Chapter 09
Measurement and decisions
45 terms
Practice by level
Terms
01Demand forecastDự báo nhu cầu
Definition
An estimate of how much the market will buy in a future period, used as the input to material purchasing and production planning.
Examples
- The June forecast for the distributor channel was 352,000 cases and actual sell in was 337,000 cases.
- The full year forecast is 5.5 million cases, equal to 132 million 500 ml bottles.
Easy to get wrong
A forecast is not a sales target. The 5.8 million case target exists to push the team, while the 5.5 million case forecast is the number materials are bought against. Using the target instead of the forecast makes the plant run about 300,000 cases too many and inflates inventory.
02Forecast horizonTầm dự báo
Definition
The span of time from today to the furthest period the forecast covers.
Examples
- The company forecasts 18 months ahead, with the first 3 months detailed by stock keeping unit and by week.
- Imported flavouring has an 8 week lead time, so the part of the forecast used to order it has to look further than 8 weeks out.
Easy to get wrong
The horizon must be longer than the longest lead time. Forecasting only 4 weeks ahead while flavouring needs 8 weeks means half the purchase orders rest on guesswork rather than on the forecast.
03Forecast aggregation levelMức tổng hợp dự báo
Definition
The level of detail a forecast is built at, across product, geography and time.
Examples
- At brand level and monthly buckets the error is 8.2 percent.
- The same numbers split to stock keeping unit level carry 21.5 percent error, and split again to unit by distributor across 42 distributors the error reaches 34 percent.
Easy to get wrong
Error always rises as you split the number down. A correct total does not mean each unit is correct, and the warehouse still runs short on one code and long on another while the total matches.
04Forecast errorSai số dự báo
Definition
The gap between the forecast and the actual for the same period at the same level of aggregation.
Examples
- The forecast was 352,000 cases, the actual was 337,000 cases, an error of 15,000 cases.
- 15,000 cases divided by the actual of 337,000 cases gives an error of 4.5 percent.
Easy to get wrong
Error is divided by the actual, not by the forecast. Dividing by 352 gives 4.3 percent, lower than the true figure, and that method always flatters periods where the forecast ran high.
05Mean absolute percentage errorSai số tuyệt đối bình quân
Definition
The average of forecast errors expressed in percent and taken as absolute values across several periods or several units.
Examples
- Over the last twelve months at stock keeping unit level the mean absolute percentage error is 21.5 percent.
- Over the same twelve months at brand level the figure is 8.2 percent.
Easy to get wrong
Taking absolute values throws away the sign. One month 12 percent high and one month 12 percent low average to 12 percent error while the two months together match exactly, so this measure has to be read alongside bias.
06Weighted absolute percentage errorSai số có trọng số
Definition
Total absolute error divided by total actual volume, so large units carry more weight than small ones.
Examples
- The portfolio shows 12.4 percent on a simple average but only 9.8 percent when weighted by volume.
- The lead unit at 46 percent of volume has 7 percent error, while the festive gift unit at 1 percent of volume has 80 percent error.
Easy to get wrong
A simple average is dragged around by small units. A unit selling 200 cases with 80 percent error is only 160 cases, while the lead unit selling 155,000 cases with 7 percent error is 10,800 cases, and that is what empties the warehouse.
07Forecast biasSai lệch có hướng
Definition
A tendency for the forecast to lean the same way period after period, consistently above or consistently below actual.
Examples
- In 12 months the forecast ran high in 9 of them, and the cumulative gap for the year was 6.2 percent high.
- 6.2 percent of 5.5 million cases is 341,000 cases of material bought against volume that did not sell in the period.
Easy to get wrong
Bias is more damaging than random error. Random error cancels out across months, while a steady overforecast builds inventory month after month with no month pulling it back.
08Naive forecastDự báo ngây thơ
Definition
A forecast that simply carries forward last period or the same period last year with no further work.
Examples
- Using May actual of 331,000 cases as the June forecast, applied across the whole year, gives 16.8 percent error.
