Targeting strategies that matter
Everyone wants perfect targeting right now. Platforms feel loud, budgets feel tight, and there is new talk about Safari limiting tracking. So let’s get practical about what actually moves numbers.
When people say targeting, they usually mean a mix of audience filters, channel choices, and timing rules. The easy trap is to stack every filter you can find until your audience is a tiny puddle that costs a fortune per impression and never leaves the learning phase. The better way is to work three levers with intent. Who you talk to, when you show up, and where you put the message. Get those levers right and your creative suddenly looks smarter, your analytics makes sense, and your spend finds a home. Get them wrong and you chase ghosts across dashboards. The big theme today is this. Precision loses to reach when you forget freshness, and first party beats third party when the clock is ticking.
That is the frame for the rest of this piece.
Start with first party, then earn the right to expand
Your first party data is your best targeting asset. Email lists, site visitors, app users, CRM segments, purchase history, churn cohorts, and any consented behavior on your properties. On Facebook you can build Custom Audiences from emails and site events, then jump to Lookalike. On AdWords you have Customer Match and Similar Audiences. In a world where Safari is about to make cookie based retargeting harder, owning the relationship matters. You get stronger match rates, cleaner attribution, and a faster feedback loop. If you sell apparel, upload your past buyers, split by men and women, exclude recent purchasers for a cooling period, and then create separate lookalikes by category. If you sell a B2B tool, sync your closed won list from your CRM, exclude active deals, build a lookalike on the smallest seed that still has quality, and keep your creative anchored to the one problem you actually solve. If you run a game, feed your payer cohort to Facebook and lift your CPI caps for the lookalike only. That is the bedrock.
Do not skip the cleaning step.
Practical checklist. Make a habit of exclusions. They are the cheapest targeting there is. Exclude buyers for a smart cooldown window. Exclude press, employees, and support domains. Exclude non core countries that are cheap clicks with zero conversions. Exclude the bottom 10 percent of placements in the last 30 days. Exclude anyone who already completed the step your ad is asking for. This keeps frequency honest and it lets your budget work on people who can actually say yes.
Exclusions build profit faster than fancy segments.
Lookalike is the new broad, if your seed is right
Lookalike audiences are the friend that does not get enough credit. Facebook and Google have real reach and real data at the account level. Feed them a clean seed and they do a lot of heavy lifting. The trick is seed quality. If you toss a giant mixed bag of visitors into the seed, the model chases noisy signals. If you give it 3 to 10 thousand people who took the action you actually want, you get magic. For retail this means high AOV buyers. For SaaS this means trialists who activated the sticky feature. For a subscription app this means users who made it past day seven and kept a streak. Then you build a 1 percent lookalike for control, plus a 2 to 3 percent for reach. If your country is small, keep the size low and let time do the work. Do not be tempted to stack tons of interest targeting on top of lookalike. You pay twice when you do that. Let the model run and watch your frequency, CPM, and conversion rate trend over two weeks before you call it.
Good seed in, good results out.
Context beats creepiness
There is a lot of talk about tracking right now. Apple showed Intelligent Tracking Prevention at WWDC and it is going to squeeze cookie based retargeting on Safari. Some people will panic. Do not. Lean into contextual targeting where people show intent by what they are doing in the moment. If you sell running shoes, aim for pages about marathon plans on premium publishers and sponsor a weekly training column. If you sell accounting software, place ads next to year end checklists and tax deadline reminders on sites that decision makers actually read. If you sell design tools, own tutorial keywords on YouTube and sponsor creators with real audiences. Context scales without looking creepy, and it keeps working when third party cookies age out.
Right message, right page, right now.
Reach, then tighten
The path I keep coming back to is simple. Start wider than your gut wants, then tighten based on real fail points. Use a broad interest that actually reflects a behavior, not a vanity topic. Use geo filters that match your ship list, not your dream map. Use device filters when the funnel tells you to. If your trial flow is smooth on desktop but clunky on mobile, push more top funnel on mobile and shift conversion spend to desktop. If your app install rate is hot on Android but your pay rate is double on iOS, split your budget like a poker player and let each OS carry its weight. Channel mix and device mix are targeting levers, not just delivery details. Let your budget learn at volume for a week or two, then pause the bottom quartile placements and move that money to a second lookalike or a stronger context.
Reach first, prune second.
Recency and frequency: the quiet power couple
If you only change one thing this month, fix recency and frequency. Most accounts either hammer people with 12 views a day or starve a good group with one view a week. Neither works. Set a daily cap that fits your funnel length. For short calls to action like a flash sale, three per day per person is plenty. For a higher ticket item, one per day with a strong follow up on day three can work. Use shorter recency windows for retargeting on mobile where attention is fleeting, and longer windows on desktop where people compare options. In programmatic tools like DoubleClick, aim for a seven day retargeting pool for cart abandoners and a thirty day pool for product viewers. On Facebook, watch Reach and Frequency on the ad set and rotate creative as soon as your CTR drops below your weekly median. This is quiet work that saves real money.
