If you are responsible for forecasting in a B2SMB company right now, you have my sympathy.

I’ve been there. 40 years worth.

But now not only feels different to me it is different.

The variables that matter most to forecast performance simply cannot be estimated from historical data.

And yet Wall Street demands quarterly guidance. Boards demand annual budgets. Sales teams need quotas. So you forecast anyway.

Stuck in the Forecaster’s Dilemma.

Three B2SMB companies reported Q3 2025 earnings. Two tried to forecast quarters in a world where quarters are unforecastable. One found a different approach that actually works.

The difference? The two that missed were forecasting quarterly results. The one that beat was forecasting where the market was going.

HubSpot: The Hidden Miss

HubSpot beat Q3 2025 estimates. Revenue of $810 million versus $786.94 million expected. So why did the stock fall 1.54%?

Because the guidance revealed what the headline number hid: growth was decelerating from 18% in Q3 to 16% in Q4 (constant currency). That is moving the wrong direction.

What HubSpot said: Strong customer acquisition. Net revenue retention will improve in Q4.

Between the lines: HubSpot is adding customers (up 17% year over year) but cannot monetize them. Average Subscription Revenue Per Customer grew only 3%. That is a 5.6:1 ratio of customer growth to revenue-per-customer growth. You are supposed to grow revenue per customer faster than you grow customers. That is how you prove pricing power.

This is the forecast failure: HubSpot’s 2025 model assumed they could raise prices as they added AI features. What they discovered is that when Salesforce announces a major OpenAI partnership (October 14, 2025) and Microsoft Copilot becomes the AI standard embedded in every enterprise’s existing tools, suddenly your new AI features are not differentiated. They are table stakes. And table stakes do not command premium pricing.

Here is what makes this variable nearly impossible to forecast: You cannot predict when your competitor will announce a partnership that redefines what baseline capability means. Most companies miss this not because their models are bad, but because competitive bar shifts are fundamentally unforecastable using historical data.

HubSpot forecasted growth. They are getting growth. But they are not getting the monetization they forecasted, and that monetization gap is the real story.ADP: When the Model Breaks

In April 2025, with a quarter yet to go in their full fiscal year, ADP lowered its annual revenue growth guidance from 6-7% to 5-6%.

This is textbook forecast failure: late-year revision because reality did not cooperate with the model.

What ADP said: International weakness + HRO softness in the domestic market.

Between the lines: ADP’s forecasting model was built for stability. Mid-market companies decide on HR tech implementations based on hiring plans. Hiring plans are predictable when economic conditions are stable. They are not predictable when companies are facing simultaneous pressures from AI-driven productivity debates (do we need more headcount or better tools?) and uncertainty about whether their customers will still exist in 12 months.

The stock dropped 5.3% after Q3 earnings, despite beating estimates. Why? Because beating the quarter does not matter when your guidance reveals you cannot predict the next quarter.

Here is what ADP’s model missed: Mid-market companies do not delay HR decisions because of macroeconomic uncertainty. They delay because the ground is shifting under their feet. When your customers are questioning whether they need your category at all (does AI eliminate the need for some of what we do?), historical patterns stop predicting future behavior.

Shopify: Forecasting the Market, Not the Quarter

Shopify posted 32% revenue growth in Q3 2025. GMV up 32%. International revenue now 49% of total (Europe up 37%). B2B on its sixth consecutive quarter of 100%+ GMV growth. Free cash flow margin: 18%, nine consecutive quarters of double-digit FCF margins.

Then they guided Q4 growth to mid-to-high 20s. Conservative. Guiding down from 32%.

This looks like every other company managing expectations. It is not.

What Shopify is doing differently: They are not forecasting quarters. They are forecasting where commerce is going, building for that future, then guiding quarters conservatively.

The bets Shopify made years ago:

In 2022, they launched B2B on Shopify, betting that B2B and B2C commerce would converge.

In 2023, they divested logistics (Deliverr) to refocus on the core commerce platform, betting that trying to compete with Amazon on fulfillment was a distraction from their real advantage.

