Amazon DSP and AMC incrementality sounds like a measurement topic. It is not. It is a budget permission topic.
That distinction matters when a brand spends €5K, €25K or €100K a month across Amazon Sponsored Ads, DSP and retail media. DSP can reach shoppers before they search, AMC can connect the path between impressions, Sponsored Products, Sponsored Brands and purchases, and the dashboard can produce very confident looking ROAS. Lovely. But the operator question is sharper: did the media create profitable demand that the SKU could actually carry?
The named mistake is what I call “view-through victory”. A DSP campaign shows strong attributed sales because shoppers saw an impression and bought within the lookback window. Everyone celebrates the ROAS. Then AMC shows that many of those shoppers were already brand buyers, already exposed to Sponsored Products, or already searching the category. Worse, the promoted SKU had only 18% contribution margin before ads. The campaign did not scale the business. It rented credit from demand that was already walking toward checkout.
This guide is written for marketplace teams and agencies managing Amazon advertising as a commercial operating system, not a media silo. The stance is simple: use AMC to decide where DSP earns the right to scale, then connect that decision to SKU margin, TACoS, stock cover and follow-on Sponsored Ads pressure. Incrementality without margin is an interesting statistic. Incrementality with margin is a budget decision.
What competitors explain well about DSP and AMC
The current market content is useful. Amazon explains DSP clearly as a demand-side platform that buys display, video and audio inventory across Amazon properties such as Twitch, IMDb, Fire TV and third-party publishers, powered by Amazon shopping and streaming signals. It also highlights clean rooms and Amazon Marketing Cloud as part of the planning and measurement stack.
Podean’s DSP content makes a strong strategic point: DSP should not be managed as an isolated programmatic channel. It needs retail alignment, inventory data, pricing, merchandising and Sponsored Ads context. BidX explains AMC custom audiences well, especially the ability to build SQL-defined segments from Sponsored Ads, DSP, Amazon Shopping Insights and first-party data, then activate those audiences in DSP. Their AMC for Sponsored Products article also points to the shift where AMC-style insight is no longer only a DSP luxury.
Other industry guides focus on the attribution and incrementality gap. Venture Forge puts it neatly: ROAS shows what happened after spend, not whether the spend caused the sale. EVA’s incrementality guide uses a useful example of an 8.3x attributed ROAS campaign that becomes a 2.5x incremental ROAS once a holdout is measured. Nectar’s positioning is also sharp: most brands use AMC for reporting, while the better use is decision-making, especially where Sponsored Ads and DSP overlap.
The gap FiveX can own is the profit layer after the incrementality layer. Many articles stop at “prove incrementality”. Operators need the next question: after incremental revenue is proven, is it still worth scaling once SKU margin, fees, returns, coupons, stock and paid-search cannibalisation are included?
The operator rule: DSP budget needs three proofs
Before DSP budget scales, require three proofs. If one is missing, keep the campaign in test mode.
| Proof | Question | Why it matters |
|---|---|---|
| Incrementality proof | Did exposed shoppers create more sales than a comparable control group? | Prevents paying for orders that would have happened anyway. |
| Margin proof | Did incremental orders remain profitable after fees, fulfilment, returns, coupons and ads? | Prevents scaling revenue that weakens contribution margin. |
| Operating proof | Can stock, Buy Box, pricing and Sponsored Ads budgets support the extra demand? | Prevents DSP from creating demand the business cannot fulfil cleanly. |
This is where FiveX is useful. The Amazon Advertising integration pulls campaign performance into the same operating view as SKU sales. Profit analytics adds product cost, fees, fulfilment and return assumptions. The Advertentie Service team uses those signals to decide whether DSP should scale, cap, retarget, prospect or stop.
Start with campaign role, not ROAS
DSP campaigns do different jobs. Retargeting a product-detail-page viewer is not the same as reaching a new in-market audience on streaming TV. If you compare both only on attributed ROAS, retargeting will usually look like the adult in the room. That does not mean it is creating the most growth.
