What it means
The method starts with things you can count or observe directly. A retailer might begin with the number of stores, the customers per store per day and the average spend per customer.
Multiplying and adding those inputs produces a revenue figure that rests on real operations, not on a market report. It is the opposite of top down analysis, which starts with a large number, such as the size of the whole market, and then assumes a percentage share.
Top down is quick and gives a sense of scale, but the share assumption is often a guess. Bottom up forces the analyst to explain exactly how each dollar will be earned.
In finance, the approach appears in budgeting, startup financial models, sales forecasting and company valuation. Analysts building a revenue forecast may use sales reps, quota per rep and conversion rates.
Investors who value a company bottom up may add up the worth of each division, asset or customer segment separately. The strength of the method is traceability.
Every number in the model can be linked to a driver, such as a price, a headcount or a production capacity, and each driver can be tested against real data. When results disappoint, managers can see which input was wrong instead of arguing about a vague percentage.
The weakness is effort and false precision. A model with fifty detailed inputs can look very exact while still being wrong if a few key assumptions are off.
It can also miss the market ceiling, so a bottom up forecast that implies a huge share of the market should be checked against a top down view. Experienced analysts therefore use both approaches and compare them.
If the two answers are close, confidence rises. If they are far apart, the gap points to an assumption that deserves more scrutiny.
In practice
Real-world examples
Example 1
A software start-up forecasts next year's sales by listing its 8 sales reps, a quota of $500,000 per rep and an expected 70% quota attainment. It reaches $2,800,000 of bookings and uses the figure in its budget. Each input can be checked against last year's results.
Example 2
A hospital group plans a new wing and builds its income estimate from the number of beds, the expected occupancy rate and the average revenue per bed day. The finance team then compares the result with regional demand figures. The two views are close enough to approve the project.
Example 3
A private equity analyst values a food manufacturer by estimating each product line separately, with its own volume, price and margin. The sum of the parts gives a total value that she presents to her investment committee, with a clear explanation of each assumption.
Think of it
Bottom-up starts from the details and builds up-analyzing individual pieces first.
Formula
Calculation
Annual revenue = Number of outlets x Customers per outlet per day x Average spend per customer x Trading days
Suppose a coffee chain plans 12 stores, each serving 150 customers a day. The average spend is $40 per customer and each store trades 300 days a year.
Daily revenue per store = 150 x 40 = $6,000. Annual revenue per store = 6,000 x 300 = $1,800,000.
Total annual revenue = 12 x 1,800,000 = $21,600,000. If the store count rose to 15 with the other inputs unchanged, revenue would be 15 x 1,800,000 = $27,000,000.Case study
Seen in the real world
Greenfield Logistics is a fictional courier start-up seeking funding. Its first pitch used a top down claim that the delivery market was worth $10,000,000,000 and that winning just 1% would mean $100,000,000 of revenue. Investors were unimpressed, because the 1% had no support.
The founders rebuilt the forecast bottom up. They counted vans, deliveries per van per day, the average fee per delivery and the number of working days, and they added drivers and depots in stages as cash allowed.
The illustrative result was a far smaller number, but each line could be traced to an operating decision. Investors preferred it, and the company raised money on the strength of a plan they could test.
Watch out
Common mistakes
- Building a very detailed model on weak inputs. Precision in the arithmetic does not make the assumptions correct.
- Forgetting the market ceiling. A bottom up forecast should still be checked against the size of the total market to see whether it is plausible.
- Treating bottom up and top down as rivals. Good analysts use both and investigate any large difference between them.
Questions
People also ask
Is bottom up analysis more accurate than top down?
It is usually easier to defend because each number has a source, but it is only as accurate as its inputs.
Where is it used outside forecasting?
It is also used in costing, budgeting and valuation, for example when adding up the costs of each task in a project.
How detailed should the model be?
Detailed enough to capture the main drivers of the result, but not so detailed that nobody can maintain or review it.
Watch