Why is "Treating the baseline as a neutral technical detail rather than a commercial decision" a common mistake when using the Promo Baseline Estimator?

Symptom: The JBP argument about promotion ROI goes round in circles, with manufacturer and retailer each producing a correct calculation from a different baseline. Fix: The method you pick for the baseline is rarely neutral, because it largely decides who wins the ROI argument before anyone opens the numbers. Agree it with the retailer before the promotion cycle, usually as part of the annual JBP, because afterwards each side has a number to defend. Five questions settle it. (a) Which statistical method: a moving average of recent weeks, a regression (a statistical model fitted to past sales), a seasonal decomposition that separates the yearly pattern from everything else, or a comparison against control stores, similar stores that ran no promotion. (b) Which comparison period: the previous 13 weeks, the previous 52, or the same period last year. (c) How the season is adjusted for, (d) how the category's own growth or decline over time is handled, and (e) how the baseline is corrected when the number of stores stocking your product changes.

Why is "Confusing MAPE, the size of the baseline error, with its direction" a common mistake when using the Promo Baseline Estimator?

Symptom: A 10 percent MAPE gets accepted without question as a small error. Yet the direction matters too: whether the baseline sits above or below the truth decides whether every ROI claim built on it is too low or too high. Fix: MAPE measures the size of the error and says nothing about its direction. A baseline 10 percent too high understates the incremental volume, by 20 percent at this tool's default settings, and makes the promotions look too expensive. A baseline 10 percent too low overstates it by the same 20 percent and makes them look too cheap. Both show the same 10 percent MAPE, so the MAPE alone cannot tell you which of the two mistakes you are looking at. Report both: the direction (over, under, or accurate) and the size.

Why is "Assuming real baseline errors are the same size every week, the way they are in this simulator" a common mistake when using the Promo Baseline Estimator?

Symptom: A new baseline method scores a 5 percent MAPE in a test where the error is the same every week. Then, once it is in real use, it misses by 20 percent or more in weeks when the number of stores stocking the product changes or a competitor runs a promotion. Fix: This simulator applies one error to every week for teaching. Real baseline errors change size from week to week: smaller in quiet weeks, larger in weeks when something is changing, and badly wrong when a competitor runs a promotion to answer yours. Test the method you actually use on the weeks where a baseline is hardest to get right. Those are (a) weeks when the number of stores stocking your product changed, (b) weeks when a competitor was on deal, (c) weeks when one of your own products took sales from another of yours, and (d) weeks with a one-off supply problem. Say a method misses by 5 percent (its MAPE) in the easy weeks and by 25 percent or more in the hard ones. On this tool's scale it grades UNUSABLE in those hard weeks, however good its average looks.

Why is "Reporting a share of promoted volume without checking that it falls between 0 and 100 percent" a common mistake when using the Promo Baseline Estimator?

Symptom: A dashboard shows estimated Merchandising Efficiency (uplift / promoted volume) at 115 percent for the last 52 weeks, and the business reads that as a sign of very productive promotions. A share of promoted volume above 100 percent is not a sign of productive promotions. No baseline can produce it, so the uplift has almost certainly been counted over more weeks than the promoted volume, every quiet week's baseline miss included. Fix: Any share of promoted volume outside 0 to 100 percent, whether it is Merchandising Efficiency or incrementality (net incremental volume / promoted volume), should send you back to how it was built before anyone reads an ROI from it. Below 0, check the baseline. Above 100, check that the uplift and the promoted volume cover the same weeks. Build the check into your reporting: if the estimate falls below 0 or rises above 100 in any period, suppress the ROI figure and flag how it was built for review instead. This tool shows the check working; your own reporting should copy it.

Why is "Running the Source of Volume Decomposer or the Promo ROI Calculator before validating the baseline" a common mistake when using the Promo Baseline Estimator?