- Using last year same month multiplied by 1.05 gives 14.1 percent error.
Easy to get wrong
The naive forecast is the benchmark every process has to beat. A process that occupies 6 people and lands at 17 percent error is spending effort to do worse than copying last month.
09Forecast value addedGiá trị gia tăng của dự báo
Definition
How much each step in the forecasting process improves error compared with the step before it.
Examples
- The naive forecast errs 16.8 percent, the statistical model brings it to 13.6 percent, so that step adds 3.2 points.
- After the sales team overrides, error is 12.4 percent, adding a further 1.2 points.
Easy to get wrong
Some steps make error worse. If the sales override moves error from 13.6 to 15.1 percent, that step has negative value, and dropping it is both faster and more accurate.
10SeasonalityTính mùa vụ
Definition
A pattern of rises and falls in demand that repeats at the same point every year.
Examples
- The month before the lunar new year ships 479,000 cases into the channel while September ships 290,000, some 65 percent higher.
- The four hottest months carry 41 percent of annual volume.
Easy to get wrong
Seasonality is not trend. September running below August is not a decline, and September can only be compared with September of the prior year.
11Seasonal indexChỉ số mùa vụ
Definition
The ratio between one month of demand and the average month of the year, used to spread an annual forecast across months.
Examples
- The month before the lunar new year carries an index of 1.42 and September carries 0.86.
- The annual channel forecast of 4.044 million cases averages 337,000 a month, and multiplying by 1.42 gives 479,000 for the festive month.
Easy to get wrong
The twelve indices must sum to exactly 12. If the set sums to 12.6, the monthly forecasts add up to 5 percent above the annual forecast and nobody can see where the extra sits.
12Intermittent demandNhu cầu gián đoạn
Definition
A demand pattern with many zero periods broken by a few periods with orders, typical of small units and occasion led lines.
Examples
- The festive gift unit takes orders in 14 of 52 weeks and none in the other 38.
- Spread across 52 weeks the average is 640 cases a week, but a week that actually has orders runs at 2,377 cases.
Easy to get wrong
An annual average is meaningless for intermittent demand. Holding stock to a 640 case week runs short through the 14 peak weeks and leaves stock dead through the other 38.
13Sales and operations planningLập kế hoạch bán hàng và vận hành
Definition
A monthly meeting process where sales, manufacturing, supply chain and finance agree one shared set of numbers for the months ahead.
Examples
- The cycle runs in 5 steps and closes with the executive meeting on working day 12 of each month.
- The output is a single number covering the next 18 months, used for both the production plan and the financial plan.
Easy to get wrong
This is not a meeting to report last month. Its purpose is to decide for the next 3 to 18 months, and last month is reviewed only to understand where the 4.5 percent error came from.
14Monthly planning cycleChu kỳ lập kế hoạch tháng
Definition
A fixed sequence of steps within the month, each with a defined input, output and owner.
Examples
- Working day 3 closes actual sales, day 6 is the demand review, day 9 the supply review and day 12 the executive meeting.
- Actual sales arriving 2 days late compresses all three later steps and strips the preparation out of the executive meeting.
Easy to get wrong
The cycle has to run on its dates even when the data is incomplete. Postponing the meeting to wait for numbers pushes the decision past the point where next month production can still be changed.
15Demand reviewRà soát nhu cầu
Definition
The step that reviews and agrees the demand forecast, before any plant or warehouse limit is applied.
Examples
- This step sets the July forecast at 368,000 cases into the channel, of which 31,000 come from promotions already signed.
- It also records June error at 4.5 percent, driven mainly by the six bottle multipack running two weeks longer than planned.
Easy to get wrong
The demand review has to produce unconstrained demand. Trimming the number because the plant clearly cannot make it destroys the record of the demand that was left unserved.
16Supply reviewRà soát nguồn cung
Definition
The step that tests agreed demand against plant capacity, material availability and warehouse space.
Examples
- Demand of 368,000 cases meets available capacity of 344,000 cases because of 2 planned maintenance days.
