Actionable insights — This Week in Visuals: WMT ADI NTES TGT AS KLAR

1. Strip the one-off windfall before judging a beat — then demand the guidance on the clean basis too

1. Whenever a beat is unusually large, find the line that made it unusual: a tariff or duty refund, an insurance recovery, a legal settlement, a warranty release, a gain on investments, a tax item. Get the dollar figure and the basis-point contribution to margin.
2. Subtract it and re-run the comparison against consensus. A beat that survives the subtraction is a business event; one that doesn't is an accounting event.
3. Do the same to the margin, not just to EPS — a windfall lands in operating income and can flatter a margin trend for several quarters if it recurs in tranches.
4. Then apply the identical test to the guidance: ask whether the raise was stated including or excluding the windfall, and insist on the ex-windfall midpoint. Management that volunteers both numbers is telling you the underlying business moved; management that only quotes the inclusive figure is hiding behind the windfall.
5. Finally, ask what the company is doing with the windfall — banking it flatters this year's margin, spending it on price or capacity trades reported profit for share.

• The quarter the refunds stop — the Y/Y margin comparison inverts even if the business is unchanged; management quoting only the inclusive guide; and any windfall recurring in instalments, which quietly becomes an assumed run-rate.

2. Decompose a disappointing headline into the part management was dealt and the part it chose

1. Take the disappointing metric and identify every component with an external, non-operating cause — a regulatory price change, a mandated reimbursement cut, an FX move, a lapped one-time event.
2. Quantify each in basis points and restate the metric excluding it. If management supplies the ex-item figure, check that it reconciles rather than accepting it.
3. Separately, identify any component that reflects a deliberate decision — price investment, deliberate wholesale restraint, a subsidy pull-back. A metric management chose to depress is information about strategy, not about demand.
4. What remains after both adjustments is the actual demand signal. Cross-check it against a volume measure that pricing can't distort (transactions, units, visits, subscribers).
5. Only then decide whether the market's reaction was to the business or to the arithmetic.

• The drag annualizing out of the comparison (the ex-item and reported figures should converge); transactions turning negative, which no adjustment can rescue; and a "deliberate" price investment that quietly becomes permanent because competitors matched it.

3. Rank a retailer's segments by margin structure, not by size — the small fast lines are the economics

1. List every revenue line the company discloses beyond the core, with its growth rate: e-commerce, advertising, membership/subscription fees, marketplace/third-party, financial services.
2. Classify each by whether serving an extra dollar of it consumes inventory, labour and shelf space (retail-like) or not (platform-like). Advertising, memberships and marketplace fees are structurally the highest-margin dollars a retailer can earn.
3. Track whether the platform lines are compounding several times faster than the core. That ratio, not the comp, determines where the company's profit comes from in three years.
4. Check the enabling asset: the physical estate doing a second job (stores as fulfilment nodes), the audience being resold (shopper data), the membership base being monetized.
5. Re-underwrite the multiple on the mix you project, not the mix reported — a retailer with a fast-growing ad business converges on a platform multiple, and the market usually notices late.

• Advertising growth decelerating toward the core's rate (the mix thesis stalls); e-commerce growth bought with rising fulfilment cost per order; and any move to stop disclosing the platform lines separately.

4. Split a comp into traffic and ticket — traffic is the honest half

1. Decompose comparable sales into transaction growth (how many visits/orders) and average ticket growth (how much per visit).
2. Weight traffic far more heavily. Ticket can be inflated by price increases, mix shifts into expensive categories, or discount-driven basket building; traffic cannot be manufactured.
3. Adjust for the comparison base — a traffic number achieved while lapping a one-time event (a console launch, a stimulus quarter, a viral product) is worth more than the same number on an easy base.
4. Then check breadth: how many merchandise categories grew, and specifically whether the historically differentiating ones participated.
5. Treat a traffic-led comp with two flat flagship categories as a turnaround in progress, not a completed one — the missing categories are the remaining thesis.

• The first quarter where the comp holds but traffic turns negative (the mix has become price); apparel/home inflecting positive, which is what converts "credible" into "broad-based"; and category breadth narrowing back toward consumables.

5. Run the inventory-versus-revenue check as a sell-through lie detector

1. Put inventory growth beside revenue growth for the same period. Inventory growing materially slower than revenue means product is leaving shelves faster than it is being built — genuine end demand.
2. Inventory growing faster than revenue is the warning: either demand is softening or the company is shipping into the channel to book revenue, which sets up future discounting.
3. Pair the ratio with the gross-margin direction. Rising margin plus lean inventory means full-price selling; rising revenue plus swelling inventory means a markdown quarter is being deferred.
4. For wholesale-exposed brands, ask separately whether management is deliberately restraining shipments to protect retailer inventory health — that shows up as a revenue sacrifice today and a margin defence later.
5. Re-run every quarter; the ratio is most informative as a trend, and it inverts early.

