The Model Isn't the Analysis. The Analysis Is Everything You Cut Out of the Model.

The first DCF I ever built had 17 tabs. I remember because I counted them, proudly, when I handed it to my new Boss. He didn't open it. He asked me what the three things were that mattered most to the valuation. I started to explain the WACC build. He cut me off. "No. The three things about the business."

I couldn't answer him. I'd spent two weeks modeling a company I didn't understand.

That was the first time someone pointed out the trap I'd walked into, though it wouldn't be the last. Junior analysts, and I was very much one of them, tend to confuse complexity with rigor. A 17-tab model feels like work. It looks like work. It produces a number at the end, and the number has decimal places, and somewhere along the way those decimal places start to feel like confidence. They aren't.

The best DCF I've ever seen fit on a single sheet of A4. It belonged to a long-only PM in London whose fund I won't name. Three revenue drivers, one margin assumption, a reinvestment rate, a terminal value, and a discount rate. No working capital schedule. No separate lease line. No quarterly phasing. When I asked him why he didn't model working capital properly, he gave me an answer I still remember almost word for word: "Because over ten years, on this business, it's worth about 30 basis points of IRR and I can't forecast it to better than 200." He had thought about it. He had decided it didn't matter. Then he had left it out.

I don't think I really understood what he was doing at the time. I thought he was being lazy, or maybe just confident enough to get away with it. It took me a few more years to realize he'd done more analytical work on that sheet of paper than most 300-line forecasts I'd seen. He'd done it in his head, before he opened Excel, and the empty rows were the proof.

Think about the tax line on a typical sell-side model. Analysts will build out an effective tax rate schedule country by country, because they can. For most businesses, the gap between modeling tax at a flat 22% and running a full jurisdictional waterfall is a rounding error against the error bars on the top line. Meanwhile the revenue assumption, which is the thing that actually drives the whole valuation, gets one row and a gut feel. We spend our effort where our skills are, not where the sensitivity lives.

Now, before I touch a cell, I try to run the same question through my head. If I'm wrong about this by twenty percent, does the answer move enough for me to care? If it doesn't, I pick a number that feels directionally fine and I stop thinking about it. That's not laziness, although it used to feel like laziness to me when I first tried it. It's triage. The scarce resource I have as an analyst isn't computation or Excel wizardry or footnote literacy. It's attention. And attention spent on a jurisdictional tax build is attention I'm not spending on whether the business actually has pricing power, which is usually the thing that will decide if I'm right.

I sometimes think about this the same way I think about writing. The first version of anything I write is everything I could say about it. The useful version is what's left once I've taken out the lines that don't earn their place on the page. The paragraphs I cut often did more for the final draft than the ones I kept, because cutting them forced me to work out what the piece was really about.

Models aren't so different, in the end. The empty space in a good one is the tell. It means the analyst got their hands dirty with the business before they let themselves start building, and then made deliberate choices, with reasons they could articulate, about what to leave on the cutting room floor. So when I flip open a DCF now and see 47 tabs staring back at me, I don't think "thorough." I think about my own 47 tabs, and I get a little worried on the analyst's behalf.