Build your career around helping an organization learn from evidence, make sound decisions and follow through on what it changes.
If you work in social impact data, build your career around helping an organization learn and act on evidence. Producing accurate reports remains part of the job. Your wider contribution is to explain what the organization is learning, which decisions the evidence supports, and what needs to change. As AI takes on more of the preparation work, I would invest in that responsibility. This applies to analysts, evaluators, impact measurement and management practitioners, and consultants—although their authority and methods differ.
Download the presentation used in the video. The examples illustrate the skills behind this career shift.
My advice is to become someone a team trusts to explain what it knows, what it does not know, and what that means for its choices. That is a broader responsibility than delivering a dashboard or completing a funder report.
A report records results. Learning changes the team's understanding of those results. A decision commits someone to a course of action. Following up shows whether that action helped. Your career can develop by taking responsibility for more of this work, with clear boundaries around who makes the final decision.
These are connected responsibilities, not a ladder of job titles. An early-career analyst can identify a misleading comparison that changes a recommendation. A senior measurement lead can produce extensive reporting that nobody uses. Seniority alone does not establish the value of the work.
| Responsibility | Question it answers | Useful contribution |
|---|---|---|
| Reporting | What happened? | An accurate account of results, with definitions and limitations. |
| Learning | What do we understand differently? | An explanation checked against evidence, including what remains unknown. |
| Decisions and action | What should change, and who can act? | Options and tradeoffs, an agreed choice and someone responsible for it. |
| Review | What happened after the change? | A check on the consequences and what the team should reconsider. |
Reporting also serves accountability, transparency and compliance. Keep doing it well. The career question is whether your contribution ends when the report is sent.
AI can assist with preparing material: drafting a framework, organizing text, writing a query and preparing an explanation. The amount it can safely handle depends on the task, the information supplied and how the result is checked. My career recommendation is to use those capabilities to make room for investigation and judgment.
If you use AI only to produce more reports, you may increase output without improving a single decision. The more valuable opportunity is to notice a problem sooner, examine more than one explanation and help the team respond while the evidence is still useful.
That does not make every practitioner a manager. The practitioner establishes what the evidence supports and explains the options. The person responsible for the program, funding or investment decides within their authority. People affected by that decision need a meaningful voice in it.
This direction predates generative AI. Utilization-Focused Evaluation explicitly centers the people who will use findings and how those findings will be used. In impact investing, Principle 8 of the Operating Principles for Impact Management connects reviewing results with improving decisions and management processes. AI offers new ways to support that work; it does not replace its purpose.
Learn how the work happens before deciding what its data means. Who is the program reaching? What do participants experience? Where do staff make difficult choices? Which results take time to emerge? A change in a chart may reflect a change in delivery, a change in who participated or a change in how someone entered the records.
This is where program experience becomes valuable in a data career. Someone who understands the service can spot an explanation that looks plausible in a report but makes no sense in practice. Build enough analytical skill to examine that concern, and enough subject knowledge to know when to raise it.
Move closer to data collection. Help decide what needs to be learned, when people can reasonably answer and how their responses will connect over time. A good collection process captures useful experience without repeatedly asking people for information the organization already holds.
Attendance, case notes, surveys and follow-up each reveal something different. Learn how to bring them together without losing the person, the context or the timing. This responsibility involves listening and coordination as much as software. If you only receive a spreadsheet after collection has ended, many of the most important problems may already be difficult to fix.
Learn to distinguish a pattern worth investigating from a conclusion strong enough to act on. Ask whether the comparison is fair, whose experience is missing and what other explanation could fit. Understand what the method can establish before recommending what to do.
The strength of evidence needed also depends on the decision. Trying a reversible change to a session reminder is different from ending a service or claiming that a program caused an improvement. Learn when routine analysis is sufficient and when you need an evaluator, statistician or other specialist.
Finding a data mismatch is useful because of its consequence. You may prevent a team from scaling a program based on an overstated result, or from cutting a service because follow-up records are incomplete. That is the career value to explain, beyond “I cleaned the dataset.”
Practice discussing findings with the people who will use them. Explain the options, the evidence for each, the uncertainty and the practical tradeoffs. Invite people to challenge the interpretation. Be able to say that the available evidence does not justify a decision yet.
A useful recommendation also needs an owner, resources and a realistic time to act. If those are absent, a better chart will not solve the problem. Learn how to identify these barriers without presenting yourself as the person entitled to make everyone else's decisions.
Stay involved long enough to find out whether the recommendation was used and what happened afterward. Sometimes the team learns that the original explanation was wrong. Sometimes an improvement for one group creates a difficulty for another. Those findings should shape the next decision.
For a career portfolio, this creates a stronger account of your work: the problem you investigated, the evidence you checked, the choice it informed and the result of reviewing that choice. Report the limits honestly. You do not need to claim sole credit for an outcome to demonstrate your contribution.
Spend less of your development time on repeated production work that your tools can already handle reliably. Put more time into the reasoning and relationships that make the output useful.
