What if your model didn’t just say what happened, but showed you what to do next?

Most early-stage companies are stuck playing defense. One surprise, and suddenly… the whole model needs reworking.

That’s why we wrote 📙 Scenario Planning for Early Stage Companies, a tactical guide to help you model tradeoffs like: Raise now or later? Hire fast or hold? Grow or preserve cash?

Download it here.

"Where is everyone?!"

I was walking the front line of the operation in the last week of the financial year. It should not have been this quiet.

The plant manager (a man of few words) responded: "One week shutdown."

"What... why?"

"To get inventory down."

"That doesn't make any sense. Has demand dropped?"

"No, we're flat out. We'll have to run overtime next week to replace the lost production."

"Why the hell are you doing it then?"

He explained the instruction had come from above. Their regional operations head had told them to do it, which had come from their business unit (BU) leadership team.

It suddenly made sense. This particular BU had blown out their profit target for the year (even though this particular plant had done well). The only part of their bonus left to salvage hinged on hitting an inventory target.

So the instruction that had come down from Darren and his team (yes, the same Darren) had been to take whatever actions necessary in the last week of the year to hit a certain inventory target by plant.

This wasn't the right thing to do at any level. But at a plant level, it led to some insane behavior, including this example of disrupting the business, and shelling out a fortune in overtime and rework to catch up from an unnatural shutdown. Customer service would suffer. Hell, it would even throw off our corporate financials as it would disrupt sales in the final week.

I was furious. With Darren for ordering it. With the plant manager for not challenging it. But most of all with myself, for building a bonus scheme that could lead to such dumb behavior on the front line of the operation.

Welcome to part 8 of this 9-week series, diving into FP&A. You can catch up with the rest of the series here.

This week, we will dive further into bridging and variance analysis. Picking up where we left off last week. If you didn’t read last week’s piece, now would be a good time to correct that unthinkable mistake.

As a quick recap, last week we covered:

Today, we will cover some other important principles for pulling together quality bridges:

Then we will wrap up with some ways you can bring this all together and use these principles practically in your business.

We’ve all been there. Your procurement team tells you proudly they “saved $100k.” Only to find out that you are actually paying $50k more than you were 6 months ago.

It would be more accurate for them to say: “we were at risk of paying $150k more, but we have managed to mitigate $100k of that.”

Having an agreed baseline to calculate savings against is important. They need to be measured, and savings are hard to measure, because what is the right base? Especially through the last 3-4 years of exceptional inflation.

Is a cost price increase of 7% in a market that has inflated by 10% a good or bad result? It’s not an easy question to answer. Especially if that market price whipsaws back again:

But you need some control over how you measure the savings. Clear principles… or it will get abused. Using the budget process to either accept or reject any inflation risks into the base numbers allows you to precisely define what your basis is for calculating any savings..

And price isn’t the only dynamic on cost. There is usage/volume, too. Which is often controlled in a completely different part of the business.

I once saw a buyer try to claim a saving on toilet paper procurement after ‘negotiating’ the cost per pack from $12.00 down to $6.50. How did he achieve this miraculous feat? He moved from buying packs of 96 to packs of 48.

I ended up having to explain it to him in terms of ‘cost per wipe.’ And then used it as an example for the rest of the procurement team to embarrass them into being honest about their savings calculation: ‘don’t be cost per wipe guy.’ I think his nickname is still ‘CPW’...

Is a cost overspend always bad? No.

Take Novo Nordisk: they almost certainly overspent on R&D in recent years. But without that, they wouldn’t have developed Ozempic, the breakthrough drug that made them the most valuable company in Europe. That’s good overspend. It’s not just a cost, it’s capital allocation into an investment.

A useful way to think about it: overspends above the maintainable free cash flow (MFCF) line are typically bad as these are conversion inefficiencies that eat into margin. Overspends below the MFCF line can be good if they drive returns. In that case, the overspend is actually a strategic acceleration.