Symptom: The [Source of Volume Decomposer](/tools/source-of-volume-decomposer) shows a net share of uplift of 65 percent (net incremental volume / uplift: of the extra units above normal sales, the share that was genuinely new). At its starting setting, a 10,000-unit uplift, it also shows an event margin of $3,840. That is the profit the promotion makes in the weeks it runs, and the next tool, the [Promo ROI Calculator](/tools/promo-roi-calculator), puts a return on the event. But the baseline behind that uplift was 1,500 units too low on a 10,000-unit base, so 1,500 of those "incremental" units were volume that would have sold anyway. Fix: Every step in this chain starts from the baseline, so check it first. Then run the [Source of Volume Decomposer](/tools/source-of-volume-decomposer), then the [Promo ROI Calculator](/tools/promo-roi-calculator), and only then read the Performance Grid's verdict on the promotion. On the Source of Volume Decomposer's own starting figures, correcting a baseline that was 1,500 units too low turns a 10,000-unit uplift into 8,500. If the extra volume still comes from the same sources in the same proportions, the net incremental volume (6,500 to about 5,500) and the event margin ($3,840 to about $3,260) both shrink by the same 15 percent. That happens before anyone has even argued about where the extra volume came from. Had the baseline run high instead, every one of those numbers would move the other way. The lesson gives a rule of thumb for a single promotion, like the one in its worked example. A 10 percent error in the baseline becomes three to five times larger by the time it reaches net incremental volume in the profit bridge, the calculation that runs from volume down to profit, and the profit figure at its bottom can move further still. So defend the baseline before you defend any number built on it. A correct calculation on top of a wrong baseline is the failure this lesson exists to prevent.

Free tool · no login required

Promo Baseline Estimator

Put a size on what an error in the baseline, the sales you would have made with no promotion running, does to the uplift you report, the extra units above that baseline. Find the point where the calculation stops describing anything a store could produce, and bring the joint business plan (JBP) review a recommendation on how the baseline should be built before the argument about return on investment (ROI) starts. It is the same interactive model the full RGM Academy course uses in Trade Promotion Optimization Lesson 3, free to use with no sign-in.

Updated 11 October 2026Extracted from the Trade Promotion Optimization module, lesson 3: Baseline vs Incremental Volume

Loading calculator...

Guided walkthrough

Explore the tool, from setup to common mistakes

Five short sections explain the scenario, what each control does, how to read the output, and the mistakes to avoid. Open whichever helps. The tool above works without them.

5.1Scenario setup

The starting SKU, market, and assumptions the model makes.

You are the trade marketing director on a mainstream juice brand. The joint business plan (JBP) review, the annual meeting where you and the retailer agree on next year's plan, is three weeks away. The retailer's category team says your last four promotions delivered 5,200 incremental units. Your own sales team says 8,000. Same 52 weeks of checkout data, same four two‑week promotions, and two different ways of estimating the baseline, the volume that would have sold with no deal running.

Your CFO wants one defensible ROI number before the review. You have three options: (a) argue against the retailer's baseline; (b) concede and take the lower number; (c) work out how much a baseline error distorts the ROI calculation, and get a shared method agreed before the ROI argument starts.

Your job by Friday: use this tool to put a size on what a 10 percent error in the baseline does to the incremental volume you report, and find the point where the calculation stops describing anything real. Then bring the review a recommendation on how the baseline should be built rather than a claim about what the promotions returned. Getting the JBP to agree HOW to build the baseline is what makes agreeing what the promotions delivered possible.

Your objective

Use the simulator to watch an error in the baseline turn into a larger error in incremental volume, and find the size of error at which the calculation stops describing anything real. Then see why a 15 to 25 percent gap between a manufacturer's baseline and a retailer's, the gap this course plans on as a starting assumption rather than a measured finding, leaves every ROI figure in dispute.