Label every DSP and Sponsored Ads campaign by role:
- Prospecting: reach shoppers who do not know the product or brand yet.
- Consideration: bring category-aware shoppers back to the product or store.
- Retargeting: convert recent product viewers, cart abandoners or video viewers.
- Brand defense: protect high-intent demand from competitors.
- Launch support: buy learning, reviews and early visibility for a new SKU.
Each role needs a different incrementality expectation. Retargeting may have high attributed ROAS but low incremental ROAS. Prospecting may have weaker short-term ROAS but create more new-to-brand demand and downstream Sponsored Products efficiency. The trick is not choosing one forever. The trick is knowing which role deserves the next euro this week.
Named example 1: Nora Home and the retargeting trap
Nora Home sells premium bedding on Amazon.nl. The team runs €18,000 a month in Amazon Ads, including €4,000 in DSP retargeting to shoppers who viewed the duvet cover PDP in the last 14 days. The DSP report shows €32,000 in attributed sales. On paper, that is 8.0 ROAS. Very tidy. Possibly too tidy.
An AMC audience holdout tells a different story. The exposed group generated €32,000 in sales. The matched control group, suppressed from DSP, still generated €24,500. Incremental revenue was €7,500, not €32,000. Incremental ROAS was 1.9.
Now add margin. The duvet cover has €21 contribution margin before ads on a €70 selling price, so the break-even ACOS is 30%. The DSP spend of €4,000 against €7,500 incremental revenue means the incremental ACOS is 53%. That campaign did create some sales. It did not create profitable sales.
The action is not “turn off DSP forever”. The action is to cap the 14-day retargeting pool, exclude past purchasers and recent Sponsored Products clickers, and move €2,500 into a broader in-market home textiles audience with a four-week test. In FiveX, that becomes an action label: cap retargeting, test prospecting, review margin after 14 days.
Named example 2: VoltGear and the search halo that actually pays
VoltGear sells USB-C chargers on Amazon.de and MediaMarkt. The team wants to know whether a €12,000 DSP video and display campaign is helping Sponsored Products or simply taking credit after Sponsored Products did the work.
AMC path analysis shows a useful sequence. Shoppers exposed to DSP and then clicking a non-brand Sponsored Products ad convert at 5.4%. Shoppers who only click the same Sponsored Products terms convert at 3.8%. The combined DSP plus Sponsored Products path creates €46,000 in incremental revenue versus a modelled baseline of €33,000. Incremental revenue is €13,000.
The margin check matters. The charger sells for €29.99 with €9.20 contribution margin before ads. After €12,000 DSP spend and €3,100 incremental Sponsored Products spend, the test is barely positive: €13,000 incremental revenue at roughly 30.7% pre-ad margin creates about €3,991 contribution before media, which is not enough to cover €15,100 in total media. If you stop there, the campaign fails.
But the SKU also gains organic rank on two non-brand terms and reduces CPC pressure the following week. FiveX would not call this a scale decision yet. It would call it a controlled learning decision: keep the audience, reduce frequency from 7 to 4, lower the DSP budget to €6,000, and only scale if TACoS falls below the SKU’s 18% guardrail after the halo period. Operator voice: the halo is allowed to be promising; it is not allowed to be vague.
Named example 3: PureSip and the stock-cover problem
PureSip sells reusable water bottles on Amazon.fr. A prospecting DSP campaign reaches outdoor and fitness audiences and looks genuinely incremental. The test group produces €41,000 revenue, the control group €29,000, so incremental revenue is €12,000 on €3,600 spend. Incremental ROAS is 3.3. The SKU has 38% contribution margin before ads, so the profit math looks workable.
Then the operating proof fails. FiveX shows only 11 days of FBA stock cover at the new run rate, and the next inbound shipment is 19 days away. If the team scales DSP immediately, the product risks stocking out, losing organic rank and wasting the learning from the test.