- The warehouse holds 4,800 pallet locations, about 230,000 cases, so space is not the binding constraint this month.
Easy to get wrong
Available capacity is not design capacity. A line rated at 24,000 bottles an hour yields 344,000 cases only after changeovers, maintenance and real running efficiency are taken out.
17Consensus forecastCon số đồng thuận
Definition
One single number that every function accepts and is jointly accountable for once the reviews are done.
Examples
- Sales proposed 382,000 cases, supply chain proposed 351,000, and the consensus landed at 368,000.
- That number feeds straight into the materials plan and into the 62.6 billion dong revenue forecast for the month.
Easy to get wrong
Consensus is not the arithmetic mean. The mean of 382 and 351 is 366.5, while 368 comes from removing 14,000 cases sales could not justify and adding 17,000 for a promotion already signed.
18Frozen periodKỳ chốt số
Definition
The near term window where the plan can no longer be changed because materials are ordered and the line is scheduled.
Examples
- The first 6 weeks of the plan are frozen and only week 7 onward can still be edited.
- Imported flavouring carries an 8 week lead time, so the flavour dependent part freezes 2 weeks earlier than the rest.
Easy to get wrong
Frozen does not mean change is forbidden. It means any change inside those 6 weeks is paid for with an extra changeover or an expedited material order, and whoever asks for it must see that price.
19Constrained planKế hoạch có ràng buộc
Definition
The plan after it has been cut back to the real limits of capacity, materials and space, which makes it the plan that can actually run.
Examples
- Demand of 368,000 cases is cut to 344,000 to match available capacity.
- The 24,000 case cut is taken from the 3 slowest units, leaving the 5 lead units untouched.
Easy to get wrong
The constrained plan must not replace the unconstrained one. Keeping both is what shows the company is leaving 24,000 cases, about 4.1 billion dong of revenue, on the table, and that is what answers whether to add capacity.
20Demand supply gapKhoảng trống nhu cầu và nguồn cung
Definition
The difference between agreed demand and what can actually be supplied in the same period.
Examples
- July shows a gap of 24,000 cases, 6.5 percent of demand.
- Two extra weekend shifts recover 16,000 cases at an added cost of 210 million dong.
Easy to get wrong
A gap is not only closed by adding supply. Moving a promotion to August or shifting 8,000 cases to an equivalent unit closes it too, with no extra shift.
21Planning scenarioKịch bản kế hoạch
Definition
A version of the plan built on a different set of assumptions so the consequences are visible before a decision is taken.
Examples
- The heavy promotion scenario delivers 402,000 cases but needs 4 extra shifts and 1.1 billion dong of budget.
- A scenario with sugar up 25 percent lifts cost of goods 2 percent and pulls gross margin from 30.5 to 29.1 percent.
Easy to get wrong
A scenario is only useful with a trigger attached. Write down that if volume passes 92,000 cases by the end of week 2 the heavy scenario starts, and otherwise the base plan holds.
22Exception managementQuản lý ngoại lệ
Definition
A way of working where only cases that breach a preset threshold reach the table and everything else runs untouched.
Examples
- The system flags only units with error above 20 percent, which in June was 11 units out of 68.
- Those 11 units account for 34 percent of the total absolute error of the portfolio.
Easy to get wrong
Thresholds have to be set on value as well as percent. A unit selling 200 cases with 40 percent error is not worth a meeting, while a unit selling 155,000 cases with 8 percent error is, even though it never trips the 20 percent rule.
23Plan adherenceTuân thủ kế hoạch
Definition
How closely actual production follows the frozen plan, measured on the right unit, the right quantity and the right week.
Examples
- Last week the plan called for 84,000 cases and 81,000 ran on the right unit in the right week, giving 96 percent adherence.
- A further 3,000 cases ran at the right quantity but slipped into the following week, so they still count as non adherent.
Easy to get wrong
Hitting total volume is not the same as adherence. Running the full 84,000 cases while the plant swaps units to suit itself still leaves the warehouse short of what is selling and long on what is not.
