If you’re new to programmatic, guessing an audience and throwing a big budget at it is the fastest way to lose trust.
A sprint gives you a different path.
You use a modest budget, spread across a few very different audiences, to answer one specific question: for this client and this offer, who actually responds?
Step 1: Pick one client and one outcome
Don’t overcomplicate it.
You’re not “going programmatic” for the whole agency at once.
You’re running one smart test for one client.
Good use cases:
Local lead gen: home services, healthcare, fitness, legal, real estate.
B2B: SaaS, professional services, agencies with a sales team.
Ecommerce: brands with a clear hero product or flagship category.
Define a single outcome in client language:
“More booked consultations from [service area].”
“More qualified demo requests from [job titles].”
“More purchases of [product/category] at or below [target CPA or ROAS].”
This outcome is the lens you’ll use when you look at your test results.
Step 2: Define 3–5 radically different audiences
This is where the learning comes from.
Because you’re using a DSP, you can target by role, industry, company size, behavior, and geo through 80+ SSPs.
That’s a lot of flexibility.
Use it to test big swings, not tiny nuances.
Examples you can copy:
Local services client (e.g., regional fitness chain):
Gym/fitness business owners.
Healthcare practice owners.
Professional services owners (law, accounting, real estate).
B2B client (e.g., SaaS selling demos):
Marketing Directors/VPs in your target industries.
Founders/CEOs at small companies.
Operations/RevOps roles in your ICP.
Multi‑location brand:
Corporate decision‑makers at HQ.
Franchise owners.
Local store or general managers.
Each of these becomes its own line item later.
If two audiences sound almost identical (e.g., “Marketing Directors” vs “Marketing Managers”), combine them for this first sprint.
You want to be able to look at a report and say: “These audiences were genuinely different bets.”
Aim for 3–5 audiences on your first run.
Step 3: Set a budget your client will actually approve
You want the sprint to feel safe and controlled.
For most verticals, a good starting structure looks like this:
3–5 audiences.
$100–$150 per audience.
Roughly $300–$750 in total media spend.
In higher‑CPM or niche markets (certain B2B, some CTV/video plays), you may need to push to $150–$200 per audience to get enough impressions to compare segments meaningfully.
How to position this budget:
“Instead of guessing and risking a $10k test on the wrong audience, we’ll invest around $500 across a few very different audiences. We’ll see which one is 2–3x more responsive, then direct the real budget there.”
That framing makes the test look like risk management, not extra spend.
Step 4: Draft one flexible message you can reuse everywhere
You are not running a creative shootout yet.
You are running an audience test.
That means you want one clear message that works for all your audiences:
It should explain what the client does in simple terms.
It should highlight a core problem and outcome.
It should lead to a basic call‑to‑action (learn more, book, sign up).
Example for an agency selling programmatic services:
Headline: “Get Enterprise‑Level Ad Reach Without an In‑House Team.”
Body: “We run your campaigns across premium sites and apps through an enterprise DSP. Start with a low‑risk test and scale only what works for your business.”
You’ll use this same core ad (with appropriate formats) for every audience in the sprint.
That way, when one audience wins, you know it’s because of who they are, not because they happened to see a different headline or image.
Step 5: Summarize the sprint as a one‑page plan
Before you build anything inside DSP Connect, you should be able to summarize the sprint in a single page or slide:
Client: who this sprint is for.
Outcome: what you’re trying to impact first.
Audiences: the 3–5 distinct segments you’re testing.
Budget: per‑audience allocation and total test spend.
Message: the one ad concept you’ll use everywhere.
That summary is easy to share internally and externally. You can send it to the client as:
“Here’s the exact test we’re going to run, the budget involved, and what we’ll learn at the end. From there, we’ll decide where to scale based on data, not guesswork.”
Now you have a clear plan and a number your client can say yes to.
The next step is turning that plan into a live campaign inside DSP Connect.