Key assumptions

MAPE is the mean absolute percentage error: the average size of the baseline's miss, whichever direction it misses in. The number you set on the Your baseline estimate slider is called the offset: how far your estimate sits above or below the true baseline. In this simulator, MAPE is simply the size of the offset, because the tool applies that same error to all 52 weeks. That is a deliberate simplification: a real baseline method misses by different amounts in different weeks, tighter in quiet weeks and looser around seasonal peaks and promoted weeks. The uniform error keeps the lesson's central point in view: a modest percentage error on the baseline becomes a large percentage error on the incremental volume.
Incremental volume is a small difference between two large numbers. At the default settings the eight promoted weeks sell 55,273 units against a true baseline of 36,848 for those weeks, so the promotions added 18,425. A 10 percent error on that baseline is 3,685 units, and every one of them lands in the incremental figure, which it moves by 20 percent. The multiplier is always the baseline divided by the true incremental: 2.0 here, because a promotion that lifts sales by half adds half a baseline. The lesson's worked example, a 5,000‑unit baseline against 8,000 actual, gives 1.7 (a 10 percent error moves the incremental by 17 percent). Its rule of thumb for a promotion that lifts sales by a third to a half is 20 to 30 percent per 10 percent of baseline error, the same rule again. Your baseline is off in the 44 weeks with no deal as well, and the chart shows it, but no promotion ran in those weeks, so none of that error lands on a promotion.
An impossible Merchandising Efficiency, a reading below 0 percent or above 100, describes nothing that can happen in a store, because Merchandising Efficiency is the uplift as a share of everything sold in the promoted weeks (uplift / promoted volume). It is the figure you see in the reports written after each promotion from store sales data. It treats every extra unit sold in the promoted weeks as a sale the promotion created. Incrementality (net incremental volume / promoted volume) also takes out volume borrowed from your own other packs, other stores and other weeks, so on a real event it reads lower. A reading below 0 says the baseline has been set too high, by more than the whole uplift: the estimate says the promotions lost volume. A reading above 100 cannot come from a baseline at all, since even a baseline of zero gives exactly 100. In a real report it almost always means the uplift and the promoted volume were added up over different weeks, for example a whole year's uplift over the promoted weeks' sales. Either way, fix how the number was built before you read it.
This course plans on manufacturer and retailer baselines for the same promotion sitting 15 to 25 percent apart, a starting assumption rather than a measured finding. The way to close the gap is for both sides to agree how the baseline will be built before the promotion runs. Keep two quantities apart when you read the tool's grade: Baseline Quality measures one baseline against the truth, while the 15 to 25 is the gap between two people's baselines. A manufacturer 10 percent low and a retailer 10 percent high are each ACCEPTABLE on the tool's scale and still 22 percent apart (1.10 divided by 0.90).
This simulator is simplified on purpose. A real method for building the baseline has to deal with seasonality that differs by product, a competitor's promotion that keeps affecting your sales after it ends, and the dip in your own sales after a deal ends. It also has to handle changes in how many stores stock you, everyday price changes, and the choice of comparison stores, the similar stores with no promotion that you measure against. Four sliders cannot carry all of that, and they do not need to. The point they carry is the lesson's: agreeing one shared way of building the baseline matters more than which statistical method you choose.

5.2Controls & toggles

Every input the calculator exposes, its range, and what it changes.

Control	Range	Default	What it changes
Baseline trend growth slider	-10% to +15% per year (1% step)	+5% (a mid-growth category)	How fast the category grows or shrinks on its own, year over year, before any promotion. It scales every week's baseline by a trend factor that rises or falls evenly from week 1 to week 52. Positive in a growing category, negative in a declining one.
Seasonality amplitude slider	0% to 40% (1% step)	15% (a moderately seasonal category such as juice)	How far sales rise above the yearly average in the peak season, and fall the same distance below it in the low season. The tool shapes the year as a smooth wave that peaks in week 25. At 0 percent the year is flat; at 40 percent the peak week sells more than twice the trough week. This simulator assumes you know the seasonal shape, which is a large assumption: a baseline built from a plain average of recent weeks misses the swing entirely, and the bigger the swing, the further such a baseline sits from the truth.
Promo uplift factor slider	1.2x to 4.0x (0.1x step)	1.5x (promoted weeks sell half as much again, a starting assumption for a shallow price cut)	How much volume multiplies in each of the eight fixed promoted weeks. 1.5x is a starting assumption, not a typical figure: in one large study of US store sales, the median product's promoted week sold about three times a normal week, with nine in ten products between about 1.3x and 9.2x. Set the uplift your own promotions deliver. In this simulator the uplift never touches the baseline; it only changes the sales the model records in those weeks. Over a real year, promoting heavily does wear the baseline down. The lesson's Baseline Erosion card covers that, and this tool does not model it.
Your baseline estimate slider	-25% to +25% against the true baseline (1% step)	+10% (set wrong on purpose so the effect is visible; drag to 0 for an accurate baseline)	The error in your own baseline method, applied to every week. How far you move it from zero, in either direction, is the MAPE. Above zero your baseline is too high, which understates the incremental volume and makes the promotions look worse than they were. Below zero it is too low, which overstates the incremental volume and makes them look better than they were. The whole teaching happens on this slider.
Baseline Quality grade (derived)	EXCELLENT < 5% / ACCEPTABLE < 15% / PROBLEMATIC 15 to 25% / UNUSABLE 25% or more	ACCEPTABLE at default (MAPE 10%)	A grade read off the size of the error: EXCELLENT under 5 percent MAPE, ACCEPTABLE under 15, PROBLEMATIC 15 to 25, UNUSABLE at 25 or more. The bands are this tool's own, and the upper limit of the ACCEPTABLE band matches the target the lesson sets for a weekly baseline: an error under 15 percent. The grade measures one baseline against the truth, which is a different thing from the gap between two people's baselines: a manufacturer 10 percent low and a retailer 10 percent high are each ACCEPTABLE and still 22 percent apart.
Merchandising Efficiency is impossible (derived)	OK / firing	OK at default (est. Merchandising Efficiency 26.7% of promoted volume)	The main card's box turns red and adds a line headed 'Merchandising Efficiency is impossible' when the estimated Merchandising Efficiency (uplift / promoted volume: the uplift your baseline implies as a share of everything sold in the eight promoted weeks) falls below 0 percent or rises above 100. That is a reading no real market can produce: a share of what sold cannot go negative or exceed the whole, so a reading outside the band means the baseline error has broken the calculation. Fix the baseline before you read anything else. It fires only when your baseline is too high by more than the whole uplift, so on this slider it appears only at a 1.2x uplift, once your estimate is about 20 percent too high (from +21 percent at the default trend and season).