24Retail measurementĐo lường bán lẻ
Definition
A service that collects sales out of a sample of outlets, projects it to the whole market and sells the result to manufacturers in the category.
Examples
- The sample covers 1,850 outlets in the city and is projected to the 68,000 outlets of the full market.
- Reports are issued in 2 month periods and reach the company 6 weeks after the period closes.
Easy to get wrong
Retail measurement counts what shoppers take off the shelf, not what the company ships into the channel. In June the company shipped 337,000 cases in, distributors sold 331,000 cases out, and retail measurement recorded 326,000, three different numbers because stock is sitting in two places.
25Market sizeQuy mô thị trường
Definition
The total volume or total value the whole category sells in a defined market and period.
Examples
- The 500 ml soft drink category in the city runs at 77.2 million litres a year.
- Valued at the price shoppers pay at the shelf, the market is 1,306 billion dong a year.
Easy to get wrong
Market size has to state which price level it uses. The 1,306 billion dong figure is at retail price, while company revenue of 950 billion dong is at distributor price, so the two neither add nor divide into each other.
26Volume market shareThị phần theo sản lượng
Definition
One brand volume as a share of total category volume, in the same market and the same period.
Examples
- Brand A sells 8.8 million litres of a 77.2 million litre category, a volume share of 11.4 percent.
- Brand C leads with 19.8 million litres, a volume share of 25.6 percent.
Easy to get wrong
Volume share has to be converted to one common unit. Adding 330 ml bottles to 1.5 litre bottles and dividing by total bottles makes any brand with lots of small packs look bigger than it is.
27Value market shareThị phần theo giá trị
Definition
One brand sales value as a share of total category value, measured in money at a single price level.
Examples
- Brand A turns over 167.2 billion dong of a 1,306 billion dong category, a value share of 12.8 percent.
- The 12.8 percent value share sits above the 11.4 percent volume share because brand A prices above the category average.
Easy to get wrong
A rising value share is not automatically good news. If the category cuts price while brand A holds, value share climbs while volume stands still or falls.
28Relative price indexChỉ số giá tương đối
Definition
The average price per measured unit of a brand divided by the average price per the same unit for the whole category, multiplied by 100.
Examples
- Brand A sells at 19,000 dong a litre against a category average of 16,920 dong a litre, an index of 112.
- That index of 112 explains the gap between the 12.8 percent value share and the 11.4 percent volume share.
Easy to get wrong
The index moves when pack mix moves, not only when prices move. Selling more 330 ml packs, which carry a higher price per litre, lifts the index although no brand has raised a price.
29Moving annual totalTổng mười hai tháng
Definition
The sum of the last 12 months, refreshed each month by adding the new month and dropping the oldest one.
Examples
- Brand A moving annual total to the end of June is 8.8 million litres.
- Moving into July, adding 0.79 million litres for this July and dropping 0.74 million for last July lifts the total to 8.85 million litres.
Easy to get wrong
A moving annual total flattens seasonality and therefore reacts very slowly. A brand losing 20 percent of one month moves the annual total down only 1.7 percent, so it is the wrong measure for spotting a problem.
30Category growthTăng trưởng ngành hàng
Definition
The change in volume or value for the whole category against the same period a year earlier.
Examples
- The category grew from 73.7 to 77.2 million litres, a rise of 4.8 percent.
- Brand A grew 6.2 percent, so its volume share edged from 11.2 to 11.4 percent.
Easy to get wrong
Brand growth only reads correctly beside category growth. Growing 6.2 percent sounds fine, but if the category grew 9 percent the brand is losing share, not gaining it.
31Like for like comparisonSo sánh trên cơ sở tương đương
Definition
Comparing two periods after stripping out differences in outlet count, trading days or product scope.
Examples
- Modern trade sales rose 14 percent, but the number of measured chains went from 5 to 6, so the like for like rise is only 5 percent.
- February has 28 days and March has 31, a 10.7 percent difference in trading days that has to come out before comparing.
