Profile picture for user Andrew Nguyen

Movies and novels are examples of where unpredictability is welcomed. But this doesn’t hold true in marketing. We like a little bit of predictability when it comes to the revenue we can expect from our campaigns.

Murder mysteries and detective stories that keep us watching or reading use unpredictability as a hook. But that shouldn’t apply to your marketing reports. Let’s leave mystery to the Hollywood writers, and focus on helping guide your organization to better decision making through forecasting.

In this post we show you how to turn your marketing data into a forecast so you can make good budgeting decisions and deliver actionable information to the leadership team.

Because forecasts are living and breathing numbers, changing as current information gets inputted, forecasts can help you meet your sales opportunity and revenue goals. You’ll know whether you’re likely or unlikely to hit your target, which is essential information for decision making and decision makers.

Step 1: Make Sure Your Data Collection Process Is Correct

You can’t build a house on a rocky foundation. And in forecasting you depend on a data collection process that is accurate and consistent. In finance for example, if you’re looking at stock prices each week, you better be sure that the weekly stock prices are collected in a consistent manner and that they are indeed the actual prices from that week.

Yes, I did just say to make sure the weekly stock prices are the weekly stock prices, that the data is clean. Sounds pretty basic but you’d be surprised at how difficult this can be in marketing.

Is 200 leads in January really 200 leads? Or does that number change depending on which platform you used to count?

In B2B marketing, the customer journey is long and thus they engage with a variety of channels and as they convert on different channels there is more room for miscounting leads. This can drastically affect your forecast if you don’t use a consistent counting method.

To avoid this, be sure to have a centralized system that tracks conversion information from all your channels and campaigns (deduping and matching leads to accounts), tracking the entire journey from anonymous first touch through the opportunity and closed-won stage.

Bad data means forecasting based on erroneous metrics. Essentially, you’re doomed from the start.


revenue forecast


Step 2:  Set Your Target Goals

A forecast is made of time series data which is simply made of three types of data: the time of the observation, metric you want to forecast and the metric(s) that predict it. For example, if we’re forecasting revenue then you’d look at revenue in January, revenue in February, etc. You’d look at a metric that predicts revenue, like Sales Opportunities.

Sounds pretty simple, but deciding on the forecast method and deciding on metrics after revenue also requires attention.

The further up the funnel you go, for example if you want use web traffic to predict revenue, this may be more difficult because there are many factors that can affect the conversion of web traffic. There are many changes, such as how much you spend on ads one month compared to the next, making it difficult to understand the relationship between traffic and revenue.

So what kind of metrics do you put in your spreadsheet?

Start with your target or commit goals. We’ve written about how to work backwards to set target Opportunity and MQL goals based on Revenue goals.

Ensure that you’re using metrics that marketing has the power to change, are reasonably predictable over time, have consistent influence on financial outcomes and have clear definitions across teams.

Step 3: Decide On The Forecast Model

There are several types of forecast models. Actually there are hundreds depending on what industry you’re in.

So how do you decide which model to use? Answer: the one that most accurately predicts real revenue.

The most important thing to remember is that the forecast only works if it predicts how much actual revenue you generate. There are several kinds of forecasts models such as:

  • Average Approach: Take the mean or average of all your past data, and that’s you’re forecasted number.

  • Naive Approach: Take value of the previous observed value. For instance if last month you generated 10K in revenue then this month you’ll generate 10K in revenue.

  • Drift Method: Essentially, drawing a line between the first and last observation and extrapolating out. The drift method takes into account the average change between observations.

  • Moving Average: Creating averages from subsets of observations and shifting them forward, essentially smoothing out short term fluctuations.

  • Historical conversion rates: Use the historic conversion rates of opportunities and apply that to the open opportunities to forecast revenue. For example if your historic opportunity conversion rate is 50% and you have 2 opportunities in your pipeline then your forecast is 1 deal X average deal size. This is how Bizible forecasts your revenue.

    Projected Revenue = (Opportunity Count X Opportunity Conversion Rate) X Average Deal Size

  • Predictive Scoring: This approach takes into account variable conversion rates based on channel or prospect information. It’s like lead scoring except for revenue. It looks at prior accounts that are closed-won to determine the predictive factors that account for the prospect becoming a customer. Bizible also uses this method for its account-based marketing feature, predicting which accounts are likely to close using your customer data.

So which method should you choose? Whichever was most accurate in predicting the revenue you made last month! If the Naive Approach accurately predicts real revenue, there’s no reason not to use it.

Step 4:  Take Action After Comparing The Forecast To Your Target Goal

If your forecast tells you you are unlikely to hit this month’s goal then it’s time to take action. But where to begin? How do you know if you should generate more leads or focus on converting opportunities at the bottom of the funnel?

Start by identifying the bottlenecks in your funnel where you have a large pool of prospects who aren’t converting. Next, choose a multi-touch attribution model to inform you on which corrective action to take. You want to choose the attribution model that includes the touchpoints of the funnel stage of the bottle neck. This is very important. Let me explain.

For instance if the bottleneck is at the very bottom of the funnel and your opportunities aren’t converting into closed deals you’ll want to use a Full Path Attribution Model that tells you which Opp to Sales channels are generating the most revenue/conversions.

If you see that there’s a bottleneck at the top of the funnel then you’ll want to use U-Shaped Attribution to see which channels or campaigns are converting people into leads most efficiently.

You’ll use W-Shaped Attribution if you need guidance on middle of the funnel conversion rates, i.e what activities or channels are converting MQLs into Opps.

These attribution models are basically a prescription for widening a bottleneck in your funnel and telling you where you need to spend in order to reach your revenue goal. You can thank your forecast for giving you the wake up call.

Getting The Attribution Models

Guess what kinds of multi-touch attribution models you gain access to with Bizible? All the previously mentioned models.

You now have everything you need to approach revenue forecasting like a pro. Remember, it’s not so much about the future as it is about having the information you need to make good decisions.

Case in point. If your forecast model isn’t accurate then you’re leaving your immediate decisions up to chance in the hopes that you hit your target. If you have an accurate forecast and you aren’t pacing to hitting your goal then you need multi-touch attribution models to help you decide on the channels and activities you can spend on to help you reach your target.

Forecasting and attribution are all about action.

So, how do you set high targets, and understand how to reach them? By using multi-touch attribution models and carefully choosing your forecasting models.