In this section
5.1OverviewFour categories account for most student success technology. They share goals, and they share risks.Key takeaway. Every tool embeds a theory of student success, whether or not anyone wrote it down.5.2Learning analyticsMost learning analytics draw on LMS activity: logins, page views, submissions, discussion posts, and time-on-page.Key takeaway. LMS data measures the quantity of involvement. Do not mistake it for quality, or for belonging.5.3Predictive analytics and early-alert systemsPredictive models estimate which students are likely to stop out, fail a course, or miss a milestone, and route them to outreach.Key takeaway. The value of a predictive model lies in the human response it triggers, not in its accuracy.5.4AI for advising, chatbots, and personalized supportConversational AI can answer routine questions at any hour, send timely nudges, and free advisors for relationship work, if it is designed to hand off to people.Key takeaway. Measure AI support by how often it connects students to people, not by how many conversations it deflects.5.5Data governance, ethics, and responsible AIGovernance aligns what a tool does with what the institution says it values about students.Key takeaway. If students would be surprised by how their data is used, governance has already failed.5.6Generative AI as a distinct categoryPredictive tools estimate who may need help. Generative tools produce work, advice, and conversation. They raise different questions.Key takeaway. For each generative AI use, ask whether it strengthens the process the outcome is supposed to measure, or let students skip it?5.7Other student success technologiesStudent success technology extends well beyond analytics and early alert. Each category touches different theoretical mechanisms.Key takeaway. Name the mechanism before choosing the category. Two tools in the same category can support entirely different theories of success.5.8From theory to enterprise architectureEach theory names a mechanism. Each mechanism depends on specific campus capabilities, and those capabilities run on specific systems.Key takeaway. If a theory’s mechanism has no owner, capability, or system behind it on your campus, a new tool will not supply one.