Learn Mark Kosoglow’s 4-Step Forecast for Sales and Success
One of the most crucial, yet difficult skills to master as a business leader is forecasting. It is both an art and a science that every leader should feel excited about getting right. And this article is meant to do precisely that—to reduce forecasting stress with a proven strategy that works for Sales and Customer Success.
Last month, Catalyst Chief Revenue Officer Mark Kosoglow hosted a private executive forum with 100 customer success and sales leaders to share his formula to manage renewal forecasts with precision—the one that helped him consistently forecast within 3% on global revenue teams, meaning 97%+ accuracy quarter over quarter.
In this article, we’re excited to share his exact framework with you.
A strong forecast rests on three pillars
Did you know that roughly one quarter (24%) of companies forecast accurately today? Plus, 55% of leaders report their organization regularly misses forecasts by 10% or more.
That means most people aren’t even close to getting their forecast right.
The good news about a massive gap like this? This is your biggest opportunity to drive more wins in this environment if you approach forecasting correctly.
6 foundational truths about forecasting (and yes, there’s more to come):
- Don’t rely on one forecast model to make calls from the top.
- Don’t make forecasting calls alone without your team’s support.
- Do commit to building the processes and opportunity rigor from the bottom up.
- Do commit to building a shared forecast methodology for both Sales and Success.
- Don’t think it’s going to be easy, this requires work. Don’t lose faith!
- Don’t get caught fighting an upstream battle at the end of the forecast period.
The most common offender among these truths is that companies often only use one forecast method—and that just leaves too much room for error. If you’re in Sales, would you strictly rely on Marketing for your entire pipeline?
That’s why Mark lives by the belief (and practice) that forecasting requires at least three vectors to pay attention to at all times. When speaking with Dan Heath on his bestseller, Upstream, Mark felt he could finally put his forecast practice into words…
“The upstream process of how you set yourself up determines the downstream forecast you will present.”
- Mark Kosoglow, CRO Catalyst
Model 1: Build "Bottoms Up" starting with each renewal and expansion deal
In a practical sense, Bottoms Up forecasting is when you take all deals across all accounts—net new sales + customer renewals + customer expansion—and add everything up to produce a forecast number.
This is what’s really happening on the field that no AI or model is going to tell you. Let’s talk about the critical components of getting this right.
There must be a standardized, shared way of managing opportunity stages, documenting risk, and putting pipeline into the forecast that can be mastered at the individual level and managed rigorously at the frontline leader level. A few ways this can be achieved, with more expanded on in our template include:
a. Exit Criteria: Clearly define a highly structured methodology around what needs to be fulfilled to consider a deal has exited one stage and moved to the next.
👉 If you’re a VP of Account Management, you should know–without having to confirm with your team–that every upsell moved to “Green/Commit” is firmly in the forecast, without feeling that knee jerk reaction to confirm directly.
b. Mutual Action Plan: Clearly define your mutual action plan that all sellers (including renewals) must go through with the account. These steps need to be fulfilled for the customer or prospect to realize your unique value deeply.
👉 Think of your structured mutual action plan like a roadmap. It allows your team to align on a process for themselves and understand how you perceive deal progression.
Once you lock down a consistent methodology, you can coach people more easily; once you can coach more easily, your team will model the right behaviors more consistently; and when deals are managed consistently, you can finally perform a strong top down review of what’s actually happening on the field.
Overall, this adds up to the biggest unlock: consistent forecast data.
The Sales and Success Pitfalls of Inconsistent Forecasting
Over eight years at Outreach, Mark produced a forecast of 97%+ accuracy. How did he do it? He attributes Bottoms Up to being the #1 reason.
“The real unlock is consistency. When everyone similarly does things, the data can actually be used at the deal/opp level which makes Bottoms Up a key vector of forecasting. Every week, we check the data with every single deal, look at the stage and close date on the forecast category, on what our risk assessment field looks like. And then determine what we can do to keep that deal on track?” - Mark Kosoglow, CRO, Catalyst
A couple of notes for data consistency in your shared methodology:
Don’t overcomplicate green/yellow/red by deal type.
- If they’re a customer or prospect, your forecast language should be uniform. This allows you to quickly tally everything up and not require multiple different dashboards to get a holistic forecast view.
Don’t overcomplicate green/yellow/red by segment.
- If they’re enterprise, you shouldn’t have to look into a matrix or legend of where the forecast category fits into that segment.
- Have a clear mutual action plan based on deal type/segment/industry but for the love of the executives looking at this from 10k feet above, keep the categories the same.
At the end of the methodology rainbow, this is the pot of gold: all deals roll up into a uniform green/yellow/red forecast that leaders can efficiently review and accurately forecast from.
Now, before you raise a glass to celebrate, Bottoms Up is the most critical piece to get right in forecasting and it certainly requires work. However, we have some good news with regard to making that effort:
- Mark’s full Bottoms Up forecast template can be downloaded at the end of this article so you don’t have to start from scratch.
