A useful resume number helps explain work you actually did. It can show volume, the scope of a responsibility, an observed change or a carefully bounded estimate. It does not need to be a dramatic percentage. UConn's bullet-writing guidance starts with what you did, how and why, then recommends quantification when appropriate. Choose evidence for the role you are applying for; do not turn every responsibility into a claim of business impact.
Choose what the evidence can establish
| Kind | What it can describe | What it cannot establish by itself |
|---|---|---|
| Workload or scope | Items handled, people supported or a defined responsibility during a stated period. | Accuracy, improvement, satisfaction or financial benefit. |
| Measured change | A difference between comparable observations with a clear baseline and denominator. | That your action alone caused the change. |
| Estimate | An approximate quantity derived from stated, reasonably representative inputs. | An exact audited result or a forecast that will occur every month. |
| Team result | A documented collective output accompanied by your own responsibility. | That one person produced or caused the entire result. |
A repeatable record-to-bullet method
- 1Identify the contribution relevant to the posting: the work, deliverable, responsibility or problem you can explain.
- 2Check records you are permitted to use, such as a task log, reconciliation, project history, approved report or performance review. Do not retrieve confidential information without authorization.
- 3Write down the measurement period, unit, denominator and inclusion rules before calculating. Check for duplicates, cancelled items and changes in scope.
- 4Calculate transparently. Keep the underlying observations, not just a polished final percentage.
- 5Separate your work from reviewers, colleagues and other changes. Choose verbs such as prepared, reviewed, contributed or approved that match your actual authority.
- 6Draft a concise bullet and check what a reader could infer. Remove a number or soften a causal claim if the evidence does not support it.
All four exercises are constructed
Every record, candidate and outcome below is fictional instructional material. The arithmetic is reproducible from the stated facts; none is a real client result, an employer benchmark or a promised hiring outcome. Adapt the method to your own permitted evidence, not the sample achievements.
Exercise 1: count workload without inventing an improvement
Underlying facts: an administrative assistant logged supplier invoices at a fictional small business from September 1–28, 2025. The intake log records distinct invoice identifiers entered by this assistant. Cancelled and duplicate entries have already been excluded. The assistant checked for missing purchase-order references and referred exceptions to a bookkeeper; the assistant did not authorize payments. There is no error-rate, payment-speed or cost measurement.
| Week | Distinct invoices logged by the candidate |
|---|---|
| September 1–7, 2025 | 48 |
| September 8–14, 2025 | 52 |
| September 15–21, 2025 | 46 |
| September 22–28, 2025 | 54 |
| Four-week total | 200 |
| Check | Worked answer |
|---|---|
| Record, period and unit | The fictional intake log; September 1–28, 2025; distinct supplier invoices entered by this candidate. |
| Calculation | 48 + 52 + 46 + 54 = 200 invoices. 200 ÷ 4 = 50 invoices per week on average during these four weeks. |
| Contribution | Logged invoices and flagged missing references for review; did not approve or pay them. |
| Final bullet | Logged 200 supplier invoices over four weeks and flagged missing purchase-order references for bookkeeper review. |
| Claim limits | Do not write 'processed 50 invoices every week,' because the weekly counts vary. Do not claim faster payment, improved accuracy or an annual total from this four-week record. |
The total is enough here. The average is shown to explain the arithmetic, not because another number improves the bullet. A workload measure helps describe scope; the completed logging and exception-review steps describe the work.
Exercise 2: compare rates without confusing percent and percentage points
Underlying facts: a fictional office coordinator introduced a document checklist for standard inventory requests. The supervisor approved the checklist. The same missing-document definition and request type were used in two consecutive four-week periods: January 6–February 2, 2025 and February 3–March 2, 2025. Each period includes all 80 eligible requests, with no requests excluded after counting. A request is counted once as incomplete if it lacked at least one required document at initial intake. Staffing and other conditions were not experimentally controlled.
| Week within each period | Before: eligible / incomplete | After: eligible / incomplete |
|---|---|---|
| Week 1 | 20 / 4 | 20 / 2 |
| Week 2 | 20 / 5 | 20 / 2 |
| Week 3 | 20 / 3 | 20 / 1 |
| Week 4 | 20 / 4 | 20 / 3 |
| Total | 80 / 16 | 80 / 8 |
| Check | Worked answer |
|---|---|
| Record, period and unit | The fictional intake log; two consecutive four-week periods; proportion of eligible requests incomplete at first intake. |
| Rates | Before: 16 ÷ 80 × 100 = 20%. After: 8 ÷ 80 × 100 = 10%. These are request rates, not individual document counts. |
| Percentage-point change | 20% − 10% = a 10-percentage-point decrease. |
| Relative percentage change | (20% − 10%) ÷ 20% × 100 = a 50% relative decrease from the original incomplete-request rate. |
| Contribution | Created and maintained the supervisor-approved checklist; other staff used it during intake. |
| Final bullet | Created an intake checklist; missing-document rates fell from 20% to 10% across consecutive four-week periods of 80 requests each. |
| Claim limits | The observed decrease followed the checklist's introduction, but this comparison does not isolate its effect from staffing or other changes. Do not say you eliminated missing documents or improved all office accuracy by 50%. |
Use the from/to rates when they make the claim easier to interpret. A 10-percentage-point fall is not a 10% relative fall: here the rate halved. If you cannot establish a comparable definition or denominator, explain the observations separately instead of calculating an improvement percentage.
