Targeting in marketing: what it is, its place in the STP model, and how the algorithm changed the craft

Targeting is the decision —not the analysis— of which market segments a company will actively go after with its products, communication, and investment. It always comes after segmentation: first you understand the map of the market, then you choose which parts to plant your flag in. That sequence is the foundation of one of the most widely taught frameworks in the discipline, formalized by Philip Kotler in Marketing Management (1967) and known as STP (Segmentation, Targeting, Positioning).

The operational importance of keeping these three layers separate is that each belongs to a different team and a different moment. A research team segments. A leadership or strategy committee makes targeting decisions. A brand and creative team builds positioning. When they get mixed up —when "segmenting" is confused with "choosing a target"— it usually ends in presentations of eight personas with exquisite biographical detail, none of which has real investment assigned to it.

If you've gotten this far looking for an answer about targeting, the most useful thing this article can do is help you make better decisions about who to serve, not give you more theory about the concept.

The operational difference between segmenting and targeting

It's worth pinning down precisely:

• Segmenting is describing the market. It's analytical: identifying groups that are significantly distinct by need, behavior, context, or attributes. The output is a map. It doesn't commit resources.
• Targeting is committing resources to one or several segments on the map. It implies prioritization: no longer actively serving others. It's a political and economic decision, not an analytical one.

The mistake you see at midsize companies is presenting targeting as the natural continuation of segmentation. It isn't. A well-done segmentation can show 12 legitimate segments; targeting requires choosing 2 or 3. That choice is rarely obvious from the data alone — it also depends on operational capacity, ambition, available resources, and possible differentiation. More depth in market segmentation.

The strategic origin: Ansoff and Kotler

The concept of targeting as a deliberate decision has older roots than the term itself. Igor Ansoff published in 1957, in Harvard Business Review, his famous matrix that crossed products (existing vs. new) with markets (existing vs. new) and named four growth strategies: penetration, product development, market development, and diversification. What Ansoff did was force executives to make explicit which market they were going to pursue, instead of treating growth as a single vector.

A decade later, Kotler integrated that thinking into his more operational marketing model and added the missing piece: once you've decided which market, you have to choose which segment of that market, and then how to position yourself. STP is that sequence of decisions, and targeting is the intermediate step where strategy goes from map to investment.

This isn't academic trivia. It's the difference between a company that "targets marketing professionals" (not a decision) and a company that "primarily serves marketing directors at European B2B companies with 50-500 employees, deliberately ignoring freelancers and enterprise" (a real decision, with budget to follow).

How to evaluate whether a segment deserves to be a target

The classic marketing literature teaches several mnemonics for evaluating segments as targets. The most used in English-speaking practice is DAMP or variants — Distinguishable, Accessible, Measurable, Profitable. The fundamental idea is that a segment can be "real" as an observation and still not deserve specific investment.

A segment deserves to be a target if it meets five conditions, which you can check honestly:

It's big enough. Not in absolute terms but relative to your business. A segment of 200 potential accounts can be huge for a five-person agency and absurd for a multinational.

It's accessible. There's some viable way to reach this segment — a channel, a community, an event, a partner. If the segment is theoretically attractive but you don't know how to get into it, it stays a nice hypothesis with no real operations behind it.

It's identifiable. You can tell when someone belongs to that segment. Not "people with sustainability values" but something concrete enough to make decisions about ad targeting, copy, or product.

It generates defensible economic value. The segment has the purchasing power, purchase frequency, or lifetime value to justify the cost of serving it. Back to the uncomfortable question any executive asks: "if we go after this segment, when do we recover the investment?".

It's coherent with your positioning. If the segment values things your brand can't credibly offer, serving it would force you to shift your positioning — and that's a decision of a different order.

If all five are yes, you have a target. If any is clearly no, that segment stays on the map but not on the priority list.

What changed in the last five years: the algorithmic era

Here is the most important shift that almost no current manual updates honestly. Until roughly 2021, targeting was primarily a manual selection practice: the media buyer chose the audience ("women aged 25-45 in Madrid, interested in X"), selected placements (sites, programs, positions), adjusted bids, and monitored.

Since the introduction of Performance Max (Google, 2021) and Advantage+ Shopping Campaigns (Meta, 2022), the dominant model on large platforms has flipped. The advertiser supplies:

• Creatives in several formats.
• A clear optimization objective (purchase, lead, subscription).
• A base audience (ideally first-party: current customer, customer lookalike).
• A budget.

And the platform decides who to show what, with which creative, at what moment. Fine-grained targeting, historically the media buyer's core skill, has become the algorithm's job. Not always with the buyer's consent.

