Amazon DSP is the most oversold product in this industry.
Agencies love it because it is expensive and it sounds advanced. Most brands do not need it yet.
Here is when it is genuinely worth it, and when it is a tax.
When it earns its keep
DSP earns its keep when you have already saturated Sponsored Products on your core keywords, when you have enough traffic for retargeting audiences to be large enough to matter, and when you have a real reason to reach people off the search page. Usually that is a considered purchase, a strong repeat cycle, or a brand-building goal with budget behind it.
It is a waste when your Sponsored Products account still has obvious inefficiency in it, when your listing does not convert, or when your monthly spend is small enough that the minimums and management fees eat the upside before it appears.
I have told brands doing $150K/month to stay away from DSP and fix their listing conversion instead. That advice made us less money and made them more.
One job it's great at
Where we do see it work: a brand with a 60-day repeat cycle used DSP to retarget past purchasers in the window before their reorder. Not prospecting. Not awareness. Just reaching people who already bought at the moment they were about to run out.
That campaign ran at a 6.1 ROAS while their prospecting DSP sat at 1.4.
Same tool. One job it is great at, one job it is mediocre at.
The rule I would give any founder: never buy a more advanced ad product to solve a problem a cheaper one is still failing at.
Is your Sponsored Products account clean enough to justify anything fancier?