The Mathematical Fallacy of Using Historic Data for 2027 Growth Projections
Planning 2027 from historic data can create false confidence when the market has already changed. This article explains why fragile forecasts create fragile quotas, and how leaders can test assumptions before a sales compensation problem
By Compswell —
We are only just entering H2, but this is also the point in the year when many organisations begin to look backwards and plan forwards at the same time. H1 performance is being reviewed, H2 gaps are being discussed, and somewhere in the background, the first assumptions for 2027 growth are already beginning to take shape. That is where the risk starts. Most 2027 planning conversations will begin the same way many 2026 and 2025 planning conversations began. Someone will open a spreadsheet, pull the last two or three years of bookings, add the current year forecast, draw a growth line through the numbers, and apply a reasonable sounding ambition for next year. The line will look clean. The logic will look familiar. The board will see a number that can be explained in one slide. Then the quota will be built on top of it. The problem is not the spreadsheet. The problem is the assumption sitting quietly inside it: that the conditions which produced the historic numbers will still be in place when the 2027 targets are supposed to land. That assumption is becoming harder to defend. A deal that closed in 2023 may have moved through a very different buying environment from one being worked in 2026. The earlier deal may have had a smaller buying committee, faster budget approval, less internal scrutiny, and fewer competing priorities. The same type of customer today may take longer to decide, involve more stakeholders, require stronger business justification, and compare more alternatives before signing. At the same time, AI is beginning to change the sales environment in ways historic data cannot fully explain. Some sellers are already using AI to research accounts faster, draft follow ups, improve outreach, summarise calls, and prepare for customer conversations. Some companies are using AI to score pipeline, forecast risk, identify next best actions, and improve account prioritisation. Buyers are also using AI to compare vendors, challenge pricing, review proposals, and involve internal teams more efficiently. This means the market is not simply continuing from the past. Parts of the system are changing. That is why using historic data as the main engine for 2027 growth projection can create false confidence. The numbers may be accurate, but the world that produced them may no longer be the same world the sales team is selling into. Historic data is still useful. It tells you what happened, where the business has performed well, where capacity turned into revenue, and where attainment patterns have repeated. But it becomes dangerous when it is treated as a clean forecast of a market that has already started to behave differently. For sales compensation, this is not a theoretical problem. The growth projection becomes the quota. The quota becomes the basis for pay opportunity. The pay curve, accelerator, territory model, and OTE expectation all sit downstream of that number. So when the growth assumption is weak, the compensation plan inherits the weakness. That is how a forecasting error becomes a sales compensation problem. The number is not only the growth percentage approved for 2027. The more important number is the quality of the assumptions behind that growth percentage. Recent planning discussions across many B2B organisations are happening against a difficult backdrop: longer buying cycles, more cautious budget approval, pressure on win rates, stronger scrutiny from finance teams, and higher expectations for forecast accuracy. Add AI into that environment, and the challenge becomes even sharper because productivity may improve unevenly across teams, roles, markets, and customer segments. One team may use AI well and create more qualified pipeline with the same headcount. Another may use the same tools poorly and see very little improvement. One region may benefit from faster account research and better customer targeting. Another may still be constrained by long procurement cycles and complex stakeholder approval. One seller may become more productive because AI improves preparation and follow up discipline. Another may simply produce more activity without improving conversion. A historic average will not see those differences clearly. It will look at past bookings, past attainment, and past conversion, then produce a number that feels objective because it comes from data. But data can still mislead when it combines different market conditions into one smooth trend. That is the problem with relying too heavily on 2022, 2023, or even 2024 performance to price 2027 ambition. Those years may still contain useful lessons, but they may not represent the buying conditions, productivity tools, competitive pressure, or customer behaviour that will shape the next plan year. The question is no longer simply: what did we grow by before? The better question is: which parts of the past still describe the market we are about to sell into? The Principle Historic data should inform the plan, but it should not become the plan. When leaders use historic data well, they separate durable patterns from outdated conditions. They ask which performance trends still hold, which assumptions have weakened, and which parts of the business are already operating under a different market reality. They look beyond bookings and attainment to understand win rate, sales cycle length, pipeline creation, stage conversion, deal size, product mix, customer segment, territory maturity, ramp time, and seller capacity. When historic data is used poorly, the past is treated as one continuous line. That is the mathematical fallacy. A linear projection assumes that the environment producing the numbers has remained broadly stable across the period being measured. But a market with changing buyer behaviour, uneven AI adoption, shifting productivity expectations, economic uncertainty, and new pay transparency obligations is not one stable environment. It is a set of different regimes. A faster cycle market with easier budget approval is one regime. A slower cycle market with larger buying committees and stronger financial scrutiny is another. A pre AI sales motion is one regime. A sales motion where some sellers are supported by AI enabled research, forecasting, coaching, and outreach is another. Blending those regimes together may create a neat average, but a neat average is not the same as a reliable growth projection. This becomes a structural issue in sales compensation because quota sits directly downstream of the growth model. When the model assumes stability that no longer exists, the quota inherits that false stability. The pay curve inherits it. The accelerator structure inherits it. Territory expectations inherit it. By the time the plan reaches the seller, the original modelling weakness has been converted into an individual performance target. That is where trust starts to weaken. The seller is not arguing with a spreadsheet. They are reacting to whether the number feels connected to the market they are actually selling into. The AI Factor AI deserves a specific place in the 2027 planning conversation because it can change both productivity and expectations at the same time. On one side, AI may improve sales productivity. Sellers may prepare faster, personalise outreach better, identify buying signals earlier, summarise customer meetings more accurately, and spend less time on administrative work. Managers may use AI to detect pipeline risk earlier, improve coaching, compare deal patterns, and challenge forecast assumptions before the quarter is already lost. On the other side, AI can also create planning mistakes when leaders assume productivity gains before they are proven. A company may look at AI adoption and decide that each seller should carry a higher quota in 2027 because the tools are expected to make the team more efficient. But if the AI mainly improves activity volume and not win rate, the quota increase will be built on hope rather than evidence. The real question is n