Revenue "up 120%" makes a great slide until someone asks from what base. Percent change mistakes are not arithmetic errors—they are story errors. The formula is short; the discipline is labeling old and new values correctly.
Old value is the denominator
Change from 80 to 100 is (100−80)/80 = 25% increase. Using 100 as the base gives 20%—wrong story. In templates, color the old column and lock the formula to it.
When the old value is zero or negative, percent change is not meaningful or is misleading. Say "new positive" or show absolute change instead of forcing a percent.
Do not compare pre-rounded numbers
Dashboard A shows 4.2M rounded from 4.18M; last month showed 4.0M from 3.96M. True change is smaller than the rounded comparison suggests. Calculate change on full values, round for display after.
Sign and language
"Decreased by -15%" doubles negative. Write "decreased 15%" or "change −15%." Consistent verbs prevent executives from cheering a negative sign.
Validate with a Percent Change Calculator on two spreadsheet cells before you paste into slides. If the percent disagrees with your mental math, check whether someone swapped old/new columns.
Percent change is not percent of
Taking 15% off a price is percent of a number. Going from 15% off to 20% off is not a 5-point "percent change" in the price—those are different bases. Keep vocabulary strict in meeting notes.
Spreadsheet QA before the board deck
Add a hidden QA sheet that recomputes every percent on the deck from raw tables. Conditional formatting flags when slide text does not match QA within 0.1 points. One tired analyst paste is all it takes to misstate growth.
Teach interns that "percent off" and "percent change" live in different worksheet tabs with different color headers—never merge the templates.
Leaders and press releases
Press releases love "fastest growth in five years" with a percent change from a COVID-low base. Ask for the absolute numbers in the appendix table. A 200% rise from a tiny base is still a tiny business in absolute terms—context prevents embarrassment in investor Q&A.
Healthcare and safety metrics should show counts alongside percents: "incidents down 40% (5 to 3)" reads honest; "down 40%" alone hides that one event is noise.
Seasonality breaks comparisons
Comparing December retail to January without noting seasonality invites false panic. Use year-over-year change with the same month last year as old value, not last month, when the question is seasonal performance. Label the comparison in the chart subtitle: "YoY, same month."
Price changes mix unit and revenue: if price rose 10% and units fell 5%, revenue change is not the sum of those percents. Decompose in three lines: price effect, volume effect, mix—percent change on revenue alone hides the lever.
Portfolio rollups
Division A up 10% and Division B up 10% does not mean the company is up 10% unless weights match. Roll up dollars, then compute change on the total. Weighted averages of percents without weights recreate the mistakes this guide warns about.
Investor slides should show revenue dollars for the two periods adjacent to the percent—always.
Color the "old" column blue and the "new" column green in every template—visual habit beats memory in late-night edits.
Goals framed as percent change
"Increase NPS by 10%" is a point change goal, not always a percent change of the old score—read the OKR literally. Mixing Net Promoter math with revenue percent change templates causes leadership reviews to compare incompatible stories on one slide.
When goals are relative, store the baseline quarter in the same cell comment every month so nobody swaps baselines mid-year.
Tables versus charts
Charts truncate axes to dramatize change; tables should carry exact values. Pair every chart slide with a two-row table for old and new so the board can audit the percent caption without squinting at pixels.
Exporting charts to PDF without the underlying table is how decisions get made on decorative graphics—break that habit in your team template.
When comparing cohorts of unequal size, weight the rollup—percent change on unweighted averages of group percents is a common dashboard bug.
Frequently Asked Questions
Can change exceed 100%?
Yes, when the old value is small. A jump from 2 defects to 9 is 250% increase—true but emotionally loud; pair with absolute counts.
What about symmetric averages?
Do not average percents from different bases unless using properly weighted methods.
How do I show change when old is zero?
Use absolute delta or "new activity" labels instead of percent.
Should I annualize a monthly change?
Only with clear labeling; multiplying a one-month change by twelve assumes steady compounding that may not exist.
How do I audit a peer's percent?
Recompute (new−old)/old with unrounded inputs. Mismatch within 0.1 point usually means rounded inputs or inverted old/new.
For slice sizing before change analysis, confirm bases with Percent of a Number Calculator so you do not mix the two question types in one table.