This doesn't mean targeting stops existing. It means the skill has mutated. What now distinguishes the advertisers who perform well from those who don't is:

• The quality of the input data. Clean first-party audiences, well-measured conversions, correctly tagged events.
• The diversity and quality of the creative. More on-brand variants, more tests, more algorithmic learning.
• Independent measurement. Marketing Mix Modeling and incrementality tests to validate that what the platform reports as performance is real.

This shift has its supporters and its detractors. Supporters point out that the algorithm finds audiences a human wouldn't have detected and optimizes faster. Detractors note that ceding targeting means losing your own learning about your market and handing the decision to a system whose incentives aren't perfectly compatible with the advertiser's.

In practice, the serious companies I've seen have adopted a mixed model: algorithmic for volume acquisition, manual for brand activations or defensible niches. More depth on how the craft has been reorganized in direct advertising.

The historical mistake: targeting "everyone"

The most expensive phrase in marketing —and the most repeated— is some variant of "our product is for anyone who values X." Operationally, that's not having decided. When a brand says it serves the whole market, it's accepting that every segment within that market will get the same attention, the same message, and the same product fit. That means no segment will get what it would need to prefer this brand over more focused alternatives.

The classic case is Tesla in its early years (2008-2012). Instead of presenting itself as an "electric car for everyone," it deliberately targeted the wealthy innovator willing to pay a premium for new technology — the Roadster at 100,000 dollars. Only once the technology, the manufacturing learning, and the brand were consolidated did it move down to broader segments. Serving the whole market from day one would have required capital and a structure they didn't have in 2008, and would probably have failed.

The operational opposite: Procter & Gamble never has "one detergent brand" that serves everyone. It has a portfolio of distinct brands —Tide, Ariel, Gain, Cheer—, each positioned for a different segment. The company serves all segments of the market, but not with a single product.

What both cases teach: useful targeting isn't necessarily choosing a small segment; it's correctly allocating resources to each segment served and accepting that no brand can be everything to everyone at the same time.

Typical mistakes in 2026

Some operational patterns worth recognizing:

Blindly trusting the base audience the platform suggests. Performance Max and Advantage+ work better when they receive clean first-party data. If the audience you load into the platform includes unconverted contacts, low-quality leads, or poorly measured events, the algorithm learns from the noise and optimizes toward the wrong audience.

Targeting without a clear value proposition for the chosen segment. The effectiveness of targeting depends on having something to say to the segment — a message, an offer, a benefit. Without that, reaching the right segment only shortens the path to indifference.

Confusing frequency caps with targeting. Limiting how many times someone sees your ad isn't deciding who to serve; it's adjusting saturation. They're different problems solved with different mechanisms.

Recycling the same segmentation for years without review. A market can shift its segments significantly in 18-24 months. Targeting based on an obsolete segmentation is an arrow well aimed at a bullseye that's no longer there.

Forgetting the segment you do not want to attract. The definition of a good target carries within it that of an anti-target — profiles you could capture but that hurt your economics or your brand. Demarketing used well is the natural complement to targeting.

Targeting and creative operations

The part that's rarely documented is that each targeted segment requires specific material: copy variants, cases by sector, demos by context, sequences by stage. When a company chooses three segments as targets and produces generic content for all three, the targeting is decorative. When it produces distinguishable material for each, the targeting works.

That specific production is an operational problem, not a strategic one. Without an editorial calendar that distributes production by segment, approval workflows that allow variants without jamming up the rest, and brand management that keeps consistency across the variants, the three segments end up receiving the same averaged message and the targeting loses its reason for being. That connects directly with the discipline of creative operations.

At Polimake this coordination has three surfaces: Studio to schedule production by target segment with assigned resources, Studio to produce the variants with a consistent brand system, Media as the repository where the assets by segment are tagged and accessible for reuse.

---

If you lead strategy, performance, or product and you've gotten this far looking for an answer about targeting, the most useful thing you can take away is probably the most uncomfortable: targeting well means saying no to legitimate segments, and accepting that the large platforms' algorithm already does part of the work that used to be yours. The skill has changed, but the strategic decision of who to serve is still human.

To round it out, market segmentation covers the previous step, digital positioning covers the next one, and direct advertising covers how the target translates into paid activation.

Quick references

• Market segmentation — the step before targeting.
• Digital positioning map — the third step of the STP model.
• Market segmentation criteria — variables that feed both segmentation and targeting.
• Demarketing — the complementary discipline of who to deliberately not serve.
• Direct advertising — how targeting is executed in paid activation.
• Market research — the data that supports an honest targeting decision.