Adobe reports its August quarter after the close today, and two widely used estimate feeds do not agree on what it is being measured against: one carries $4.86 for the quarter, the other $6.07. They also do not agree about the past. On the same two quarters Adobe has already reported, one feed records two misses and the other records two beats — and neither has made an arithmetic error.
The obvious question tonight is whether Adobe beat. That question leads nowhere on its own, because "beat" is a subtraction, and the subtraction has no meaning until both of its terms sit on the same basis. The question worth asking first is narrower and answerable before the print: how many earnings-per-share series exist for this quarter, and does the page you are reading tell you which one it is using? For Adobe the answer is three, and most pages do not.
Adobe publishes two of them itself
On 11 June Adobe guided this quarter twice in the same table: GAAP earnings per share of $4.40 to $4.45, and non-GAAP of $6.05 to $6.10, both on revenue of $6.67bn to $6.72bn. Neither is a Street figure. Both are Adobe's, for the same three months.
The distance between them is $1.65 a share at the midpoints, and Adobe itemises it: $1.47 of stock-based and deferred compensation, $0.15 of intangibles amortisation, $0.04 of acquisition-related costs, less a penny of tax adjustment. Stock compensation is nine-tenths of the wedge, which is why it travels with the Rule of 40; the general mechanism is the one we set out in August. Our preview earlier today flagged that a third number was circulating and should not be paired with tonight's print. This piece is what that number turned out to be.
The third series matches neither, in any quarter
Read across the table above. In four consecutive quarters the estimate feeds' figure is never Adobe's GAAP result and never Adobe's non-GAAP result. It sits between them, and not at a fixed distance: $0.11, $0.10, $0.23 and $0.33 above GAAP as the quarters run forward.
Barchart footnotes it precisely and without apology — its earnings numbers "reflect diluted earnings per share, reported before non-recurring items". That is a real convention, and it can be rebuilt out of Adobe's own reconciliation table, which is worth doing before concluding the feeds are careless.
Start from Adobe's non-GAAP figure, take the stock-compensation add-back out again, and take out Adobe's income-tax normalisation as well. What is left reproduces the feeds' number in every one of the four quarters: $4.32 against their $4.29, $4.57 against $4.55, $4.84 against $4.83, and $4.58 against $4.58. The largest residual is three cents, which is about what rounding each per-share line to the penny will produce on its own.
So the series is coherent, and it is not a mistake. It is earnings before non-recurring items and before the company's own tax normalisation, and it leaves out the largest adjustment in software: Adobe's stock compensation has run between $1.20 and $1.47 a share a quarter across these five periods. That is a defensible thing to publish. It is simply not the number Adobe guides to, and nothing on the page where it appears says so.
Which is how the same quarter becomes a beat and a miss
Each feed's own surprise percentage, as each publishes it
| Quarter ended | Nasdaq / Barchart | Investing.com |
|---|---|---|
| 29 Aug 2025 | +1.90% beat | +2.51% beat |
| 28 Nov 2025 | +3.64% beat | +1.85% beat |
| 27 Feb 2026 | −0.41% miss | +3.41% beat |
| 29 May 2026 | −3.38% miss | +2.41% beat |
Left column as published on Barchart's Adobe earnings-estimates page on 9 September 2026, whose figures match the Nasdaq feed for all four quarters; each percentage is that feed's reported figure against its own estimate. Right column as published on Investing.com's Adobe earnings page on 10 September 2026, whose four reported figures are Adobe's published non-GAAP earnings per share exactly. Neither column is wrong on its own terms: each compares a result to an estimate built on the same basis. They disagree because the two bases are different, not because either feed made an arithmetic error.
February and May are the interesting rows. On the Nasdaq and Barchart basis Adobe missed both times, by 0.41% and 3.38%. On Investing.com's basis — whose four reported figures are Adobe's published non-GAAP numbers exactly — Adobe beat both times, by 3.41% and 2.41%.