Easy to get wrong
Not every difference can be stripped out. Opening 60 new outlets also pulls sales down in nearby existing ones, so the like for like number still has to be read next to the total.
32Growth contributionĐóng góp vào tăng trưởng
Definition
The part of total growth created by one brand, unit or channel, stated in absolute terms or as a share of the total increase.
Examples
- The category grew 3.5 million litres, of which brand C contributed 1.6 million, or 46 percent of the increase.
- Brand A contributed 0.51 million litres, 15 percent of the increase, ahead of its 11.4 percent share.
Easy to get wrong
A high growth rate does not mean a large contribution. A small brand of 0.8 million litres growing 40 percent adds only 0.32 million litres, less than brand C adds by growing 8.8 percent on 1.6 million litres.
33Category development indexChỉ số phát triển ngành hàng
Definition
Category consumption per head in one region against the average for the whole market, multiplied by 100.
Examples
- The city indexes at 100 and the neighbouring province at 78.
- With a population of 8.6 million, the city consumes 9 litres per person a year against 7 litres in the province.
Easy to get wrong
A low index can mean opportunity and can equally mean the region does not suit the category. Check income and distribution coverage before reading 78 as headroom.
34Data projectionNgoại suy dữ liệu mẫu
Definition
The step that scales measurements from the outlet sample up to the whole market using a projection factor set separately for each outlet group.
Examples
- The 1,850 outlet sample is projected to 68,000 outlets, an average factor of 36.8 times.
- Small grocers carry a factor of 52 times while supermarkets carry 4.2 times because nearly all of them are measured.
Easy to get wrong
A wrong projection factor bends the whole picture. If the true outlet count is 74,000 rather than 68,000, market size is understated by 8.1 percent and every share computed on it is off.
35Cost of goods soldGiá vốn hàng bán
Definition
All cost attached directly to the goods sold in the period, covering raw materials, packaging and conversion.
Examples
- Cost of goods sold for the year is 660 billion dong on 5.5 million cases, or 120,000 dong a case.
- Within that 120,000 dong, materials and packaging are 82,000 dong and conversion is 38,000 dong.
Easy to get wrong
Cost of goods sold follows what was sold, not what was made. Producing 5.7 million cases while selling 5.5 million leaves the 200,000 case difference sitting in inventory, not in this period cost.
36Gross marginBiên lợi nhuận gộp
Definition
What is left of revenue after cost of goods sold, expressed either in money or as a percent of revenue.
Examples
- Revenue of 950 billion dong less cost of goods sold of 660 billion leaves 290 billion, a gross margin of 30.5 percent.
- Per case, an average selling price of 173,000 dong less a cost of 120,000 dong leaves 53,000 dong.
Easy to get wrong
Percentage margin and money margin can move in opposite directions. Selling another 400,000 cases to modern trade at 158,000 dong pulls the percentage to 30.1 while adding 15.2 billion dong of gross margin.
37Fixed and variable cost splitPhân tách chi phí cố định và biến đổi
Definition
Sorting cost by whether it moves with volume or stays the same regardless of volume.
Examples
- Of the 120,000 dong cost per case, 96,000 dong is variable and 24,000 dong is allocated fixed factory cost.
- Fixed factory cost is 132 billion dong a year and does not change whether the plant runs 5.5 or 6 million cases.
Easy to get wrong
Fixed cost per case is only true at the volume it was allocated on. At 6 million cases the fixed share drops to 22,000 dong, and comparing two periods without noticing this looks like the plant has cut cost.
38Contribution marginBiên đóng góp
Definition
Selling price less all variable cost, leaving the amount available to cover fixed cost and create profit.
Examples
- An average price of 173,000 dong a case less variable cost of 100,400 dong gives a contribution of 72,600 dong, or 42 percent.
- Variable cost is materials at 82,000 dong, variable conversion at 14,000 dong and freight at 4,400 dong.
Easy to get wrong
Contribution margin always reads higher than gross margin, which makes it easy to misuse in pricing. An order at 130,000 dong a case still contributes, but a whole year sold that way leaves the 132 billion dong of fixed factory cost uncovered.
