Attributes Of A Good Model
Hopefully by now I’ve convinced you why building a model is worthy of your time; now let’s discuss some best practices and attributes of excellent startup forecasts. The best startup financial models are:
Logical: The architecture, assumptions, and inputs used should all be logical, with the model accurately reflecting your business and its economics. To achieve this, build the model bottom-up—for example, number of salespeople X monthly quota X price. Also, try to base your key business metrics on proxies from “real world” companies or established benchmarks. For example, let’s say you have an ad supported website. It’s generally not too difficult to search for average CPM data (or play around with Adsense) and figure out what the going ad rates are for content sites in your market. For other key inputs, you might ask an advisor or investor—someone who sees multiple businesses in the space and has a feel for typical metrics.
Reasonable: Here, we look at things like margins (gross, operating, and net), revenue growth (and rate of increase of that growth), hiring plan, etc. Take the outputs and filter them with simple sanity checks; e.g., are your margin forecasts in line with those found in the 10-k’s of analogous, publicly-traded companies in your space? For example, if your startup is a SaaS business, are you showing Salesforce-type numbers? Based on your full-year unit sales forecast, is your implied market share percent reasonable, or does it show you’ll own 80% of the market? Is the model showing a required funding amount that that you could realistically raise?
Simple: Good models are readily understood by model users or future model developers. This is especially important when we develop a model that will later be handed off to a startup CEO. Likewise, good models distill the number of key business metrics down to just a handful of inputs; this takes discipline, and a desire to model only that which really matters. There is beauty in simplicity; try to avoid over-engineered, complicated and unwieldy tools that will ultimately frustrate model users.
Navigable: Related, good models have an intuitive navigation system and well designed layout. The more complex a model, the more important this becomes. In addition, a clean presentation signals a solid and logical architecture—I can usually tell within seconds if the formulas are spaghetti code simply by looking at the format and presentation.
User friendly: We like to separate the model into three sections: Inputs (Assumptions), Calculations (Logic) and Outputs. Our favorite approach is to create one main page for (almost) all inputs, which we call the “dashboard;” this is then followed by the revenue buildup and monthly / quarterly / annual rollup. We also employ a few tricks to make it even more user-friendly, such as having all inputs be in a blue font, all hard-coded entries in red, and all formulas in black. This way, the end-user knows exactly what (s)he can play around with, and what not to mess with.
Sensitivity & Scenario Analysis Capabilities: as discussed in the “Why build it” section, having robust scenario analysis capabilities is a critical feature of most forecast models to facilitate decision making; this is particularly valuable for startups still figuring out the optimal business model. In many cases this takes the form of visuals such as charts and graphs; for example, a line graph is a nice a way to visually show where the revenue and cost lines cross, i.e. where we start to make a profit on each user or customer (which ultimately is what gets investors excited to pour more fuel on the fire), and where the ’search for business model’ crosses over into ‘execution of the business model.’
Good Output Page: Finally, in most cases we will want to extract some data from the model to present to investors, lenders, partners, etc. in the business plan or investor presentation. However, we rarely want to send the entire model over—the burden is on us to present the data in an appealing, digestible manner. For this reason, good models have a very clear summary page that might contain the key business metrics alongside such tables as Summary Income Statement, hiring plan, customer growth line, breakeven point (in units and/or number of customers), total capital required, and tables showing per-unit economics such as average revenue per user (ARPU) and cost-to-acquire (CPA) metrics over time.
Ok, that’s about it for the moment. I’ve run out of steam on my mission to evangelize startup financial models.
I now turn this over to you—what value have you received from working on your financial forecast? What “model hacks” have you found that you’d like to share?