In 2024, they launched Markets (unified international/retail/B2B expansion in one interface), betting that merchants wanted one operating system for all channels, not fifteen different tools.

Also in 2024, they integrated with Roblox and Perplexity, betting on agentic commerce (AI assistants that shop for you) before anyone else was talking about it.

Every one of these bets took 18-36 months to pay off. B2B did not hit 100%+ growth overnight. International did not become 49% of revenue in one quarter. These are multi-year theses about where commerce was going.

Between the lines: Shopify’s Q3 2025 beat was not luck. It was the result of forecasting correctly in 2022-2023 about where the market would be in 2025. They bet on unified commerce when everyone else was still selling point solutions. They bet on international when others were still optimizing for North America. They bet on B2B when B2B e-commerce was still an afterthought.

And here is the kicker: Their Q4 guidance (mid-to-high 20s) is conservative not because they are unsure about the market. It is conservative because they have learned that the right strategy is to forecast the long-term market direction, build for it, and then guide quarters cautiously so you beat and raise.

What This Reveals About Forecasting

Most B2SMB companies forecast the way ADP and HubSpot do: extrapolate recent trends, adjust for known variables (seasonality, new product launches, pricing changes), add a buffer, and call it guidance.

This works fine when the world is stable. It breaks completely when:

Your customers are rethinking whether they need your category (AI is not just automating tasks, it is eliminating job functions)

Your competitive position shifts faster than your product roadmap (a major partnership announcement redefines what table stakes means)

Your market structure is changing (B2B commerce is no longer a separate category from B2C, so your forecasting model based on B2C only is now incomplete)

Shopify does not avoid these problems. They just forecast differently. Instead of asking what next quarter will look like based on last quarter, they ask where commerce is going in 3-5 years, then work backward.

That does not mean their quarterly forecasts are more accurate. It means they care less about quarterly forecast accuracy because they are optimizing for a different goal: be positioned correctly for where the market will be, not where it is today.

The Uncomfortable Truth

Here is what the data reveals: forecasting in B2SMB right now is impossible if you are trying to forecast quarters accurately.

The variables that matter most (will AI eliminate this job category? will our biggest competitor get acquired by Microsoft and suddenly have unlimited distribution? will our customers decide to consolidate vendors and we are number three on the list?) are not predictable from historical data.

But Wall Street demands quarterly guidance. Boards demand annual budgets. Sales teams need quotas. So companies forecast anyway, knowing the model is built on assumptions that could become obsolete in sixty days.

The smart ones do what Shopify does: forecast the market direction, build for that, and guide quarters conservatively so you have room to beat. The honest ones do what some companies are starting to do: stop giving annual guidance entirely and admit that visibility is too limited.

The ones in trouble are still pretending their model works. That their pipeline coverage gives them confidence. That they have accounted for competitive dynamics. They have not. They are using yesterday’s tools to predict tomorrow’s market.

What to Watch For

As we head into 2026 guidance season, here is how to read between the lines on forecasts:

Companies that give specific guidance but wide ranges (example: 15-20% growth) are telling you they have no visibility. Narrow the range or widen the honesty.

Companies that cite macro uncertainty but give precise guidance are pretending to have a model that works. If you genuinely do not know what the macro will do, your guidance range should reflect that.

Companies that guide conservatively, beat consistently, and raise guidance incrementally are playing the game correctly for a world where quarterly forecasting is theater. Watch whether they are also making multi-year bets that could redefine their position.

Companies that miss guidance mid-year and attribute it to one specific factor (like ADP blaming international weakness) are giving you the polite explanation. The real explanation is usually that their model was built for a world that no longer exists.

The lesson from Q3 2025 earnings is not that some companies are better at forecasting than others. It is that the ones who beat are forecasting different things. Most are forecasting quarters. The winners are forecasting markets.

And in a world where AI can shift competitive dynamics in 90 days, M&A can redraw market boundaries overnight, and your customers are questioning whether they need your entire category, forecasting markets is the only kind that matters.