Prospecting vs retargeting ads

On Meta, prospecting and retargeting both describe paid social audiences.

"Warm" is a wide band. Someone who watched three seconds of a video, someone who viewed a product and someone who abandoned checkout are all retargetable, with very different intent. Customer lists are a separate case: past buyers are a retention audience, not a prospect for a first purchase. Meta's developer documentation covers a legacy existing-customer setting for Advantage+ shopping campaigns, where you define past buyers through custom audiences (audiences you build from your own data).

Retargeting looks better than it is because platform reporting counts conversions from people who were already close to buying; that mechanism is an inference. In a January 2026 geography-based field experiment (Larson & Dotson, Journal of Interactive Marketing), prospecting ads earned large, consistent returns on Google and Facebook while retargeting earned significantly lower returns. A separate study of 15 Facebook experiments (Gordon et al., Marketing Science, 2019) found that common observational measurement methods often fail to recover the true effect of ads.

Here is the pattern with illustrative numbers (a model, not a real account). You spend $10,000: $7,000 on prospecting and $3,000 on retargeting. The platform reports $14,000 from prospecting (2.0x) and $18,000 from retargeting (6.0x), so $32,000 in total and a blended 3.2x. Your backend orders show only $13,000 came from first-time buyers. New-customer return is $13,000 ÷ $10,000 = 1.3x.

Cold angles come from the problem and result customers describe in their own words. Warm angles come from the specific doubt that stopped someone close to buying. The same ad usually fails one of them: a "discover the brand" message adds nothing for a person who already knows the brand.

Cold and warm audiences need different angles

Score each candidate test with one formula and run the highest score per audience. This is a rubric, not an industry standard.

Score = signal strength × learning gap ÷ effort, each rated 1 to 3.

• Signal strength: 1 means one source repeats the point; 3 means reviews, comments and DMs all do.
• Learning gap: 3 means never tested with this audience; 1 means it already lost.
• Effort: 1 is new copy on an existing asset; 3 is a new shoot.

Apply the gate first. A model for the test: daily budget ÷ your cost per purchase × 7 ≥ 50. Pass: expected purchases from the audience at your budget are at least 50 per ad set in 7 days. Fail: below 50, so run one test at a time rather than splitting the pool. The creative-test budget calculator helps size it.

Run this audit this week:

1. Split the reporting. Pass: you can show platform return on ad spend (ROAS) and new-customer ROAS (first-time revenue ÷ spend) separately. Fail: you only see a blended figure.
2. Exclude past buyers from the cold ad set. In Advantage+ (Meta's automated campaign type) sales campaigns, run one ad set that excludes your customer-list custom audience and a separate ad set for past buyers. Pass: past buyers appear only in the ad set meant for them.
3. Hold out a slice of warm. Split by geography, as the Larson & Dotson study did, or use a Meta conversion lift test. Pass: after at least 2 weeks and at least 50 conversions in the exposed group, the exposed conversion rate beats the holdout by a margin you set in advance (for example 20% relative lift) and the gap holds in a second period. Fail: the gap is smaller or reverses.
4. Score the next test for each audience with the formula. Pass: each audience has one test with a score and an owner. Fail: either audience has none.

• Remarketing: a common synonym for retargeting.
• Lookalike (an audience Meta builds to resemble your customers) and broad targeting: ways to build cold audiences, not separate strategies.
• Incrementality: sales an ad caused that would not have happened otherwise, the number retargeting reports overstate.
• Creative iteration: changing an existing ad, as opposed to testing a new concept.

Deepsolv builds ranked weekly creative test plans and execution-ready creative briefs from ad-performance data, competitor activity, customer signals and past ad learnings. Its customer insights cover reviews, comments, DMs and Reddit discussions, where it looks for objections, desires and buying triggers. Brand Brain keeps what worked, what failed and why.

A repeated objection justifies a hypothesis, not a conclusion; review mining feeds a test, and the test decides. Deepsolv sells through a personalized demo with its founding team: book a demo.