Importantly, this isn’t about where the spend hits in the P&L. It’s about the nature and intent of the spend. Good overspend should be shown in a bridge alongside its associated return, so you see the red of the investment next to the green of the outcome.

Good overspend must be ringfenced, and the returns against it must be tracked. That’s the only way to prove it was actually good.

Not all P&L impacts are created equal. Some are recurring, i.e., they’ll continue month after month. Others are one-time. Think: a ‘one-and-done’ event. Examples of one-time impacts: insurance rebate, fire-related disruption, legal settlements, project costs, M&A transaction fees (the list is endless)…

Why this matters:

Let’s unpack that last point with a 2x2:

One-Time Adverse Impacts: Ringfence these, confirm they are genuinely one-off, and avoid them clouding your assessment of run-rate performance

One-Time Positive Impacts: Call these out explicitly. Prevent the business from “free-riding” on a temporary windfall, and make sure weak underlying performance isn’t masked by a short-term gain.

Recurring Negative Impacts: These signal a structural issue. You need to prioritize mitigation. And fast. Consider budgeting for them if unavoidable.

Recurring Positive Impacts: Decide whether to lock in (protect) the gain or reinvest it strategically. Either way, treat them as part of the new baseline.

Each of these scenarios demands a different response from the business, so it must emerge from the financial storytelling. And that can only happen if they’ve been well identified on the bridge.

This is a close cousin of the one-time vs recurring distinction, but it’s not the same. Sometimes, the event you expected does happen, just not when you expected (timing difference). Other times, the shift introduces a permanent change in financial structure or performance.

You expected a one-off sale in August. It slips and lands in October instead.

If all else is equal, this is a neutral variance over time: August shows an adverse variance, October shows a favorable variance, and the total year impact is zero.

Ringfencing these is important, because they can create unnecessary noise on a bridge. In practice, monthly bridges are full of timing noise, especially where cut-off timing (like delivery date or contract sign date) shifts performance across periods.

And you don’t want the business to over-react to timing differences. They are much less important than structural issues. As long as you are confident they are temporary and self correcting.

PVM analysis could easily warrant a full post of its own, and one day, it will.

For now, the key takeaway is this: isolating price, volume, and mix variances in your bridge is essential because these effects cascade throughout the P&L. A volume variance versus prior year or budget will ripple down into revenue, gross margin, and any variable opex.

As discussed last week, the nature of the variance is more important than where it lands in the P&L. So, instead of having volume-driven impacts scattered across multiple lines, strip them out and present them clearly.

I like to visualize the concept of PVM using a simple chart:

Note - I don’t normally use a chart like this in practice, it’s more a device for explaining the concept.

In this simple single product, you have a favorable price variance and an adverse volume variance vs budget.

By plotting one against the other, you can clearly isolate the price and volume elements of the variance.

Now extend this to a multi-product business. You’d have multiple charts like this, one for each product, stacked on top of each. When you aggregate them, you get the total price and volume variance. But a third factor also emerges: mix. The effect of changes in weighting between the products.

Mix is often misunderstood or, worse, misused. If you ever see a line in a bridge labeled “mix/other,” it should be a yellow flag. Mix effects are real and can be significant. But “mix” should never be used as shorthand for “we didn’t really understand what happened.”

While this example focuses on price, volume, and mix on the sales line, the same logic applies to cost of sales. That means when you're analyzing margin movement, you're juggling changes in sales volume, sales price, cost price, and product mix. It gets complex fast.

One of the core challenges with any multi-layered bridge is identifying and eliminating variance overlap.

When you’re dealing with cost prices, volumes, sales prices, mix effects, cost efficiencies, one-time items, and recurring shifts, the risk of double or even triple counting is very real.

This is not an exercise in perfect math. No bridging exercise is. As we said last week, your job is to extrapolate your understanding of the business into its drivers, and then quantify those drivers as best you can.