- Once you get Bottoms Up forecasting right, 50% or more of the heavy lifting will be done to begin reaping the rewards.
Model 2: Model-Driven Forecasting
The second forecast vector is model-driven. Most of us have or currently use one of three types: a financial cohort, AI-based predictions, and a bespoke data-driven model.
Here’s a rundown of where we stand with model-driven forecasting.
❌ Stay away from Financial Cohort analysis
A financial cohort model is never fun to work with for the following reasons:
- It’s very rigid, inflexible, and has a limited scope.
- Historical cohort models don’t account for changes in the market.
- Historical cohorts also don’t take into account real-time customer behavior.
- There’s limited applicability as your business scales because it can only apply to people with the same situation in the past.
To expand on that last point, if you sell only into manufacturing and try moving into consumer packaged goods, your cohort based model doesn't apply to this new industry and deal motion.
So, in a nutshell, your forecast misses all the relevant data that's happening internally in your business and externally in the market.
That’s why we recommend staying away from this one altogether.
⚠️ Proceed with caution on AI-Based Forecasts
AI forecasts mostly look at factors like time in stage, the number of calls made, the number of meetings/email activity, and then weight how much of an impact these inputs have on close rates, deal velocity, etc. Then, this data gets simulated and tested thousands of times to develop a prediction. Finally, this data is applied to your forecast.
The problem with that is it could be more transparent. You can't see how it's really working. Plus, the biggest input for data quality is historical. It all comes back to how good your Bottoms Up data is to have trust in the mysterious AI prediction.
Here’s our brief assessment of AI forecasting:
The only way AI works is to learn from historical data to make assumptions about what might happen in the future–this causes a similar problem to the Financial Cohort analysis trap.
- If historical Bottoms Up data isn't tight, the AI cannot learn to forecast correctly.
- If your Bottoms Up data is tight, proceed cautiously and verify the prediction (so yeah, still put in some manual work).
Consider a machine-learning algorithm for stock trading. If it has been trained using data only from a period of low market volatility and high economic growth, it may not perform well when the economy enters a recession or experiences turmoil—say, during a crisis like the Covid-19 pandemic. As the market changes, the relationship between the inputs and outputs also may change.
Source: Harvard Business Review
✅ Use a Data-Driven Model for Your Specific Business
A data-driven model
You aren’t slapping AI onto your current model here. However, you can speed up your efforts with our data-driven Catalyst Play that walks you through how Heap has blazed this trail using Catalyst Health Scores.
Here’s what this play will unlock for you:
- How to use customer data you already have to make data-driven predictions on churn, expansion, & health
- How to build a data-rich predictive health score for customers (including inputs to download here)
- Understand how Heap operationalizes this information to predict renewals with 95%+ accuracy
- How to get started when replicating this Play within your business.
Overall, data-driven models are often built for your specific business, industry, ideal customer profile, target segments, and more. That means they’re not easily stood up overnight—but they are highly flexible and accurately aligned to your specific business as it evolves.
- One big forecasting truth to call out here: Your Product and Customer Success team should lead this charge because they are the most informed about signal risk, expansion, and more—too many other teams involved could clog the process.
Catalyst keeps our team focused on customers’ outcomes by highlighting risks and opportunities from day one. I use its data to power my executive meetings and hold our company leaders accountable for critical outcomes.
- Ahmed Quadri, CCO, Heap
Model 3: Top-Down Forecast
Top-Down forecasting is a discussion usually during annual or quarterly planning. The key here is how you manage it and how you apply it.
Even in difficult times, a Top-Down approach is always helpful to show a gap in the business and determine the performance optimizations needed to fill it.
You should know what’s possible with capacity, what pipeline you have to work with, what the data inputs are signaling, and what resources you should ask for if you need to multiply your performance—and actually hit plan!
Top-Down forecasting notes:
- Use Top-Down goals to influence better performance from the bottom up.
- Use Bottoms Up and data-driven inputs to handle forecasting with clarity and accuracy.
- During forecast reviews, always show a worst-case 5% margin to plan for the unexpected (e.g., extremely healthy customer but went bankrupt and didn’t renew).
- Don’t set yourself up for failure. Question your targets before accepting.
If your CFO wants you to double your performance, use your data to assess the inputs needed.
Bonus vector: your gut. Yes… YOU!
Mark’s final word (and bonus vector) is… trust your gut! Whatever fancy models say, remember YOU hold the power with this information. Your gut instincts are often telling you something; pay attention.
If something feels off in your forecast, go investigate that area and pinpoint what's not sitting right. That’s it. The thing about listening to your gut is that it’s useless without data. But once you’ve conducted a thorough analysis, your gut is a powerful final layer on top of your forecasting data.
Now, go download Mark’s Bottoms Up forecast template to get started 👇