5.3Step-by-step exploration

7-step guided exploration of the scenario.

Read the default scenario
Leave every control where it starts: trend +5 percent, seasonality 15 percent, uplift 1.5x, baseline offset +10 percent. Read the main results card: the two incremental figures, the two baseline tiles under them, and the amber box at the foot of the card.
Expected outcome: In the eight promoted weeks, True baseline is 36,848 units and Your estimate 40,533, which is 10 percent higher. True incremental is 18,425 units; Est. incremental 14,740, which is 20 percent lower. The amber box says the baseline estimate is high by 10 percent, which understates the uplift and makes the promotions look worse than they were. The downloadable PDF grades this baseline ACCEPTABLE (MAPE 10 percent). The box stays amber, with no 'Merchandising Efficiency is impossible' line. The estimated Merchandising Efficiency (uplift / promoted volume, the extra units as a share of everything sold in the deal weeks) is 26.7 percent against a true 33.3 percent. So a baseline only 10 percent too high shrinks the incremental volume you can report from 18,425 to 14,740, and a fifth of the promotions' value disappears from the ROI calculation before anyone has looked at a single promotion.
Set the error to zero and see an accurate baseline
Drag Your baseline estimate to 0 percent. Read the main card again.
Expected outcome: Your estimate matches True baseline exactly (both 36,848 in the eight promoted weeks). Est. incremental equals True incremental (both 18,425). The amber box disappears, and the downloadable PDF's Baseline Quality grade reads EXCELLENT (MAPE 0 percent). This is the baseline the tool is built to help you defend. In a real planning cycle you get there by agreeing the method with the retailer before the trade plan is signed, rather than by arguing about the number after the promotion has run.
Find the point where the calculation breaks
Put Your baseline estimate back to +10 percent and set Promo uplift factor to 1.2x, a thin promotion, leaving trend and seasonality at their defaults. Then drag Your baseline estimate upward until the main card's box turns red.
Expected outcome: At +10 percent the error is already half the true incremental: 3,685 of 7,371 units. The box first turns red at +21 percent. Your estimated baseline for the eight promoted weeks, the volume you think would have sold in them with no deals, is now 44,586 units, more than the 44,219 that actually sold. So the estimated incremental volume, promoted sales minus your baseline, turns negative: the calculation says the promotions lost sales, which no promotion at a 1.2x uplift can do. That reading is the tool's way of saying the calculation no longer describes a real market. Note the size of error that gets you there: 21 percent, which the PDF's Baseline Quality grade only calls PROBLEMATIC. Now set the uplift back to 1.5x and drag the estimate to +25 percent. The box stays amber, with half the incremental gone, because at 1.5x the error would have to pass 50 percent, more than the whole uplift, before the calculation breaks. The thinner the promotion, the smaller the baseline error it takes to break the result. This course plans on manufacturer and retailer baselines starting 15 to 25 percent apart, and on a thin promotion one side of that gap is enough on its own to break the calculation.
Test trend growth
Put Your baseline estimate back to 0 percent so the baseline has no error, then drag Baseline trend growth, the category's own growth before any promotion, from +5 to +15 percent a year, the pace of a fast‑growing category.
Expected outcome: The true baseline of the eight promoted weeks climbs from 36,848 to 38,698 units, about 5 percent more, because the trend factor now reaches 1.15 by week 52. Actual sales rise with it. Now put the +10 percent offset back. The amber warning returns, and the error in units has grown from 3,685 to 3,870, because the same 10 percent now sits on a larger baseline. MAPE still reads 10 percent, because it measures the error as a percentage, and the error is still 20 percent of the true incremental. The same percentage error in your method costs more units in a growing category, so getting the baseline right pays off more there.
