After a Holland Code test, pick one activity you’d like to do more often. Find two postings with the same job title, compare how each describes that activity, then ask for a recent example or try a small version of the work. Cleveland and Colorado both advertised for a Data Analyst. One posting emphasized city dashboards and open data. The other emphasized statistical analysis for Medicaid. For someone who likes investigating why a number changed, the next step depends on who actually gets to investigate and revise the work.
The postings and public materials are real. Maya, her decisions, the feedback, and the practice data are fictional.
A job-posting check you can use right away
Copy the blank row and start with one duty. The completed row shows how Maya would investigate a real Cleveland posting. She’s deciding which kind of work to explore first.
| Posting evidence (source and duty) | Question to ask | Evidence to look for and whom to ask | What would change my next step |
|---|---|---|---|
| City of Cleveland bulletin 11045-HR, fifth bullet under Essential Duties & Job Functions. “Test data and/or applications to validate data accuracy/quality.” | When a dashboard number is challenged, who traces the discrepancy and who decides whether to revise the measure? | Ask a city analyst or their manager to walk through a recent correction. Look for what the analyst checked, how often this happens, and who signed off. | If analysts at my level regularly investigate discrepancies and propose fixes, explore city reporting first. If they mainly run predefined checks, compare a program-analysis role. If I only have the ad, keep the question open. |
Use four questions to make the comparison specific. Task frequency asks how often the work comes up. Share of work time asks how much of a typical week or month it takes. Importance asks what depends on getting it right. Decision authority asks which parts the analyst can decide and who approves a change. A quarterly quality check can take an hour and still stop a misleading report from going out. A prominent bullet gives you a reason to ask about the duty. The answers need evidence from the team. These four questions are FermatMind Editorial’s method for reading job ads. For occupation-level ratings, O*NET defines its task relevance, frequency, and importance scales separately. O*NET task scale definitions
Turn your Holland Code result into a prediction you can check
Maya remembers a student project where attendance suddenly jumped. She enjoyed tracing the jump to a change in how repeat visits were counted and explaining the correction to the event team. Refreshing the same report after the counting rule was settled held less appeal. Her Holland result points toward investigation and organizing information. Her project narrows that down to a question about the work itself.
She starts with a prediction. Program evaluation will give me more chances to investigate why a result changed. She’d reconsider if analysts at her level in evaluation mainly prepare established reports while city analysts get to investigate disputed metrics. A word like predictive gives her a lead. A recent assignment can show who posed the question, dealt with ambiguous data, and got to propose a change.
Holland’s six interest areas describe the kinds of work activities people report enjoying. The O*NET Interest Profiler manual explains the forms, scoring, and measurement evidence for that specific assessment. Across 105 studies, Hoff and colleagues estimated a modest positive correlation between interest fit and overall job satisfaction (ρ = 0.19, 95% confidence interval 0.16 to 0.21). That estimate summarizes a group-level association. Maya uses her result to decide which activity to investigate. She assesses her skills, eligibility, and working conditions separately. O*NET Interest Profiler manual overview · Full technical manual, chapters 2, 3, 5, and 6 · Hoff et al., 2020
If you’re still figuring out what the letters or scores mean, start with the guide to what a Holland career interest test can tell you.
How two Data Analyst jobs describe different work
Both vacancies are closed. The first column of each table summarizes advertised duties in source order. The second shows what the work could produce and gives you a way to picture it. Public tools and documents add context about the organization. To learn how much of that work falls to one analyst, ask about an assignment at the same level.
