When the makeup counters closed in 2020, the damage didn't show up in one place, it showed up several branches deep in the revenue model. Here's how to draw a driver tree that shows where a change hits your buyer, and which lever is yours.
A value driver tree links an outcome metric, like revenue, to the input metrics that drive it, branch by branch. In sales, you draw one to find where a change hits your buyer's business, which input you can credibly move, how far it sits from revenue, which other paths you're competing with, and the story each path lets their executive tell, so your point of view matches how they already run the business.
Placeholder answer
Short, practical frameworks for executive messaging, business cases, multithreading, and sales enablement.
For decades, a big part of how makeup got sold was a woman at a department store counter, testing a shade on the back of her hand, then on her face, and asking the person behind the counter what they thought.
Then, in the spring of 2020, the counters closed.
Testers came off the shelves everywhere, because nobody wanted to share a lipstick with a stranger in the middle of a pandemic. By the end of June, roughly one in five stores in Estée Lauder's Americas business was still closed. And with half the world in masks, the products that depend most on being seen took the hardest hit: makeup sales fell 18% that fiscal year, while skin care grew 13%, and in the worst quarter, total sales dropped 32% (fiscal 2020 results).
At the time, I was selling a data platform that tracked emerging companies and technologies (think Crunchbase) to innovation teams inside large enterprises. We learned fast that data alone didn't create much value for our customers, because a list of startups is only useful if you know which one matters to your business, and why. So we wrapped advisory services around the platform, recommending specific initiatives tied back to each customer's business model. Their whole job was to find the biggest threats to the business, fend them off, and think about the future alongside their executive team.
Picture doing that job for Estée Lauder in the spring of 2020. Online sales in the region grew double digits for the year and made up 40% of sales, and as CEO Fabrizio Freda put it, "We quickly pivoted to capture consumption online during COVID-19 as retail stores around the world temporarily closed." But the website had the same problem the counters had quietly solved for decades. You can't try on lipstick through a screen, so shoppers showed up, couldn't tell whether a shade would work on them, and left.
Which makes the initiative you'd recommend pretty clear: stand up virtual try-on, so a shopper can see a shade on her own face before she buys it. (Estée Lauder did move fast here, by the way. It went on to build a virtual try-on platform that let its brands launch try-on in as little as ten weeks, down from about six months.)
Here's a simplified version of how that looks, drawn out:
- Stores: ~1 in 5 still closed (Americas) - Travel retail - Online: growing double digits - Traffic - Conversion: where makeup is bleeding - Bounce rate - Shade confidence: the input the pandemic broke - Average order value
A driver tree links "outcome" metrics to a series of related "input" metrics, simplifying a complex goal into a set of cause-and-effect relationships. The left side is what you'd like to influence. The right side is how you'll drive that change.
Finance teams have used them for decades (you'll hear them called KPI trees, or value driver trees), because it's how an executive team actually runs a business. Which is exactly why you want to learn to draw one: it lets you think the way the person signing the contract already thinks.
It's also how you diagnose what's really going on. If revenue's down, is it because fewer people are buying, or because the same people are spending less? For Estée Lauder, the tree makes it plain. Stores closing hit the far left side, but the damage online lived several branches to the right, in one very specific input: shade confidence. That's the node virtual try-on could move.
It's rare to sell something that moves a number right next to revenue. Most deals involve agreeing on why a driver a few levels to the right is the right path to a first-level outcome, and since your message gets more indirect, and more complex, the further right you go, every link has to be clear, specific, and believable.
If you say, "Better website UX drives cross-sells!" Well, sure, that's probably true. But it's less clear why that path gets the buyer to the outcome better, faster, and more reliably than another approach, like adding more inventory to match what their top-spending customers are leaving to buy from a competitor.
Same with Estée Lauder. A CFO looking at makeup down 18% and skin care up 13% could reasonably ask: why fix makeup conversion at all? Why not pour everything into the category that's already growing? Your path doesn't just have to work. It has to beat that.