• The gap closing or inverting; gross margin slipping while inventory builds; and any quarter where wholesale grows much faster than direct-to-consumer, which usually precedes the build.

6. Separate a below-the-line miss from an operating miss before reacting

1. When EPS misses but revenue and gross margin beat, walk down the income statement to find where the loss appeared.
2. Classify it: operating (demand, pricing, cost) versus non-operating (investment marks, FX, one-time charges, tax). Non-operating items say nothing about the business and often reverse.
3. Check the reverse case with equal suspicion — investment gains flattering EPS are the same distortion pointing the other way, and are far more commonly waved through.
4. Re-underwrite on operating profit and gross margin, then verify with a demand metric the accounting can't touch (users, units, bookings).
5. Note whether the company holds a large marked portfolio at all; if so, treat headline EPS as structurally noisy and build the model on operating lines permanently.

• Investment losses recurring quarter after quarter (a portfolio problem is still a shareholder problem); operating margin diverging from gross margin; and any quarter where the miss migrates from below the line to above it.

7. In a lender or payments platform, follow margin dollars and cost of credit — not volume — and discount any improvement that arrives with a redefinition

1. Separate the three layers: volume (GMV/TPV), revenue (the take on that volume), and transaction margin dollars (revenue minus funding and credit costs). Rank their growth rates against each other.
2. Margin dollars growing faster than revenue, and revenue faster than volume, is the healthy sequence: the business is earning more per unit, which can offset a volume slowdown. The reverse sequence is volume bought at deteriorating economics.
3. Read cost of credit as a percentage of volume, plus the direction of recent delinquency cohorts. In lending this is the metric that decides whether the growth was worth having.
4. Check where the slowdown is geographically and whether it is discretionary-spend driven — a soft core market caps the volume recovery regardless of unit economics.
5. Apply the redefinition audit: if a metric improves in the same quarter its definition or accounting treatment changes, treat the improvement as unproven until a like-for-like figure is published.
6. Finally, price the guidance shape — front-loaded spending that makes one quarter look terrible is not the same as deterioration, provided the promised strong quarter is specified.

• A like-for-like transaction margin excluding the Fair Financing treatment; provisions turning back up as the slower-growth cohorts season; Germany stabilizing or not; and the CFO and CMO departures in early 2027, which remove the people who set and explained these metrics.

8. Own the boom second-hand: track a segment's mix share inside an incumbent, not the boom's headline

1. For any capital cycle, list the physical constraints it creates beyond the headline product — in an AI buildout: power delivery, thermal management, optical interconnect, memory, switching.
2. Find the established supplier whose reported mix is being re-cut by that constraint, and track the disclosed share (e.g. "data center is X% of segment revenue") quarter over quarter. A rising share is the cleanest evidence the cycle is reaching that layer.
3. Require the sub-lines to be growing faster than the segment, ideally at triple digits, before treating the incumbent as a cycle participant rather than a beneficiary.
4. Check margin behaviour: earnings growing at roughly twice revenue growth means the incremental cycle revenue is arriving at very high contribution margin, not at a discount to win share.
5. Look at M&A for confirmation of scope — an acquisition that extends the franchise into an adjacent stage of the same constraint tells you management is underwriting the cycle as structural.

• The data-center share of Communications flattening; optical or power growth dropping below triple digits; memory/substrate cost pressure showing up in gross margin; and whether the promised FY27 momentum is guided or merely expected.

9. Judge a delayed product launch by whether the installed base needs it

1. When a company delays a flagship launch, first measure the base without it: is the existing recurring/live-service revenue growing, and is its margin expanding?
2. A growing, margin-expanding base converts a delay into optionality — the launch becomes upside. A shrinking base converts the same delay into a solvency-of-the-story problem.
3. Read the stated reason for the delay and check it is a quality decision with a specific fix, not a vague slip.
4. Verify the base's durability with engagement metrics that predict future spending (registered users, MAUs, chart position after content updates) rather than revenue alone.
5. Then decide how much of the pipeline you are willing to capitalize: with a healthy base, zero — treat every unreleased title as free upside.

• Live-service revenue decelerating while the pipeline stays unshipped (patience becomes drift); a second delay without a stated fix; and gross margin reversing, which would signal the base is being sustained by marketing spend.