Cleveland, Ohio — dashboards and open data
Cleveland’s Office of Urban Analytics & Innovation advertised the full-time role from September 28 to October 13, 2026. Its bulletin separates Examples of Duties from Essential Duties & Job Functions. The first two rows summarize the examples. The next seven follow the essential duties in order. City of Cleveland job bulletin 11045-HR
| Advertised duty | What the work could produce |
|---|---|
| Set file specifications and standard field names, advise project leads, and help technical teams implement files | A file specification other teams can follow. |
| Monitor database performance, report improvements, document standards, and support reorganization, recovery, and backups | A performance report or a procedure the technical team can use. Ask who owns recovery and backups. |
| Analyze business or service performance using several data sources | An explanation of a service problem with a proposed action for managers or frontline staff. |
| Work with stakeholders and analysts to deliver lasting solutions | A question the department agrees to solve and a solution teams can maintain. |
| Help departments prepare items for the public Open Data Portal | A public dataset with clear field definitions and metadata. |
| Build dashboards for departmental goals and performance measures | A dashboard managers can use to track progress. |
| Test data or applications for accuracy and quality | A quality check that explains whether a result is ready to use. |
| Track projects from idea to completion | A plan showing requirements, milestones, and outstanding work. |
| Lead status meetings and communicate progress | Updates that help business, technical, and leadership teams coordinate. |
Qualifications to check separately. Cleveland listed a relevant bachelor’s degree and two years of database management systems experience. The bulletin describes a substitution of two years of equivalent education, training, or experience for each missing year of college. It also lists Windows and Microsoft Office proficiency, a lifting requirement, and a valid Ohio driver’s license. Power BI, Tableau, or similar software appears among the preferences. Cleveland bulletin, minimum and preferred qualifications
Cleveland’s 311 page links to a service request dashboard and a map explorer. The city’s Data Analytics team describes building tools for city staff and publishing open data. A department could use a tool like this to examine a backlog, while a resident might look up a request. Both audiences need clear definitions. Cleveland 311 · Cleveland Data Analytics
The city’s published inspection outcomes show why data quality can involve judgment. The Cleveland Department of Public Health defines No Access Available as an attempt where the inspector couldn’t access the location and No Code Violation Observed as an inspection where no violation was observed. Grouping them under “no violation found” would treat an unobserved condition as an inspection result. It’s the kind of distinction Maya wants to investigate. She’d ask whether analysts help settle definitions like these or check them against rules someone else has set. Cleveland public health inspection outcome definitions
Colorado — program analysis and risk assessment support
The Colorado Department of Health Care Policy & Financing advertised a full-time Data Analyst role from September 17 to 28, 2026, under the state classification Statistical Analyst I. The posting says analysts may work for any unit and identifies support for the Office of Community Living under an agreement with the federal Department of Justice. Its six advertised duties follow below. State of Colorado job bulletin UHA 4779 - 9.2026
| Advertised duty | What the work could produce |
|---|---|
| Query and analyze data under supervision, including statistical analysis | A result and an explanation a reviewer can check. |
| Use tools such as Cognos or Toad to query relational databases | An extract with the records and fields a question requires. |
| Write code to prepare data for Colorado Medicaid operations and evaluation questions | An analysis dataset with clear rules about what to include. |
| Prepare reports and communicate findings to internal and external audiences | A report explaining the finding, who it describes, and where the evidence is limited. |
| Support the Office of Community Living and work for other units as needed | Analysis using one program’s definitions with room for assignments elsewhere in the department. |
| Support efforts to prevent long-term facility admissions through predictive analytics and risk assessment | Analytical work that supports the program. Ask whether the role builds models, checks inputs, or works with model outputs. |
Qualifications to check separately. Applicants had to be Colorado residents when they applied. The bulletin lists four years of relevant professional experience or the equivalent through related education, relevant experience, or both. A qualifying bachelor’s degree counted for four years under its substitution rules. SQL or R, health-sector data analysis, statistical evaluation, and Tableau appear among the preferences. These terms describe this vacancy, so check the requirements in any new posting you plan to act on. Colorado bulletin, minimum qualifications and substitutions
Colorado’s At-Risk Diversion memo describes how analysis reaches people who can act on it. The department developed a model to help identify some Medicaid members receiving home and community-based services who may need nursing facility care. The memo calls for lists to go to case management agencies. Case managers contact members, record the outreach, and help arrange support. This is evidence about the program’s workflow. Whether the job holder develops the model, checks the inputs, or reports on outreach remains a question for the team. The effect on admissions would require a separate evaluation. HCPF operational memo OM 25-002 · HCPF’s January 2026 case manager update, an HTML alternative for the outreach workflow
Compare the assignments you’d actually want to do
Maya wants to know who gets to investigate a number when it looks wrong. The ads give her two versions of that question.
| What Maya wants to know | Cleveland | Colorado |
|---|---|---|
| Who uses the work? | The bulletin names city staff and leaders. The city’s public 311 tools also serve residents. | The bulletin names internal and external report audiences. The program memo separately identifies case management agencies as users of the at-risk list. |
| What might I deliver? | A dashboard or a public dataset with definitions that other people can follow. | A query, analysis, or report for a specific program question. |
| What would make this work interesting to me? | Ask how often analysts get to investigate a disputed measure and suggest a correction. Routine updates and project coordination may take much of the week. | Ask whether an analyst at this level can test competing explanations or mainly prepares data and reports using an established method. |
| What example would help me decide? | “Tell me about a recent number someone challenged. What did the analyst check, and who approved the change?” | “What did a Statistical Analyst I produce on a recent assignment, who chose the method, and what changed in review?” |
The two vacancies are in different states and classification systems. For your own shortlist, look for a third posting at a similar level in the same market to see whether the pattern repeats. Occupation profiles can give you more duties to look for, including adjacent work such as maintaining reports and dashboards. Use them for background as you compare individual vacancies. FermatMind’s Business Intelligence Analysts profile · O*NET’s Business Intelligence Analysts profile
Each source answers a different question. A job posting records advertised duties and qualifications. An organization’s public work shows an output or workflow. A recent example from the team can clarify who does the work, how much time it takes, and who approves changes. Your own practice sample shows what you did and what you wanted to keep working on. Keeping those answers separate makes it easier to see what you still need to check.