It makes you agree on the approach before the product, which is the step most executive messaging skips (the soundbite article covers why that skip is so costly). So before you talk about any features, get everyone aligned on the higher-order questions: What's the right driver to focus on? What must be true about how we impact that driver? That's less "use our content automation," and more "shift your mix of leads from paid to organic sources."
Then ask, "What's the story this exec wants to tell?" followed by, "How can I make sure they get to?" The point of view you've built tells you what's changing, and the tree tells you where it hits, and which lever is yours.
Grab a deal where you already know what's changing. By the end, you'll have their revenue model on one page, the one path you can credibly move, and the story it lets their executive tell.
1. Draw their revenue model, by hand. Start with revenue on the far left, break it into its first-level drivers, and keep branching right, two or three levels deep. Use AI to fill in the numbers, not to draw the tree:
I'm drawing a driver tree for [COMPANY]'s revenue model, and I want it in their language, not a textbook's.
Start with total revenue. Break it into the first-level drivers [COMPANY] actually uses in its own 10-K, investor presentations, and earnings calls, using their exact metric names. Then go two levels deeper on each branch.
For every node, give me the latest reported number, the trend over the last 4–8 quarters, and a source.
Then flag two things:
1. Any metric they used to report and quietly stopped, or redefined. That's usually where the pain is. 2. Any node their executives keep bringing up on earnings calls, and the exact words they use about it.
Then put the pen to paper yourself, because the drawing is the thinking. If you let a model draw it, you'll get a pretty tree and no idea where the change actually hits. "Revenue = customers × price" could be any company. "Net sales = active customers × net sales per active customer," using the exact metric names from their 10-K, is their business, and they'll recognize it.
2. Mark where the change hits, then highlight your path. Find where the change lands on the tree, circle the one input you can credibly move, and trace the line from that node back to revenue. Then count the branches. If there are three or more, every link needs to be specific enough that an executive could check it. "Our platform drives digital revenue" skips every step in between. "Virtual try-on raises shade confidence, which cuts bounce, which lifts conversion on the channel that's now carrying the business" takes one checkable link at a time.
3. Map two or three other paths they could take. Your customer isn't choosing between you and nothing. They're choosing between you, a handful of completely different approaches, and doing nothing at all. So get those on the page too:
Play the CFO of [COMPANY]. Here's the change hitting your business:
[paste the change from your point of view]
Here's the outcome metric you're on the hook for:
[the left-side node you traced back to]
List the 3–5 paths you'd realistically weigh to move that number. Include doing nothing, building it in-house, and at least one path that has nothing to do with buying software.
For each: which input on the driver tree it moves, how long until it shows up in the numbers, what it costs, and what could go wrong.
Then tell me which one you'd actually pick, and why. Be honest, even if it's not mine.
Then answer the question that matters on your own: which path gets them to the outcome fastest? If another path really is faster, you'd rather know now than in a closed-lost note. For Estée Lauder in 2020, the paths on the table probably looked something like this:
4. Write the story each path lets them tell. Every path turns into a different sentence an executive gets to say to their board, or on their next earnings call:
Now look at which one they'd want to say out loud. That's the story you're selling, and your product is just how it comes true.
5. Draw it with them. On your next call, bring the tree, or better yet, draw it live together. Then ask the question that turns it from your framework into theirs: "What did I get wrong?" You'll be confirmed or corrected, and both are wins. Either way, update the tree, and you've got a picture of their business nobody else on the deal has.
I know this can feel a little obscure. Drawing out a customer's revenue model sounds like homework for their finance team, not yours. But it works.
Kishan's deal with one of the largest e-commerce sites in the US (I told the full story in the soundbite article) turned on a node just one branch from revenue, and it handed their executives good news for Wall Street. Most of the time, like at Estée Lauder, you'll be further out, which is exactly why the tree matters.