Test seasonality
Reset, then drag Seasonality amplitude from 15 to 40 percent, a heavily seasonal category.
Expected outcome: The weekly baseline now runs from about 2,750 units in the trough (week 1) to about 6,450 at the peak (week 25). The year's total barely moves, 239,716 against 239,800, because the seasonal highs and lows cancel each other out across a full year. What changed is how the volume is spread through the year. With the offset still at +10 percent, the error in units is still 10 percent of the promoted weeks' baseline (3,669 units, against 3,685 at the default season), because this simulator applies the same error to every week. That is the limit to notice. A baseline built from a plain average of recent weeks would sit too high in the trough weeks and too low at the peak, an error that changes size week by week and that this tool's uniform error cannot show. In a strongly seasonal category, build the baseline with a method that allows for the season before you calculate any ROI from it. A seasonal index does this: it says how far each period normally sits above or below the yearly average. A regression, a statistical model fitted to past sales, does it too when the season is one of its inputs.
Test the size of the uplift
Reset, then drag Promo uplift factor, how many times its normal volume a promoted week sells, from 1.5x to 4.0x. That is the kind of uplift a deep discount brings when the retailer also gives it full support, such as a display and a place in the retailer's flyer.
Expected outcome: Sales in the eight promoted weeks climb from 55,273 to 147,395 units and True incremental from 18,425 to 110,547. Read the error with the offset still at +10 percent: 3,685 units, unchanged. The unit error depends on the size of the baseline and the size of the bias, and on nothing else, so the uplift does not move it. What the uplift does change is how large that error is compared with the incremental volume. When promoted weeks sell 1.5x a normal week, the error is 20 percent of the incremental volume; at 4.0x it is 3.3 percent. Valued at $0.86 of margin a unit, what a promoted unit earns the brand in the Source of Volume Decomposer, the error is worth about $3,200 at either uplift. So the promotions most exposed to a baseline argument are the small ones, where the same error is a larger share of a thinner uplift. Drag the uplift down to 1.2x, leaving Your baseline estimate at +10 percent, and the same 3,685 units are half the incremental.
Map it onto Source of Volume and Promo ROI
Open the related concept pages (Promo Baseline, Promo ROI, Source of Volume, and the ten levers of trade promotion optimization, or TPO). Then run two more tools on this site, the Source of Volume Decomposer and the Promo ROI Calculator, on the same promotion.
Expected outcome: The baseline is the first input in the whole chain of trade promotion optimization (TPO) calculations. This tool answers: how much volume did the promotions add above the baseline (the uplift)? The Source of Volume Decomposer splits that added volume by where it came from. It answers the next question: how much of it was worth having (shoppers switching to you, the category growing, people consuming more), and how much was shoppers stocking up early on what they would have bought anyway? The Promo ROI Calculator answers the last question: after you subtract the margin you lost by discounting units that would have sold at full price anyway, what profit is left? Every one of those numbers starts from the incremental volume this tool produces, so a baseline error travels into all of them. The profit bridge is the step‑by‑step calculation that runs from volume down to profit. The lesson gives a rule of thumb for a single promotion, like the one in its worked example. A 10 percent error in the baseline becomes three to five times larger by the time it reaches net incremental volume in that bridge, and the profit figure at the bottom can move further still. That is why a 15 to 25 percent gap between two sides' baselines leaves every result on the scorecard open to dispute.

5.4Reading the output

Every KPI, the formula behind it, and how to interpret a positive or negative value.

