evidoria

← Back to browse

Good practice Imported

LOLA — AI administrative chatbot for student admissions at Universidad de Murcia

Spain · Murcia · See the Spain profile

Evidence: Observational / pre–post Top 82% 33/100 · Ask Evidence Copilot about this practice

NLP chatbot deployed June 2018 at Universidad de Murcia to handle 28,000 annual admissions queries. Achieved 91%+ accuracy across 74,000+ conversations with 25,000 students; freed staff from peak-period overload and provided 24/7 guidance to applicants.

28,000 queries/year
Admission-related queries handled annually before LOLA (pre-2018)
11,988 conversations
Conversations handled in first four-month pilot (June-September 2018)
4,183 students
Students reached in first four-month pilot (June-September 2018)
35,653 queries
Individual queries resolved in first pilot (June-September 2018)
93.4% accuracy
Accuracy rate in first pilot (June-September 2018)
74,000+ conversations
Cumulative conversations (by 2019)
25,000 students
Cumulative students reached (by 2019)
94.2% accuracy
Accuracy rate (by 2019)
LOLA — AI administrative chatbot for student admissions at Universidad de Murcia

Details

Maturity
Established
Promoter
Universidad de Murcia
Period
2018–present
Region (NUTS)
ES62
Keywords
higher education, chatbot, admissions, student services, NLP

Context

LOLA is a natural language processing chatbot built by 1MillionBot and deployed by Universidad de Murcia in June 2018. The university's Student Information Service (SIU) handled around 28,000 admission-related queries each year, concentrated into short summer peaks when EBAU exam results, cut-off grades and pre-registration deadlines triggered a flood of student contact.

Objectives

LOLA was built to absorb routine admissions queries around the clock, so temporary summer staff would no longer be needed to manage the surge and students who could not reach the office during business hours could still get answers.

Activities

Built on Google Dialogflow, LOLA operates 24/7 and answers questions about entry scores, score reviews, enrolment procedures and campus services, and the university describes it as learning continuously from each interaction.

Results

In its first four-month pilot (June–September 2018), LOLA handled 11,988 conversations from 4,183 students, resolving 35,653 individual queries at a 93.4% accuracy rate. By 2019, cumulative use reached 74,000+ conversations with 25,000 students, with accuracy rising to 94.2%.

Conclusions

Universidad de Murcia describes LOLA as the first AI administrative assistant deployed at a Spanish public university, and it has been cited as a case study by Fordham University Law School and other academic publications on AI in higher education; staff report being freed to focus on non-routine cases while LOLA absorbs routine queries.

Implementation
Indicative cost
Medium (€50k–€500k) — Vendor-built NLP chatbot (1MillionBot) on Google Dialogflow, replacing the need for temporary summer admissions staff; no public budget figure disclosed.
Time to results
Long (> 3 years) — Deployed June 2018 and in continuous operation to the present (2018-present).
Staffing & skills
1MillionBot (chatbot vendor/developer), Universidad de Murcia Student Information Service (SIU) staff

Conditions for success

  • Built on an established NLP platform (Google Dialogflow) rather than bespoke infrastructure
  • Deployed ahead of predictable seasonal demand peaks (EBAU results, pre-registration deadlines)
  • Continuous learning from interactions allowed accuracy to improve from 93.4% to 94.2% over the first year

Common failure modes

  • No published algorithmic audit, bias assessment, or student-data minimisation policy despite operating under GDPR
  • No documented replication at other Spanish universities to date

Where it fits

Governance type
single public university administration
Scale
one institution of roughly 30,000 students
Income level
high-income (Spain)

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful