Top down vs bottom up: lender-ready forecasting for founders
For a Canadian lender-ready business plan, the professional standard is a hybrid approach: use top-down forecasting to establish your market opportunity, then use bottom-up forecasting to prove you can actually repay the loan. Neither method alone will satisfy a bank.
Here is what each method needs to deliver in your file:
- Top-down: A credible TAM/SAM/SOM narrative that shows the market is large enough to carry your revenue targets
- Bottom-up: Unit-level cash flow projections built from real conversion rates, average order value (AOV), and customer counts that feed directly into your Debt Service Coverage Ratio (DSCR)
- Documented assumptions: Every input figure needs a source, a date, and a brief rationale
- Sensitivity analysis: At minimum, a three-scenario stress test showing how a moderate drop in a key driver affects your ability to service debt
Canadian lenders are not impressed by a large TAM on its own. Bankers view pure top-down claims as unconvincing when they are not backed by unit economics that demonstrate repayment capacity.
Table of Contents
- What is the difference between top-down and bottom-up forecasting?
- When should you use top-down vs bottom-up?
- How to build a bottom-up forecast lenders will accept
- How to build a defensible top-down market size for Canada
- How to reconcile both forecasts into a single lender-ready submission
- Common forecasting mistakes that make lenders nervous
- A worked Canadian example: retail services business
- How AI tools speed up lender-ready forecasting
- Key takeaways
- The gap between the two forecasts is where the real work happens
- LenderReady builds your hybrid forecast in 15 minutes
- Useful Canadian sources for TAM and input validation
What is the difference between top-down and bottom-up forecasting?
Top-down forecasting starts with a total market size and works downward, allocating a share of that market to your business to arrive at a revenue estimate.
Bottom-up forecasting works in the opposite direction. It aggregates unit-level inputs to build total revenue from the ground up. The core formula lenders want to see is:
Or, for product businesses: Price × Quantity sold (or orders × AOV).
| Dimension | Top-down | Bottom-up |
|---|---|---|
| Starting point | Total market size (TAM) | Individual unit drivers (customers, AOV) |
| Primary use | Strategic planning, market entry | Operational cash flow, loan repayment |
| Data required | Industry reports, Statistics Canada | CRM data, transaction history, pilot results |
| Lender credibility | Context only | Primary evidence for DSCR |
| Time horizon | Annual / multi-year | Monthly / quarterly |
| Key risk | Over-optimism on market share | Garbage-in, garbage-out from poor inputs |
The conceptual difference is the starting point. Top-down tells a story about the market; bottom-up tells a story about your operations.
When should you use top-down vs bottom-up?
The choice depends on your planning horizon and the quality of your data.
Top-down is best for long-term strategic planning and market entry when you have little or no transaction history. If you are pre-revenue or entering a new market, a TAM narrative gives lenders and investors a sense of scale before you have real numbers to show.
Bottom-up is the right tool for monthly and quarterly cash flow planning once you have any operational data at all. Even a 90-day pilot, a handful of paying customers, or a comparable business’s public financials can anchor your unit assumptions. Bottom-up is more effective at surfacing local performance issues that a top-down model would simply average away.
Practical checklist for choosing your primary method:
- Pre-revenue, no pilot data → start top-down, supplement with comparable unit benchmarks
- Early revenue (under 12 months) → bottom-up using actual transaction data, even if the sample is small
- Established business seeking growth capital → bottom-up primary, top-down for market context
- Seasonal or lumpy revenue → bottom-up with monthly timing; see guidance on seasonal revenue modelling
Pro Tip: If full bottom-up coverage feels impractical for every product line, use a middle-out hybrid: build bottom-up for your top two or three revenue drivers and apply a top-down allocation to the rest. This balances granularity with speed.
How to build a bottom-up forecast lenders will accept
Start with the inputs that actually move cash. Lenders want to see the drivers, not just the totals.
- Segment your customers by type (new vs. returning, channel, geography) and estimate the count for each segment per month.
- Assign conversion rates from lead to paying customer, sourced from your CRM, pilot data, or published industry benchmarks.
- Set your AOV and purchase frequency using transaction history or comparable market data.
- Map the sales ramp, how many months before a new customer reaches full purchase frequency?
- Time your cash receipts, when does revenue actually hit your bank account relative to the sale date?
- Calculate monthly DSCR using net operating income divided by total debt service for each month in your projection period.
| Input field | What lenders expect | Where to source it |
|---|---|---|
| Monthly new customers | Segmented by channel | CRM, ad platform data, pilot results |
| Conversion rate | Lead-to-sale percentage | CRM history, industry benchmarks |
| Average order value (AOV) | Per transaction, not blended | Transaction records, comparable businesses |
| Repeat purchase / churn rate | Monthly or annual | Subscription data, cohort analysis |
| Sales cycle length | Days from first contact to payment | CRM pipeline data |
| Cash receipt timing | Days from invoice to deposit | Accounts receivable aging report |
Pro Tip: Keep the model limited to the drivers that materially move cash, customer acquisition cost, conversion rate, AOV, and churn. Avoid SKU-level detail unless it changes your DSCR materially. A model with 40 tabs is harder to defend than one with four.
For sensitivity testing, build a simple three-scenario table showing your base case, a 15% drop in conversion rate, and a 20% drop in AOV. Show the DSCR result for each scenario. Stress-testing assumptions this way demonstrates repayment capacity under downside conditions, which is exactly what a Canadian lender needs to approve a file.
How to build a defensible top-down market size for Canada
TAM/SAM/SOM is the standard framework. Here is how to make it credible for a Canadian lender file.
- TAM (Total Addressable Market): Use Statistics Canada census data, BDC industry reports, or sector-specific association data to establish the full national or provincial market size in dollars.
- SAM (Serviceable Addressable Market): Narrow TAM to the geographic and demographic segment you can realistically reach. A Toronto-based service business serving SMEs is not competing for the entire Canadian market.
- SOM (Serviceable Obtainable Market): This is your realistic share in years one through three. Justify it with pilot conversion rates, local penetration estimates, or competitive density analysis.
Pro Tip: Never claim a percentage of TAM without a capture pathway. “We will capture 2% of a $500M market” means nothing to a lender unless you explain exactly how you acquire those customers, at what cost, and over what timeline. Tie your SOM directly to your bottom-up customer acquisition model.
For local businesses, market sizing at the neighbourhood or city level is often more persuasive than national TAM figures. A lender reviewing a Halifax café does not need a national food-service TAM; they need to know how many households are within a 3 km radius and what the average spend per visit looks like.
How to reconcile both forecasts into a single lender-ready submission
The most defensible funding submissions run both models in parallel and reconcile them in a structured review. Here is the workflow:
- Run your top-down model independently and record your SOM revenue estimate for year one.
- Run your bottom-up model independently and record your projected year-one revenue.
- Quantify the gap between the two figures.
- Investigate the gap: does it reflect an unrealistic conversion assumption, a missing customer segment, or an overstated market share claim?
- Adjust assumptions in both models until the gap is explainable and documented.
- Present both models in your plan with a one-paragraph reconciliation note.
Assumptions lenders want to see documented:
- Customer acquisition cost (CAC) and the channel it applies to
- Conversion rate source and date
- Sales ramp timeline and rationale
- Retention rate or churn assumption and basis
- Pricing rationale and any planned changes
A large divergence between your two forecasts is a diagnostic signal, not a problem to hide. Investigating it often reveals the single most important assumption in your model. Show the lender you found it and addressed it.
| Reconciliation step | Top-down output | Bottom-up output |
|---|---|---|
| Year-one revenue estimate | SOM × assumed capture rate | Customers × AOV × frequency |
| Primary risk | Overstated market share | Poor input data quality |
| Lender focus | Market opportunity narrative | DSCR and repayment capacity |
| Sensitivity driver | Market growth rate | Conversion rate and AOV |
Common forecasting mistakes that make lenders nervous
Top-down over-optimism is the most common red flag. Claiming 5% of a billion-dollar market in year one, with no acquisition plan, tells a lender you have not done the operational work.
Bottom-up garbage-in, garbage-out is equally damaging. If CRM and transactional data are clean, bottom-up yields the most accurate short-term forecast; if the data are poor, the model is fiction dressed as precision.
Other errors lenders flag:
- Revenue timing that ignores accounts receivable lag (booking a sale in month one but receiving cash in month three)
- No seasonality adjustment in a clearly seasonal business
- A single-scenario forecast with no sensitivity table
- Missing DSCR calculation or a DSCR below 1.25x with no explanation
Pro Tip: Before submitting, run a lender scoring check against your plan. Lenders weight DSCR, collateral, and assumptions documentation heavily. A plan that scores well on narrative but poorly on financial rigour will stall at the credit desk.
A worked Canadian example: retail services business
Scenario: A Toronto-based mobile dog grooming business applying for a $75,000 BDC loan.
Top-down estimate:
- Toronto pet services market: approximately $180M annually (Statistics Canada retail trade data, NAICS 812910)
- SAM (mobile grooming, Toronto): estimated $22M
- SOM (year one, single operator): a small percentage of the total addressable market, resulting in a realistic first year revenue estimate
Bottom-up build:
| Month | New clients | Returning clients | Total sessions | AOV, | Monthly revenue |
|---|---|---|---|---|---|
| 1 | 12 | , | 12 | , | , |
| 3 | , | 20 | 38 | , | , |
| 6 | 22 | , | , | , | , |
| 12 | 20 | , | , | , | , |
Year-one bottom-up total: approximately $72,000
Reconciliation: The $16,000 gap between the top-down SOM ($88,000) and the bottom-up projection ($72,000) flags a realistic constraint: client ramp-up takes longer than the market-share assumption implies. The bottom-up figure becomes the loan repayment basis. At $72,000 annual revenue with $38,000 in operating costs, net operating income is $34,000. Monthly debt service on a standard loan amount at typical interest rates over typical terms, giving a DSCR well above the minimum Canadian lenders require.
How AI tools speed up lender-ready forecasting
AI accelerates the parts of forecasting that eat the most time: data consolidation, scenario generation, and assumptions documentation.
Specific tasks where AI adds real value:
- Pulling and formatting Statistics Canada or BDC data into a TAM table
- Generating three-scenario sensitivity tables from a single base-case input
- Calculating DSCR across monthly periods automatically
- Drafting the assumptions narrative that accompanies your financial tables
Validating AI outputs is non-negotiable. Always attach the source citation for every market figure, keep your raw input tables separate from the model output, and show your assumptions list to the lender as a standalone document.
Pro Tip: AI is fastest when your inputs are clean. Spend 30 minutes organising your transaction history and CRM data before you start. The model output is only as credible as the numbers you feed it.
LenderReady’s AI generates the bottom-up model, sensitivity tables, and DSCR calculations through a conversational Q&A, then exports a lender-ready PDF. Operating businesses can add a source-backed Application File before submission.
Key takeaways
A hybrid approach combining top-down market context with a rigorous bottom-up unit economics model is the standard Canadian lenders expect in a fundable business plan.
| Point | Details |
|---|---|
| Use a hybrid approach | Top-down establishes market opportunity; bottom-up proves repayment capacity through unit economics. |
| Document every assumption | Source, date, and rationale for each input, lenders reject undocumented figures. |
| Calculate DSCR scenarios | Stress-test a 10–30% drop in conversion rate and AOV; show DSCR stays above 1.25x. |
| Treat divergence as a signal | A gap between your two forecasts reveals the assumption most worth investigating before submission. |
| LenderReady speeds the process | LenderReady’s AI builds bottom-up models, sensitivity tables, and DSCR calculations in 15 minutes. |
The gap between the two forecasts is where the real work happens
Most founders treat a divergence between their top-down and bottom-up numbers as an embarrassment to smooth over. That is the wrong instinct. The gap is the most useful thing your model produces.
When I look at a funding file and the TAM-derived SOM is 30% higher than the bottom-up projection, that difference is pointing at something specific: an acquisition cost that has not been accounted for, a sales cycle that is longer than assumed, or a churn rate that has been ignored. Founders who investigate that gap and document what they found are the ones whose files move forward. Founders who average the two numbers and hope no one notices are the ones who get a call asking for more information.
The practical habit worth building: run both models before you finalise any forecast, write one paragraph explaining the gap, and attach it to your assumptions table. It takes 20 minutes and it signals to a lender that you understand your own business. That signal matters more than a polished cover page.
LenderReady builds your hybrid forecast in 15 minutes
Getting the hybrid model right, documented assumptions, reconciled forecasts, DSCR scenarios, is exactly the kind of work that takes founders days to do manually and still comes back from the lender with questions.
LenderReady’s AI walks you through a conversational Q&A that captures your unit drivers, builds the bottom-up model, runs sensitivity scenarios, and calculates DSCR automatically. The output is a lender-ready PDF with the financial tables and assumptions documentation Canadian banks expect. If you are already operating, the Financial Statement Scan adds source-backed readiness findings before you submit. Start with a free Business Plan Check to see where your current forecast stands, or generate your full plan today.
Useful Canadian sources for TAM and input validation
When you cite market data in a lender file, the source matters as much as the number. Here is where to look:
- Statistics Canada, census data, retail trade surveys, and NAICS-coded industry revenue figures. Use for TAM and SAM calculations; always cite the table number and reference year.
- Business Development Bank of Canada (BDC), industry benchmarks, small business financial ratios, and sector reports. Lenders respect BDC data because it is Canada-specific and regularly updated.
- Industry associations (e.g., Restaurants Canada, Canadian Federation of Independent Business), sector-specific spending and margin benchmarks useful for AOV and cost assumptions.
- Municipal open data portals (e.g., City of Toronto Open Data, Vancouver Open Data), local population, business licence, and foot traffic data for SAM and SOM calculations.
- Paid reports (IBISWorld Canada, Mordor Intelligence), useful for niche markets where Statistics Canada coverage is thin; always note the report date and methodology.
When presenting sources to a lender, include the table or page reference, the date the data was published, and a one-sentence note explaining how you converted the market estimate into your specific share assumption. Attaching a screenshot or extract as an appendix item removes any doubt about where the number came from.
This article provides general information for educational purposes and does not constitute financial or professional advice. Confirm current lending criteria and program details with your lender or a qualified financial adviser.
See where your financing file stands
Fifteen questions, four minutes, no documents and no email. You get a readiness stage, the gaps a lender would raise, and the document list for your request.
Check my readiness, freeLenderReady is an educational service, not a lender, broker, or financial advisor. Lending criteria vary by institution and change over time; treat this as a starting point, not a guarantee.