Conclusions and recommendations

Six conclusions about the Mexican legal market

1. Maturity is built along different paths depending on the audience. Law firms and legal departments have similar maturity levels, but they reach it in different ways. In law firms, what most accompanies a high index is the team's knowledge. In legal departments, what weighs more is a set of management discipline (effective tool usage, governance and orderly measurement) rather than a single factor. Firms mature through knowing. Corporate departments, through orderly doing.

2. Efficiency and client experience go hand in hand. It is the strongest association in the study. Organizations that feel more efficient also report more satisfied clients. We cannot state which causes which with these data, but both move together and probably reinforce each other.

3. There are shared professional convictions. Human oversight of AI work, the centrality of human connections and resistance to change as a barrier are perceptions that do not distinguish advanced organizations from those just starting out. They are values of the profession, not signals of maturity, and that is why they are analyzed separately from the index.
4. Enthusiasm for AI does not correlate with its governance. Perceiving benefits from AI does not go hand in hand with having formalized ethical practices, especially in law firms. This does not mean that firms have worse practices than departments (on the contrary, they assess ethical risks more frequently), but that perceived benefits and governance advance separately. In a context of emerging regulation, this deserves attention.
5. Digitalizing is not the same as modernizing. The percentage of digitalized documents does not accompany greater efficiency, a better client experience or a higher index. Scanning documents without transforming processes might not produce perceptible results.

6. In law firms, knowledge accompanies maturity more than frequent training. The depth of the team's understanding is related to maturity more than the mere frequency of training sessions. Understanding matters more than accumulating course hours.

Study limitations

Additional limitations:

1. Small, non-probabilistic sample. With 72 law firms and 39 legal departments, and participants recruited through the networks of the convening organizations, the sample likely overestimates the market's effective adoption: those connected to legal technology providers tend to be more digitalized than average.
2. Self-reported, single-point data. All measurements are perceptions, not objective metrics (billing, turnaround times, productivity), and were taken at a single point in time (June–November 2025). That is why, throughout the report, we speak of variables "going hand in hand" and not of one "causing" the other: it is not possible to establish causality with this design.
3. Partial coverage of hypotheses. Of the 56 hypotheses in the original mapping, only 17 could be formally tested (one more was left without a result due to insufficient sample). It is a relevant reduction in scope worth keeping in mind when reading the findings.
4. The Index is an internal tool, not a standard. Being a sum of components, its parts relate to the total by construction. That is why, when we compare which factors explain it, we always use the corrected version that excludes each component from its own total.
5. The AI adoption rate depends on an imputation criterion. Counting as users those who left the question blank but declared using AI tools raises the reported rate. Without that imputation, the rates drop several points (see the AI chapter).

The rest of the methodological detail, the statistical tests and the literature review are in the methodology appendix.