Workday WDAY

The judgment

During the SaaS selloff, Workday looked to me like a rare case where the market had all three wrong: magnitude, immediacy, and inevitability. I thought the selloff overstated the damage AI would do to revenue and earnings, and underestimated how much had to happen before customers would replace the platform.

I had been screening SaaS companies for a pattern of positive earnings, revenue, and cash flow, with a separate check on debt relative to enterprise value. What drew me to Workday was customer dependence. I was looking for software with deep roots in the organization using it.

Replacing an HR platform meant a long runway of management meetings, approvals, budgeting, staff involvement, and implementation planning. Even if AI could produce a capable replacement, the customer would still have to take responsibility for running it. The company would effectively have to become an HR software company on top of whatever it did as its day job.

My own experience with AI made me skeptical of how easy that would be. Simple games and applications I had tried to build were disjointed and needed hours of handholding. I had also seen AI overlook explicit instructions and reference material, then acknowledge the mistake when I pushed it. That shaped my assessment of how much supervision was still required, particularly for software dealing with employees' money and compliance obligations.

I gave weight to Workday's experience serving customers: the relationships, the knowledge accumulated through those interactions, and the familiarity with laws and regulations. I expected that experience to remain valuable as software became easier to build.

When the price became compelling

The size of the drop and the resulting P/E were the aha moment. I considered what historical earnings multiples suggested might be reasonable. I read earnings transcripts to judge how much of management's outlook was optimism and how much was realistic. I compared past guidance with actual earnings and looked at the differences. Signed contracts for revenue yet to be recognized also mattered.

I opened a small canary position on February 9. Having a small position keeps me more engaged than having a stock on a watchlist and mentally prepares me to act. By February 23, my thesis hadn't changed, but the price had. At an average of $126.69, I was ready to commit substantially more capital.

I expected the market to reassess how quickly, how extensively, and how certainly AI would damage Workday's business, and the stock to recover part of its fall.

Trading the position

I maintained my underlying view of Workday while trading the price swings. I looked at historical trading ranges and volume, and used retracements as a barometer for the move. At times I reduced the position to little more than a canary, then rebuilt it when the price became attractive again.

After the February 23 build, I sold shares over the next three trading days. I also bought again, rebuilding part of the position before reducing it to a canary on March 2. The first sequence was profitable.

I rebuilt during March and added in early April, then increased the position substantially on April 9 at an average of $111.54. It was the same theory. The price drop increased my perception of the margin of safety and the profit potential. On April 15, I sold most of the position at an average of $124.34. The earlier purchases had been at higher prices, so the rebound from the April addition does not describe the result of the whole sequence.

Further additions and reductions followed through the summer. By the time I sold shares near $194.61 in August, the position was much smaller than it had been earlier. That later price cannot be used to describe the return I earned on the original large purchase.

What the record establishes

The available records show profitable realized trading, with a small residual position remaining. The reconstruction matches sales against the oldest purchases within each account and separates realized results from the remaining holdings. It does not establish broker tax treatment or a portfolio return.

The transactions show repeated changes in exposure. The author identifies an ongoing fundamental view and the use of historical ranges, volume, and retracements to inform those changes. The records alone do not establish the precise signal behind each execution.

What would have changed my mind

This was a high conviction trade. News of large customers leaving Workday or widespread cancellations of contracts and the future revenue attached to them would have made me reconsider. Customer dependence was central to my thesis. Evidence that those relationships were breaking down would have challenged it directly.

The objection that deserves attention

This section is an editorial assessment developed during the reconstruction. It is not presented as a risk framework documented before the trade.

Workday could retain its customers and still disappoint its shareholders. Customers might spend less, reduce workforce counts, or buy fewer additional products. AI could also raise the cost of defending the franchise. Customer dependence helps explain why replacement could take time, but does not establish how much profit Workday would retain.

The author's experience building small applications informed the judgment. It is not a direct test of what a specialist enterprise software team could accomplish. The stronger argument concerns the organizational work and ongoing responsibility involved in replacing an embedded system.

The profitable trades do not resolve those questions. The business thesis needs to be assessed against subsequent operating evidence, while the trading decisions need to be assessed against their own prices, exposure, and outcomes.