39Break even pointĐiểm hòa vốn
Definition
The volume at which total contribution exactly covers total fixed cost and profit is zero.
Examples
- Fixed cost of 340 billion dong divided by contribution of 72,600 dong a case gives a break even of 4.68 million cases.
- Actual volume is 5.5 million cases, so volume can fall 15 percent before the company loses money.
Easy to get wrong
Break even moves when sales mix moves, not only when fixed cost moves. Selling more through modern trade at lower prices cuts average contribution and pushes break even above 5 million cases.
40Net working capitalVốn lưu động ròng
Definition
Inventory plus receivables less payables, the money tied up in running the business day to day.
Examples
- Inventory of 27.6 billion dong plus receivables of 60 billion less payables of 54 billion gives net working capital of 33.6 billion dong.
- That 33.6 billion dong is 3.5 percent of the 950 billion dong annual revenue.
Easy to get wrong
Cutting working capital by stretching suppliers only moves the problem. A supplier pushed from 30 to 60 days prices that interest back into the material at the next negotiation.
41Days sales outstandingSố ngày phải thu khách hàng
Definition
How many days of average revenue are sitting in receivables not yet collected.
Examples
- Average receivables of 60 billion dong against annual revenue of 950 billion gives 23 days.
- Distributors pay in 15 days and carry 71 percent of revenue, modern trade pays in 45 days and carries 24 percent, weighting out to 23 days.
Easy to get wrong
The average hides the problem accounts. The headline is 23 days while 4 distributors are more than 60 days overdue on 6.2 billion dong, and that balance is the actual risk.
42Cash conversion cycleChu kỳ chuyển đổi tiền
Definition
The number of days between paying a supplier and collecting the cash from a customer.
Examples
- Inventory days of 15 plus receivable days of 23 less payable days of 30 gives a cycle of 8 days.
- Shortening the cycle from 8 days to 3 releases roughly 13 billion dong of cash.
Easy to get wrong
A shorter cycle is not always better. Cutting inventory from 15 days to 8 pulls the cycle to 1 day but drops delivery in full, and the revenue lost usually exceeds the cash released.
43Cost to serveChi phí phục vụ
Definition
The full cost of getting product to one customer or one channel, covering warehousing, freight, the sales team and trade support.
Examples
- A distributor buying 16 billion dong a year costs 640 million dong to serve, 4 percent of its sales.
- A modern trade chain buying 96 billion dong a year costs 11.3 billion dong to serve, 11.8 percent of sales, driven by 6 percent listing and display fees plus direct delivery to 540 stores.
Easy to get wrong
A big customer is not automatically a cheap one to serve. The 96 billion dong chain buys at 158,000 dong a case, 12,000 below the distributor price, and costs almost three times as much to serve as a percent of sales.
44Cost of qualityChi phí chất lượng
Definition
The total of what is spent to prevent defects plus what has to be paid once defects occur.
Examples
- The annual total is 14.2 billion dong, 1.5 percent of revenue, made up of 2.1 billion prevention, 3.8 billion appraisal, 5.4 billion internal failure and 2.9 billion external failure.
- A single product recall costs 480 million dong and sits entirely in external failure.
Easy to get wrong
Cutting prevention and appraisal raises total cost of quality rather than lowering it. Saving 600 million dong of incoming inspection can buy several batch holds at 128 million dong each and one recall at 480 million.
45Return on capital employedLợi nhuận trên vốn sử dụng
Definition
Operating profit divided by the capital put into the business, being fixed assets plus net working capital.
Examples
- Operating profit of 59.3 billion dong on capital employed of 298.6 billion gives 19.9 percent.
- Capital employed is fixed assets of 265 billion dong plus net working capital of 33.6 billion dong.
Easy to get wrong
This ratio improves from either side. Adding 5 billion dong of profit lifts it to 21.5 percent, while taking 30 billion dong out of inventory and receivables lifts it to 22.1 percent without selling a single extra case.