You still need enough technical and methodological understanding to check the work. Delegating a calculation does not remove responsibility for its meaning. Choose your depth: a specialist evaluator, a database analyst and a program learning lead will need different training.
The shared direction is better use of evidence. The responsibility you take on should fit your role.
For an investment-focused path, Impact Frontiers' curriculum connects impact management with investment decisions and impact-financial integration. It is one example of learning that needs to fit the decisions your role supports. For organizations choosing external support, our social impact consulting guide covers services, selection and handoff.
Consider this illustrative situation: a youth program sees lower participation at one location. The reporting task is to calculate attendance correctly and explain the change. The broader learning responsibility is to establish whether the program has a participation problem, a recording problem or both.
The practitioner checks the records with staff, asks participants about their experience and compares plausible explanations. Suppose that investigation supports a scheduling problem. The practitioner explains the evidence and its limits, then works with the program lead to consider an alternative time. The lead decides whether to try it and takes responsibility for delivery.
The practitioner then checks whether participation improves and whether the new time makes access harder for anyone else. If the result does not support the explanation, the team revises its understanding. The professional contribution extends from an accurate count to a better-informed choice and a review of its consequences.
The same principle applies to a funder comparing partners. A lower reported outcome may reflect different populations or definitions. The practitioner helps establish whether a comparison is defensible before it influences funding. “We cannot make that comparison yet” can be a valuable finding.
You cannot create a learning organization solely by improving your own skills. Your role needs access to the people who collect information, the people it represents and the people who make decisions. You also need time to investigate findings and permission to surface uncomfortable results.
When considering a role or consulting assignment, ask what happens after a report is delivered. Who discusses it? Can findings change a program or budget? Are practitioners involved before the questions are finalized? Is there capacity to act, or will the same person simply be asked to produce more reports?
A team may still need separate collection, analysis and evaluation specialists. The important change is that they work together early enough to correct problems. If every question travels through a long chain of handoffs, AI at the final reporting stage will not repair what was never collected.
I would also ask employers to broaden how they assess these roles: include the quality of evidence, risks identified, clarity of recommendations and usefulness to decision makers. Avoid judging an analyst solely by whether outcomes improved; implementation, resources and external conditions are not all under their control. A defensible decision to wait, retain a service or gather more evidence may be the right result.
Use one existing piece of work to demonstrate a wider contribution. If you already produce reports, choose one recurring finding that deserves investigation. If you work in programs, identify a question your experience suggests the current records cannot answer. If you consult, agree on one use of the findings and include time to support it.
Build a short account of that work: what the team needed to understand, what you checked, what changed in its understanding and which decision followed. Include the evidence that challenged your initial view. If the work has not yet led to action, say why and explain the next step.
Use that example to choose your next learning investment. A weakness in study design calls for methods training. Difficulty interviewing participants calls for qualitative research practice. A database constraint may call for SQL. Difficulty getting findings used may call for facilitation, program knowledge or a clearer agreement with leadership. The next course should address a real gap in the responsibility you want to take on.
A platform can support this way of working when collection, records and analysis remain connected. It can reduce the repeated preparation needed to revisit a question as new information arrives. It cannot establish the organization's learning priorities or take responsibility for its decisions.
Sopact's approach brings data collection and AI-assisted analysis into a connected workflow. For the practitioner, the relevant skill is designing that workflow: what should be collected, how records connect, who can access them and how an answer can be checked. The platform supports the work; people remain responsible for the interpretation and response.
No. Reliable reporting supports accountability and provides a record of what happened. The career opportunity is to help people use that information as well. This may mean investigating an unexpected result, making uncertainty visible or helping a team review a choice. Automation should preserve reporting quality while creating capacity for those responsibilities.
Not necessarily. Start with the scope of your current role and the authority you have. An analyst can contribute to a decision without owning it. Ask for involvement in interpreting findings and reviewing their use. If a prospective role offers only production work, consider whether that matches the experience you want to build.
Yes, some roles emphasize program knowledge, research, measurement design or helping teams use evidence. Others require substantial programming or statistical expertise. Read the responsibilities of the roles you want and learn accordingly. You still need to understand definitions, comparisons and limitations when a platform or AI performs the analysis.
Learn to choose appropriate tasks, provide the relevant context and verify the output. Practice identifying omissions and unsupported explanations, not just writing prompts. Understand your organization's rules for using participant information. Then examine whether the time saved is helping you investigate important questions or simply increasing the amount of material produced.
Show how your work helped a client understand a problem, assess its options or build a reliable way to learn. Explain your contribution, the client's responsibilities and the limits of the evidence. Agree on suitable follow-up rather than promising an outcome you cannot control. A useful handoff includes the ability to maintain and question the work.
No. Evidence informs a choice, but decisions also involve priorities, resources, obligations and the perspectives of people affected. Your contribution is to make the evidence and tradeoffs clear, including uncertainty. Do not disguise a value judgment as a calculation or treat an AI recommendation as the authority to act.
Bring a reporting process your team wants to improve, the evidence behind it and the decisions it needs to support. Explore how to keep collection and learning connected.
Explore connected data collection