The right decision is counterintuitive: do not scale the winning campaign yet. Hold DSP spend at €3,600, shift Sponsored Products to exact-match harvest on the strongest non-brand terms, and increase DSP only after stock cover is above 28 days. Incrementality can be real and still not deserve more budget today. Inventory gets a vote. It is annoyingly powerful like that.
How to measure DSP incrementality with AMC without overcomplicating it
You do not need a science project every week. You need a repeatable measurement rhythm.
- Define the decision first. Are you deciding whether to scale prospecting, cap retargeting, defend branded search or support a launch?
- Create comparable audiences. Use AMC and DSP controls where available. Keep audiences similar by recency, category behaviour, geography, purchase history and exposure rules.
- Measure incremental revenue. Compare exposed versus control revenue, but separate new-to-brand, repeat buyers and branded search paths.
- Calculate incremental contribution margin. Apply SKU margin, referral fees, fulfilment, return reserve, coupon cost, DSP spend and extra Sponsored Ads spend.
- Check operating constraints. Add stock cover, Buy Box stability, price position and content readiness.
- Assign an action label. Scale, cap, isolate, retest, exclude, harvest, fix or stop.
The last step is where many teams are too soft. “Analyse further” is not an action label. It is a meeting invitation wearing a tiny hat.
The budget reallocation playbook
Once you have incrementality and margin together, budget decisions become much calmer.
| Finding | Likely problem | Action |
|---|---|---|
| High ROAS, low incremental ROAS | Retargeting or brand demand is over-credited | Cap frequency, exclude recent buyers, move budget to prospecting tests. |
| Good incrementality, weak margin | SKU cannot afford the media cost | Improve price, reduce coupons, fix fulfilment cost or cap spend. |
| Good incrementality, stock risk | Demand will outpace availability | Hold budget until stock cover clears the threshold. |
| Moderate ROAS, strong new-to-brand margin | Upper-funnel campaign is under-credited by standard reporting | Protect the test and judge over a longer window. |
This is also why a percentage-of-spend agency model can be dangerous without margin guardrails. More DSP spend is not automatically better work. Better work is moving budget toward the campaigns that create profitable incremental demand.
What to ask your agency before scaling DSP
If an agency manages Amazon DSP and AMC for you, ask these questions before approving more budget:
- Which campaigns have high attributed ROAS but weak incremental ROAS?
- Which audiences overlap with branded search, Sponsored Products or past purchasers?
- What is incremental contribution margin after DSP and follow-on Sponsored Ads spend?
- Which SKUs are excluded from DSP because margin, stock or Buy Box is not ready?
- Which budget moved because AMC proved a different decision?
- What action labels will we review next week?
A strong answer connects AMC, DSP, Sponsored Ads, SKU P&L and operations in one story. A weak answer says “the ROAS is strong” and hopes nobody asks what caused the sale.
Where FiveX fits
FiveX helps marketplace teams make this practical. The platform connects Amazon Advertising, marketplace sales, SKU profitability, inventory and operational data so DSP and AMC insights do not live in a separate analytics cave. Ad managers can see which ASINs can afford more demand, which campaigns are borrowing from organic demand, which products are too close to stockout and which actions should go into the weekly queue.
For brands spending from €5K a month on Amazon, bol or MediaMarkt ads, our Advertentie Service uses the same logic in managed accounts: margin-aware campaign structure, TACoS guardrails, SKU eligibility, incrementality checks and plain-English action labels. Not because dashboards are bad. Dashboards are lovely. They just need to earn their keep.
The takeaway
Amazon DSP and AMC are powerful because they reveal what standard ad reporting hides. But the best teams do not stop at “the campaign was incremental”. They ask whether the incremental demand was profitable, whether the SKU can carry it, and which budget should move next.
That is the operator standard: DSP does not scale because it can claim sales. DSP scales when AMC proves incremental demand, the P&L proves contribution margin and operations prove the business can fulfil the demand without creating a new problem. Slightly less glamorous than a ROAS screenshot. Much better for profit.