The key is to establish a clear set of principles for how you identify and isolate variances, and then to make sure your math holds up. If this feels easy, you are probably doing it wrong.

Even in that simple bridge example from last week, there were multiple valid ways to present the answer, and different approaches to allocating cost or margin across the drivers.

Let me explain using a simple example:

A common mistake is calculating each variance using actuals. For example:

Each method is technically valid in isolation. But if you add them together, you double-count the interaction between price and volume. Alternatively, using only budgeted values as the multiplier will eliminate (or double count, depending on the direction) part of the real variance, too.

And this is just with two of the simplest variables. It doesn’t take long to see how complex this gets across a full bridge. That is why clear principles and tight math are critical.

There is no universal right answer to how you do this. But there probably is a right answer for your business and your dynamics.

There is no generic template for good variance analysis. Many have tried, many have failed. Good variance analysis starts with the business. YOUR business. You can template it for your business, but only once you understand it.

The principles outlined in this post and the last post are just that. Principles. How you build them into your bridge-building process is deeply business-specific and subjective. But as long as you have given proper thought to the principles outlined and how they apply to your business, you can’t go too far wrong.

To help bring this together, I am sharing an example of what I would consider a good bridge. This should help you see the thinking that connects the analysis to the storytelling. Yours could look very different, its the application of the principles that is important.

Let’s say we are bridging year-over-year EBIT. 2024 EBIT was $1,000, and 2025 EBIT was $1,000. Flat.

But under the hood, there is much more happening:

You’ll notice I like to see related variances grouped together. It makes the storytelling cleaner.

First up, you normalize the base:

That gets you to an adjusted 2024 base.

Next, you roll over the full-year effects of actions taken in 2024. We covered this last week. This takes us to a 2024 exit rate position, i.e., a ‘clean’ 2024 full year adjusted to account for the impact of anything that happened during the year.

Now you can start building up the things that have happened in 2025:

Volume and mix effects (total to a $150 adverse effect):

This distinction is important because it makes clear that this significant effect is driven by an issue that appears to be localized to one customer. Rather than a more endemic sales issue.

Next up, you look at price impacts (net $50 benefit):

Now let’s look at marketing spend ($50 adverse year on year):

And cost efficiencies show a different story ($200 worse):

Finally, we can see the business benefited from a large one-off rebate, which we isolate from the core operating performasnce.

Pulling the story together:

If I saw this bridge, I’d be asking: why can’t this business execute its big moves? The day-to-day is strong. But the high-impact initiatives (pricing, capex, growth investment) are where value is leaking - and its hurting.

That’s a change management problem, not a financial one. Fix that, and EBIT starts to grow.

Note how we started with a simple statement (Actual of $1,000 vs Last Year of $1,000). And ended up with a simple story. But to get there, there was a LOT of thinking and math in between. That is the formula for good variance analysis. It is often complication to get to just a simple story, explaining a simple variance.

Once you carve up the P&L the right way, the story reveals itself. But that only happens if the analysis is built bottom-up and grounded in real operational understanding. Top-down pontification almost always leads to garbage bridges, unless it’s anchored in a solid grasp of the actual business drivers.

Next week, we’ll dive into the highest-stakes use cases for these bridges: earnings calls, board meetings, and investor updates. We’ll explore how to walk the line between telling the story you want to tell and staying firmly on the right side of truth and your ethical code.

:::::::::::::::::::::::::::::::::::::::::::::::::::
:: Thank you to our sponsor ::
:: RUNWAY ::
:::::::::::::::::::::::::::::::::::::::::::::::::::

If you enjoyed today’s content, don’t forget to subscribe.

Disclaimer: I am not your accountant, tax advisor, lawyer, CFO, director, or friend. Well, maybe I’m your friend, but I am not any of those other things. Everything I publish represents my opinions only, not advice. Running the finances for a company is serious business, and you should take the proper advice you need.