing multiple AI generators at the same time to verify the content and quality
• Creating personas is important (custom GPT personas)
• Tools
  • Chat GPT
  • Dall-E, and Midjourney
  • Captioning: descript and capsho
  • Video tools: GlossAi, OpusClip
  • runway – text to video
  • HeyGen – Animate images with your voice and body motion
    • He uses it to translate the video to various languages)
    • “I don’t speak any of those languages and it took maybe 3 minutes to generate 3 new languages” (brings up questions for me about creating inaccurate content)
  • Optimizing based on data
    • Used data from UA, GA4, and other sources
    • First combined in looker to easily export data
    • Promoted AI to find content that could be removed
      • No traffic, bad ranking (though is seems like this could be done by just sorting the sheet)
      • ChatGPT suggested merging the files with database joins. Was able to perform the merge with database prompts directly in ChatGPT.
    • Looker template for search console (in presentation files)
      • ChatGPT can be prompted to ingest this data and creat new story ideas.
      • The personas can then be used to write articles or SEO suggestions.
• It’s up to us to decide what tasks can be automated and which tasks need human intervention.

Turning data into value: how to make your metrics matter to senior leadership

Matt Herzberger — VisionPoint Marketing

We will delve into the crucial task of aligning marketing metrics with the business bottom line from a leadership viewpoint.

Notes:

• Marketing is a driver for higher Ed and only data can show the success story
• Leadership make decisions on data instead of vibes
• Key players to engage with includes Institutional Research along with the upper administration
• Move marketing from the “creative team” to the revenu generators
• Determining the right metrics
  • Tie impact to goals
  • Connect 2 points not 25 (same with the number of slides)
  • Goal values
    • Per user goal value = goal value / users
    • User conversion rate = goal completions / users

Bringing the human side and analytics together: rebuilding and shrinking your website

Jessica Barch — University of Waterloo

Knowing where to start when planning a website overhaul can be overwhelming, especially when you’re trying to determine what content to keep, and what content to get rid of.

Notes:

• Website – new backend and new brand (300 page site)
• Lots of duplicate pages that outlined the same topics, but in different ways
• Overwhelmed by the amount of duplicate content
• Need to look at at least a year of data

Botched Chat: A Comedic Tale of AI Misadventures

Joshua Charles — Rutgers University-New Brunswick

During the height of the COVID-19 pandemic in 2020, and as we all pivoted to remote learning and work, our school began exploring innovative ways to support and engage our students remotely. Enter chatbots.

Notes:

• Horrible use rates and feedback. Average ratings (out of 5):
  • Advising bots: 2.7, 2.6
  • Career management: 3.7
  • Supply chain management: 3.4
• Misidentifying the core issue:
  • Faculty/staff defined the issues:
    • Students can’t find key information
    • Departments are bogged down with the same questions over and over
    • This shows the real issue is communication and culture
  • Misunderstanding the tool dependencies
  • Limited buy in from stakeholders

Higher Ed’s GA4 Challenges: Solutions for Common Problems

Cassandra Golda — SimpsonScarborough
Adam Mothersbaugh — SimpsonScarborough

Transitioning to Google Analytics 4 (GA4) is a critical step for higher education institutions aiming to leverage data-driven insights for enhancing digital experiences.

Notes:

• Best practices when setting up
  • GA4 properties can’t communicate with each other
    • Recommend global GA4 property for all subdomains
  • Recommend a single data stream per property
  • Enhanced measurements
    • Unfortunately this automation does not lead to data accuracy
    • Insead, set up exact desired events in GTM
      • Can use custom event parameters and custom dimensions
      • But, predefined event parameters automatically have predefined dimensions
  • Segment internal traffic
    • Unfortunately ip address rules are too complicated
    • Instead use “audience feature”
      • If they visit the portal we can consider them an internal user

Optimizing Data Collection and Capability: Server-Side Tagging & GA4 Modeling

Jonathan Carroll — Arizona State University ( ASU )

Learn about GTM’s server-side tagging for better data control and combine it with GA4’s modeling to uncover actionable insights even in a privacy-centric future.

Notes:

• Running server-side GTM (as a proxy) gives you faster pages and more control over the data
• Browsers do not speak directly with 3rd-party vendors
  • Reduce risk of endpoint hijacking
  • Reduce risk of cross-site scripting
  • Reduce risk of sending unintended data via HTTP headers
• Overview
  • Client-side GTM calls library from GTM GCP cloud
  • Data sent to GCP cloud to be sanitized or enhanced
  • Data passed on to advertisers and analytic collectors
• Reduced load time by 25% by offloading gtm ad tech (pixels) to the proxy
• SEO position improved 18%

One Dashboard to Rule Them All

Alan Etkin — British Columbia Institute of Technology

This presentation is for dashboard evangelists. It’s for folks who get a kick out of identifying data worth sharing, and figuring out techniques on how to share it with style. While the concepts are platform-agnostic, I’ll be focusing on what you can do with Looker Data Studio.

Notes:

• When switching to GA4 they needed to know which dashboards were worth keeping
• You can track Looker dashboards just like you track a normal website
  • The URLs are useless (pape title is better)
  • But GA4 isn’t great – there is no drilldown
• Looker data in looker?
  • Pull it in as a data source
  • Clean up titles and pull up counts for each page (use regex and sql)
  • Clean up urls to remove redundant users data
• Quota limits!
  • We have hit the token limit too
  • Unless you pay, this may mean just waiting until the top of the next hour
  • Be sure to not publicly share the Looker URLs
  • Scheduling emails can help. It includes a helpful picture of the important data and most people don’t even bother to click the link to the real dashboard.
    • You can also stagger the delivery to help avoid the limits
• You can create a dashbord of dashboards for each area of the college
• And the one “dashboard to rules them all”
  • A looker dashboard that shows the usage of all the other dashboards