Exercise 3: estimate staff-hours without claiming cash savings
Underlying facts: a fictional reporting assistant built an Excel validation template for a recurring report. Five comparable preparation runs before the template took 9, 10, 11, 10 and 10 minutes of active staff time. Five after took 4, 5, 6, 5 and 5 minutes. Both sets cover the same preparation steps and batch size; unattended import time and independent review are excluded from both. The September 2025 report log contains 24 completed batches. These ten timed samples may not represent every batch or later month.
| Observation | Before template | After template |
|---|---|---|
| 1 | 9 | 4 |
| 2 | 10 | 5 |
| 3 | 11 | 6 |
| 4 | 10 | 5 |
| 5 | 10 | 5 |
| Total / mean | 50 / 10 | 25 / 5 |
| Check | Worked answer |
|---|---|
| Record, period and unit | Fictional timed preparation samples and September 2025 report log; active staff-minutes per batch and batches per month. |
| Calculation | Before mean: 50 ÷ 5 = 10 minutes. After mean: 25 ÷ 5 = 5 minutes. Estimated monthly difference: (10 − 5) × 24 = 120 staff-minutes = 2 staff-hours. |
| Assumptions | The timed runs reasonably represent those 24 batches and the template is used for the same preparation work. The estimate does not cover independent review, unattended processing or annual activity. |
| Contribution | Built and used the validation template. The supervisor retained final report review. |
| Final bullet | Built an Excel validation template; estimated 2 fewer staff-hours of monthly report preparation from timed samples and 24 recorded batches. |
| Claim limits | This is an estimated reduction in active preparation time, not a measured reduction in headcount, wages, budget or total report turnaround. Do not multiply by a salary rate and present cash savings without separate financial evidence. |
The estimate label belongs next to the result. Retain the inputs and name the month in your supporting notes; do not imply that the same workload recurred indefinitely. If batch sizes vary widely, use a better-matched sample or omit the aggregate rather than hiding that uncertainty.
Exercise 4: separate a team result from personal contribution
Underlying facts: a fictional support team migrated 120 knowledge-base pages between April 7 and May 18, 2025. A page inventory assigns one final content reviewer to each page, avoiding double counting. The candidate reviewed 30 distinct pages: 18 needed updates and 12 were confirmed current. Colleagues reviewed the remaining 90. The project lead approved publication. There are no measurements of customer satisfaction, support resolution time or revenue.
| Inventory item | Distinct pages | Responsibility |
|---|---|---|
| Candidate: revised content | 18 | Candidate updated instructions; project lead approved publication. |
| Candidate: confirmed current | 12 | Candidate checked existing instructions; no content revision was needed. |
| Colleagues: remaining pages | 90 | Other assigned reviewers. |
| Team total | 120 | Project lead coordinated and approved the migration. |
| Check | Worked answer |
|---|---|
| Record, period and unit | Fictional page inventory and revision history; April 7–May 18, 2025; distinct pages reviewed or revised. |
| Calculation | 18 revised + 12 confirmed current = 30 personally reviewed pages. 30 candidate pages + 90 colleague pages = 120 team pages. |
| Contribution | Reviewed 30 pages, including 18 revisions; did not lead the project or approve publication. |
| Final bullet | Reviewed 30 support articles, including 18 content revisions, within a six-week team migration of 120 knowledge-base pages. |
| Claim limits | Do not claim to have written all 120 pages, led the migration or reduced support demand. Reviewing 30 pages is not evidence of causing 25% of an unmeasured business improvement. |
A deliberate decision to leave the number out
A separate fictional coordinator wrote an approved handover guide after taking over inconsistent notes. The permitted evidence is the guide and the supervisor's review. There is no reliable count of old notes, timed comparison or record of questions prevented. A workload estimate would add uncertainty without explaining the useful work.
Qualitative final bullet
Wrote a handover guide that separated open requests, owner responsibilities and escalation steps for the incoming coordinator.
The deliverable, content and purpose are specific without promising faster onboarding. Do not add 'cut questions by 40%' or 'saved 10 hours' because those results were not measured. Achievements without tracked numbers provides more qualitative alternatives.
Copy this evidence worksheet before drafting
| Field | What to record |
|---|---|
| Original task or contribution | What did you personally do, prepare, check, change or coordinate? |
| Available record and access | Which permitted record supports it? Can you retain notes or an approved aggregate without disclosing confidential information? |
| Measurement period | Exact start/end dates or a clearly defined reporting period; note unusual weeks or changes in workload. |
| Unit and counting rules | Requests, invoices, pages, staff-minutes or another meaningful unit; specify duplicates, exclusions and the denominator. |
| Original observations | Counts, baseline and comparison values, or the input values used in an estimate. |
| Calculation | Show the arithmetic with units. Label a relative percentage, percentage-point difference or estimated total accurately. |
| Individual contribution | Your actions and authority; colleagues' work, approvals and other relevant changes. |
| Limitations | What was not measured? Does the comparison support an association, a causal claim or only scope? |
| Final wording | Write the useful supported claim, put an estimate label beside an estimate, and omit a figure that does not improve the explanation. |
Check the bullet before submitting
- ▸Can you explain the source, period, calculation and responsibility without inventing a missing input?
- ▸Does an average sound like an exact recurring count? Does a '+' sign claim a lower bound you can actually establish?
- ▸Have you confused hours of active work with elapsed processing time, or time with actual money saved?
- ▸Does the wording credit a team result entirely to you or imply approval authority you did not hold?
- ▸Would a specific qualitative action be clearer than an uncertain number?
Use the 51 quantified bullet examples for role-specific patterns and the XYZ guide for an optional sentence structure. The work-experience guide handles titles and dates; the tailoring worked example helps select relevant evidence. A checker score cannot validate your numbers or predict a hiring decision; see the score methodology.