Both columns are internally honest. Each compares a result to an estimate gathered on its own basis, and each feed's surprise percentage is correct against its own inputs. The damage is done only by crossing them, and tonight that is unusually easy to do: Adobe's guided non-GAAP midpoint of $6.075 against the feeds' $4.86 is a 25.0% beat that will not have happened, and Adobe's guided GAAP midpoint of $4.425 against the same $4.86 is a 9.0% miss that will not have happened either.
It is not one vendor, and the labelled feeds get it right
This is a convention, not a mistake by a single site. The figures Barchart shows are the ones the Nasdaq feed carries, to the cent, for all four quarters and for the estimates too. MarketBeat prints $4.84 from three estimates for tonight — and prints it on the same page as Adobe's own guidance of 6.050 to 6.100, with nothing between them to say the two are measured differently. Investing.com, on the other basis, carries $6.07 for the same quarter, within half a cent of Adobe's own non-GAAP midpoint.
The tell is labelling. StockAnalysis states plainly that its earnings and forward multiples are based on non-GAAP adjusted numbers, and its FY2026 consensus of $24.41 across 33 analysts sits inside Adobe's own full-year non-GAAP guide of $24.35 to $24.45. Barchart's FY2026 figure is $19.83 — $1.83 above the top of Adobe's GAAP guide and $4.52 below the bottom of its non-GAAP guide, at home in neither. Two vendors, the same fiscal year, $4.58 a share apart.
And it does not stop at earnings day
Trailing four quarters to 29 May 2026, against the 9 September 2026 close of $254.86
| Basis | Trailing EPS | P/E at $254.86 |
|---|---|---|
| Adobe non-GAAP | $22.83 | 11.2 |
| Estimate feeds | $18.25 | 14.0 |
| Adobe GAAP | $17.48 | 14.6 |
Each trailing figure is the sum of the four quarters in the table above on that basis, and each ratio is that close divided by that sum; the division is ours. The estimate-feed row reproduces the trailing earnings figure Barchart prints in its own key statistics, $18.25, and its printed ratio of 14.10 is that figure against the previous close of $257.26 rather than the 9 September close used here. The price is this site's capture of the 9 September close and is not a live quote.
The basis propagates into every ratio built on top of it. On Adobe's trailing four quarters to 29 May, the same closing price supports a price-to-earnings ratio of 11.2, 14.0 or 14.6 depending only on which earnings series you pick up. The highest is 30.6% above the lowest, and none of the three is wrong.
What would settle it
Adobe prints both of its own figures in one sentence, so the wedge is readable within minutes of the release tonight. Two things would change the conclusion here. The feeds' series and Adobe's non-GAAP converge only when the stock-compensation add-back and the tax normalisation both go to zero, and Adobe has already guided $1.47 of stock compensation for this quarter, so they will not. And if the feeds labelled the basis on the page where the number appears, as StockAnalysis does, the number would stop being a trap without a single figure changing.
Until then the safe comparison is the one Adobe supplies: revenue of $6.67bn to $6.72bn, GAAP of $4.40 to $4.45, non-GAAP of $6.05 to $6.10. A print inside those ranges is an in-line quarter, on whichever line you care about, whatever the surprise column says.
Every Adobe figure here — the quarterly GAAP and non-GAAP earnings per share, the guided ranges, the per-share reconciliation lines and the full-year targets — is from Adobe's own quarterly releases of 11 September 2025, 10 December 2025 and 11 June 2026, each filed as Exhibit 99.1 and each printing both bases together; the February-quarter figures are the comparative column of that June release rather than a separate document. The estimate figures and surprise percentages are read first-hand from the vendors' own pages on 9 and 10 September 2026 — Barchart, MarketBeat, Investing.com and StockAnalysis — and from the Nasdaq estimate feed; each is that vendor's own number on its own stated basis, and this site stores no consensus series of its own. The reconstruction test, the trailing sums, the three price-to-earnings ratios and every percentage gap are ours. The $254.86 price is this site's capture of the 9 September close, not a live quote.
