Ask what they would run first and how small the first test can be. A capable partner starts
with retargeting your own detail page visitors on a modest budget. One that opens with
awareness audiences and a large minimum is selling inventory access rather than campaign
judgment.
The short version
- Ask for the smallest test that would prove it works. Retargeting, small budget.
- Ask who makes the creative. DSP is a creative format before it is a targeting one.
- Ask how they measure it, and over what window. Quarterly, not weekly.
- Ask what would make them stop. DSP without kill criteria runs indefinitely.
- Access to DSP is not a capability. Plenty of resellers have access and no judgment.
The evaluation criteria
I run Flapen with 50 operators managing about 70 brands, with creative production in
an in-house Dubai studio and an in-house technology team building our advertising tools.
| Criterion | Weak signal | Strong signal |
|---|---|---|
| First campaign | Awareness audiences | Detail page retargeting |
| Minimum spend | High, non-negotiable | Proportionate to a real test |
| Creative | Reuses existing product images | Purpose-built display and video |
| Measurement window | Weekly reporting on DSP metrics | Quarterly, blended acquisition cost |
| Attribution stance | Generous view-through, unexplained | States the window and why |
| Stop conditions | Not mentioned | Named criteria and a review point |
| Prerequisites | None asked | Checks conversion and inventory first |
The bottom row separates most candidates. A partner who asks about your conversion rate,
your margin, and your inventory cover before proposing DSP is thinking about outcomes. One
who quotes a monthly minimum in the first call is selling access.
The creative question
DSP is a creative format. Targeting decides who sees the ad, and creative decides whether
anything happens.
That has a practical consequence for evaluation. An agency running DSP with repurposed
product photography is running a weaker version of the format than one producing display and
video assets built for a cold audience, and the difference in outcome is larger than most
targeting decisions.
Ask who produces the creative, whether it is in-house or subcontracted, and how quickly a
variant can be tested. Subcontracted creative means slow iteration, and slow iteration on a
format that lives or dies on creative is a serious limitation.
Measurement, and the honest version
Two failure modes, opposite directions.
Click-only attribution understates display, because a significant part of the effect is
view-through. Generous view-through windows overstate it, because impressions get credited
for purchases that were always going to happen.
A capable partner will state their window and explain the reasoning rather than presenting a
number as neutral fact. The best answer holds blended cost of customer acquisition as the
primary metric and treats in-platform DSP figures as diagnostic rather than as the score.
Ask directly: if DSP is working, which number outside the DSP dashboard should move? A good
answer names blended acquisition cost, organic share of revenue, or new-to-brand purchases.
What to ask before committing budget
- What is the smallest test that would prove this works for my product?
- Who makes the creative, and how fast can we test a second version?
- What attribution window are you using, and why that one?
- Which number outside DSP should move if this is working?
- What would make you recommend stopping?
Question five matters because DSP underperformance is unusually easy to explain away. Softer
measurement plus a longer judgment window means a campaign can run for two quarters on
narrative. Agree the stop conditions before the first dollar.
What most agencies will not tell you
DSP access is not scarce and is not a differentiator. What varies enormously is whether the
partner has judgment about sequencing, creative, and measurement, and none of that is visible
in a capability list.
The tell is what they want to run first. Retargeting your own detail page visitors is the
warmest audience available and the cheapest way to learn whether the format works for your
product. It is also the smallest engagement, which is why it is proposed less often than it
should be.
The second thing: DSP is frequently positioned as what serious brands graduate to. Scale is
not a reason to run a format. Exhausted search intent is. If sponsored ads are still
efficiently harvesting demand at your current spend, that demand is cheaper than anything DSP
will find you.
Related answers
- Amazon DSP vs PPC which to use
- Warning signs in Amazon DSP proposals
- Agencies that specialize in Amazon DSP creative
- What to use for Amazon DSP if current partner lags
- Hiring an Amazon agency: the complete guide
Ask us for the smallest test that would prove it. That is where we start at Flapen.