KPI	Formula	How to read it
True baseline (the eight promoted weeks)	for each week, 4,500 x a factor for the growth trend x a factor for the season, then added up over the eight promoted weeks (the chart draws all 52)	The model's own answer to what would have sold with no promotion at all. Nobody ever observes this number in real life. Estimating it from the weeks with no deal running, adjusted for season, trend and distribution, is the whole craft of baseline estimation.
Your estimate (the eight promoted weeks)	true baseline x (1 + your offset / 100)	What your baseline method produces. In this simulator, that is the true baseline raised or lowered by one fixed percentage. In real life it comes from one of several methods. Three common ones work from your own sales history: a moving average of recent weeks, a seasonal index, and a regression (a statistical model fitted to past sales). A fourth compares the stores that ran the promotion with control stores, similar stores that did not. The offset slider stands in for the error in whichever method you use.
True incremental	sales in the eight promoted weeks - their true baseline	What the promotions added above the demand that was coming anyway, which is the uplift. This model has no cannibalization, store switching or dips, so all of it counts as new here; on a real event you would take those out to reach net incremental volume. This is the number every ROI claim to a retailer should be built on.
Est. incremental	sales in the eight promoted weeks - your estimated baseline for them	What your baseline says the promotions added. If your baseline is too high this reads too low, and the promotions look worse than they were; if it is too low this reads too high, and they look better. It sits above or below True incremental by exactly the error in your method in those weeks, in units. Your baseline misses in the 44 weeks with no deal too, and the chart shows it, but none of that lands on a promotion: those quiet weeks are where you check a baseline against what actually sold.
Merchandising Efficiency (estimated)	your estimated uplift / total volume sold in the 8 promoted weeks x 100	The share of everything sold in the deal weeks that your baseline credits to the promotions. It is divided by total promoted volume, the base the lesson measures every such share against. At the default settings it reads 26.7 percent against a true 33.3 percent. It counts every extra unit as new, which is how store‑data event reports print it. Incrementality (net incremental volume / promoted volume) also takes out volume borrowed from your own other packs, other stores and other weeks, so on a real event it reads lower. A real share sits between 0 and 100 percent. The tool calls a reading outside that range impossible: the baseline error has grown so large that the result could not happen in any real market.
MAPE (Baseline Quality)	in this simulator, the size of the error you set on Your baseline estimate	Mean absolute percentage error: on average, how far the baseline misses the true figure, whether it is too high or too low. Grades: under 5 percent EXCELLENT, under 15 ACCEPTABLE, under 25 PROBLEMATIC, 25 or more UNUSABLE. The bands are the tool's own. A 15 to 25 percent gap between a manufacturer's baseline and a retailer's does not put either side in the PROBLEMATIC band on its own: two ACCEPTABLE baselines on opposite sides of the truth can sit 22 percent apart.

Read Est. incremental against True incremental first. Est. incremental differs from True incremental by exactly the number of units your baseline gets wrong, and that is the number the retailer will challenge. Then read Baseline Quality, the one‑word grade you can take into a review. Last, check the red alert: if it is showing, the baseline is broken, and an ROI built on a broken baseline has nothing to stand on, so fix the method before any promotion ROI number leaves the room. Every later calculation in trade promotion optimization starts from this baseline. The Source of Volume split uses it to show where the extra volume came from. The profit bridge uses it to work out the profit. The Performance Grid then uses those results to decide whether to scale, hold, fix or cut the promotion. A broken baseline breaks all of them.

5.55 common mistakes to avoid

Diagnostic patterns that catch the most common misuse of this calculator.