Try each kind of work and revise it once
Use the same data, comparable time, and the same tools for both tasks. Save part of the time for revisions. Ask a willing reader to check what they understood and where they’d want more evidence. A practitioner may also tell you which parts resemble their own work. If you’re working alone, use the example objections below and record that you did the check yourself.
The practice data cover fictional community service requests. One group was offered follow-up calls starting in June. Every included request has a status recorded at day 30. The unit is the service request.
| Intake month | Group | Requests tracked | Still open at day 30 |
|---|---|---|---|
| May | Group later offered calls | 40 | 12 |
| May | Comparison group | 50 | 10 |
| June | Group offered calls | 40 | 8 |
| June | Comparison group | 50 | 9 |
Make a dashboard note an operations lead can use
Sketch a chart for the two intake months. May had 22 of 90 requests still open at day 30 (24.4%). June had 17 of 90 (18.9%). Label the measure and include the counts. Ask your reader what the chart counts and which figure they would check before sharing it.
In the example, the reader suggests the title “Residents still waiting.” Maya realizes that one resident can submit multiple requests. She changes the title to “Share of service requests still open 30 days after intake,” adds the counts, and makes two checks for the next update. Each request ID should be counted once, and every included request should have reached day 30. Ask your reader to explain the revised chart. If they can identify what was counted and when, the new title resolved the misunderstanding.
Write an evaluation note a program lead can use
Show the two groups separately. The group offered calls went from 12/40 (30%) to 8/40 (20%). The comparison group went from 10/50 (20%) to 9/50 (18%). The decline was 8 percentage points larger in the first group. Ask your reader which sentence would guide a decision about expanding the offer.
In the example, the reader tells Maya that the offer was assigned by service area and asks whether the calls explain the larger decline. Maya revises the note. “The area offered calls had a larger drop in the share of requests still open after 30 days. Before expanding the offer, compare earlier trends and check whether the areas handled different requests or introduced other service changes.” She asks for earlier months by area. The counts show how the groups changed. The data here leave the effect of the calls unresolved.
How the comparison changes Maya’s next step
Maya expected to prefer program evaluation. During the dashboard revision, she kept wanting to find out how duplicate requests and incomplete follow-up could change the chart. She also enjoyed testing rival explanations in the evaluation note. What mattered to her was getting to follow an uncertain number through to a correction or a defensible conclusion.
She moves a conversation with a city data analyst to the top of her list and asks for a recent disputed metric. If city analysts at the relevant level mainly maintain existing measures and program analysts regularly get to test explanations, she’ll put program evaluation first again. Her next search is for a role that gives her a recurring part in investigating questionable numbers. She’ll check skills and eligibility before treating any vacancy as an option.
Use the same check on jobs near you
- Pick two comparable postings. Save the employer, location, date, and link. Keep required qualifications and preferences separate from the activities you’re interested in.
- Choose one duty and fill in the card. Note what goes into the task, what it produces, who reviews it, and who uses the result. Use public work to see an example of an output and mark what still needs checking.
- Ask about a recent assignment. “What did you deliver, who reviewed it, and what changed?” If it was an unusual week, ask for a second example. Record whose work it describes and whether the answer changed your prediction. UC Santa Barbara’s guide to reading job descriptions · UC Santa Barbara’s guide to informational interviews
- Try a small version of the task and revise it. Record the feedback, what you could do with the tools and support available, and what you’d want to try again. If one exercise used tools you already knew and the other didn’t, record that difference before deciding which felt better.
You can start with an activity you already know you enjoy. If you’d like a starting profile, take the Holland Code career interest test. To keep track of interviews, practice tasks, and when to revisit your choice, use the career exploration worksheet.
Source check by FermatMind Editorial, November 10, 2026. The two employer bulletins support the duty and qualification comparisons. The Cleveland city pages and the Colorado agency materials describe examples of organizational work. Maya’s decisions and both practice tasks are editorial illustrations.