Mistake 1Treating the baseline as a neutral technical detail rather than a commercial decision
Symptom: The JBP argument about promotion ROI goes round in circles, with manufacturer and retailer each producing a correct calculation from a different baseline.
Fix: The method you pick for the baseline is rarely neutral, because it largely decides who wins the ROI argument before anyone opens the numbers. Agree it with the retailer before the promotion cycle, usually as part of the annual JBP, because afterwards each side has a number to defend. Five questions settle it. (a) Which statistical method: a moving average of recent weeks, a regression (a statistical model fitted to past sales), a seasonal decomposition that separates the yearly pattern from everything else, or a comparison against control stores, similar stores that ran no promotion. (b) Which comparison period: the previous 13 weeks, the previous 52, or the same period last year. (c) How the season is adjusted for, (d) how the category's own growth or decline over time is handled, and (e) how the baseline is corrected when the number of stores stocking your product changes.
Mistake 2Confusing MAPE, the size of the baseline error, with its direction
Symptom: A 10 percent MAPE gets accepted without question as a small error. Yet the direction matters too: whether the baseline sits above or below the truth decides whether every ROI claim built on it is too low or too high.
Fix: MAPE measures the size of the error and says nothing about its direction. A baseline 10 percent too high understates the incremental volume, by 20 percent at this tool's default settings, and makes the promotions look too expensive. A baseline 10 percent too low overstates it by the same 20 percent and makes them look too cheap. Both show the same 10 percent MAPE, so the MAPE alone cannot tell you which of the two mistakes you are looking at. Report both: the direction (over, under, or accurate) and the size.
Mistake 3Assuming real baseline errors are the same size every week, the way they are in this simulator
Symptom: A new baseline method scores a 5 percent MAPE in a test where the error is the same every week. Then, once it is in real use, it misses by 20 percent or more in weeks when the number of stores stocking the product changes or a competitor runs a promotion.
Fix: This simulator applies one error to every week for teaching. Real baseline errors change size from week to week: smaller in quiet weeks, larger in weeks when something is changing, and badly wrong when a competitor runs a promotion to answer yours. Test the method you actually use on the weeks where a baseline is hardest to get right. Those are (a) weeks when the number of stores stocking your product changed, (b) weeks when a competitor was on deal, (c) weeks when one of your own products took sales from another of yours, and (d) weeks with a one‑off supply problem. Say a method misses by 5 percent (its MAPE) in the easy weeks and by 25 percent or more in the hard ones. On this tool's scale it grades UNUSABLE in those hard weeks, however good its average looks.
Mistake 4Reporting a share of promoted volume without checking that it falls between 0 and 100 percent
Symptom: A dashboard shows estimated Merchandising Efficiency (uplift / promoted volume) at 115 percent for the last 52 weeks, and the business reads that as a sign of very productive promotions. A share of promoted volume above 100 percent is not a sign of productive promotions. No baseline can produce it, so the uplift has almost certainly been counted over more weeks than the promoted volume, every quiet week's baseline miss included.
Fix: Any share of promoted volume outside 0 to 100 percent, whether it is Merchandising Efficiency or incrementality (net incremental volume / promoted volume), should send you back to how it was built before anyone reads an ROI from it. Below 0, check the baseline. Above 100, check that the uplift and the promoted volume cover the same weeks. Build the check into your reporting: if the estimate falls below 0 or rises above 100 in any period, suppress the ROI figure and flag how it was built for review instead. This tool shows the check working; your own reporting should copy it.
Mistake 5Running the Source of Volume Decomposer or the Promo ROI Calculator before validating the baseline
Symptom: The Source of Volume Decomposer shows a net share of uplift of 65 percent (net incremental volume / uplift: of the extra units above normal sales, the share that was genuinely new). At its starting setting, a 10,000‑unit uplift, it also shows an event margin of $3,840. That is the profit the promotion makes in the weeks it runs, and the next tool, the Promo ROI Calculator, puts a return on the event. But the baseline behind that uplift was 1,500 units too low on a 10,000‑unit base, so 1,500 of those "incremental" units were volume that would have sold anyway.
Fix: Every step in this chain starts from the baseline, so check it first. Then run the Source of Volume Decomposer, then the Promo ROI Calculator, and only then read the Performance Grid's verdict on the promotion. On the Source of Volume Decomposer's own starting figures, correcting a baseline that was 1,500 units too low turns a 10,000‑unit uplift into 8,500. If the extra volume still comes from the same sources in the same proportions, the net incremental volume (6,500 to about 5,500) and the event margin ($3,840 to about $3,260) both shrink by the same 15 percent. That happens before anyone has even argued about where the extra volume came from. Had the baseline run high instead, every one of those numbers would move the other way. The lesson gives a rule of thumb for a single promotion, like the one in its worked example. A 10 percent error in the baseline becomes three to five times larger by the time it reaches net incremental volume in the profit bridge, the calculation that runs from volume down to profit, and the profit figure at its bottom can move further still. So defend the baseline before you defend any number built on it. A correct calculation on top of a wrong baseline is the failure this lesson exists to prevent.

Related concepts

Go deeper on the theory

Continue with the lessons

Go further inside Trade Promotion Optimization

This calculator is the sandbox slice of Lesson 3: Baseline vs Incremental Volume. Each of the other 7 Trade Promotion Optimization lessons teaches a complementary concept that sharpens how you read the output above.

See Promo Baseline Estimator inside the full lesson

RGM Academy lets you pull every commercial lever yourself inside a senior-practitioner simulator, with the AI RGM Strategist coaching every decision you make.

Unlock the full Baseline vs Incremental Volume lesson

Or see what a team rollout includes