Control the drumbeat.
Segment by journey, not by job title
Titles lie. Journeys do not. If you sell to marketers you already know that a Director can be a spectator and a Specialist can be the buyer. Segment by the step someone is in, not the noun in their bio. Typical journey slices look like this. People who just learned your name. People who showed interest. People who tried but did not finish. People who finished and need a reason to return. Each slice deserves different copy, different offers, and different rules of engagement. Top slice wants proof you are real. Mid slice wants a sharp comparison and a clear win. Finishers want a nudge and some reassurance. Returners want value without noise. Map that to your channels. You can run broad video to the first slice, product demos to the second, time bound retargeting to the third, and loyalty messages to the fourth. The targeting trick is not a fancy interest. It is sending the right story to the right step.
Steps beat titles every time.
Geo and time matter more than people think
Set your geo targeting to where you can deliver. That sounds obvious until you watch a campaign chew through two thousand clicks from a place you do not serve because the CPM looked cheap. Start with your ship list, your sales territories, your language comfort, and your support hours. Then layer in time of day. If your sales team calls back within an hour during business hours, shift more budget to local daytime and slow down at night. If you sell food delivery, own the zones around big office clusters at lunch and the residential rings after six. If you run a mobile game, push nights and weekends. Platform tools let you skew bids by hour and by city. Use that. It is not flashy but it compounds weekly.
Spend where you can follow through.
Creative rotation is targeting too
People act like targeting sits in a box and creative sits in another box. They are the same box. A product video with a clear demo is targeting people who need proof. A simple image with a strong offer is targeting people who already know you. A carousel that shows three use cases is targeting three mindsets at once. Rotate creatives by journey slice and device. Short caption and tight visual on mobile. Longer copy and comparison table on desktop. Keep at least three live variants per ad set so the platform can route impressions to the right message. The day you see CTR slide or comments repeat the same confusion, swap the ad. That tweak is a targeting change because you are changing who raises a hand.
Change the story, change the crowd.
DMPs, CDPs, and a quick word on data plumbing
If you have scale, a DMP like Krux or Lotame or BlueKai can centralize segments for programmatic buys. A newer wave of customer data platforms is trying to make that easier for mid sized teams. Use them for unifying events and identities, not for collecting every click in the world. Keep your events clean, push the segments that actually drive spend, and keep consent and privacy front and center. The trend line is clear. Browsers are making cross site tracking harder. So your smartest bet is to get your house data right, earn direct signals, and buy context with quality inventory. That mix will age well while the rest shifts under your feet.
Good data in, steady nerves out.
Measurement: pick an attribution rule and move
Attribution can turn into a swamp. Last click makes search look like a hero and hides upper funnel. First click flips that story. Position based finds a middle ground. The mistake is changing the rule every week. Pick one model that fits your sales cycle and stick with it long enough to learn. Use platform lift tests when you can. Facebook brand lift for awareness. Facebook conversion lift for performance if your spend is big enough. In AdWords, use Drafts and Experiments to split bids and see deltas for real. For landing pages, Google Optimize is live and good enough to ship clean A B tests without a dev queue. Whatever you choose, write it down and share it with the team so your readouts match your buys.
Commit to a rule, then improve the inputs.
Three field guides you can steal today
Retail example. A mid sized apparel shop uploads a six month buyer list, splits by category, removes anyone who bought in the last fourteen days, and makes 1 percent lookalikes. Campaign structure. Prospecting with lookalike and interest in fashion magazines and YouTube hauls. Retargeting with a seven day cart window and free returns message. Creative. Lifestyle on prospecting. Product detail on retargeting. Rules. Frequency cap at three per day on retargeting, one per day on prospecting. Geo. Only ship zones with stock depth. That alone can lift ROAS without a single new tactic.
Simple beats clever.
SaaS example. A workflow tool maps the trial journey. Sign up, invite teammates, complete first project. Seeds. Lookalike built from users who invited two teammates and completed a project in week one. Prospecting mix. Facebook lookalike plus search for pain keywords plus partner blogs. Retargeting. People who started trial but did not invite anyone get a team value story. People who invited but did not complete a project get a template demo. Budget. Sixty percent prospecting, forty percent retargeting. Analytics. Position based attribution and a simple daily cohort sheet. This setup pulls cost per qualified trial down without slicing audiences to dust.
Journey over vanity titles.
Mobile game example. The studio splits by payer and non payer behavior. Seed. Last thirty day payers who made two payments. Lookalike at one percent for the United States and two percent for Canada. Creative. One ad shows moment to reward and the store. One ad shows skill mastery loop. Placement. More on video for awareness and more on static for retargeting. Bids. Higher caps on the payer lookalike and lower caps on broad. Frequency. Two per day on prospecting and three per day on retargeting. This mix reduces CPI on the right people and keeps the ad server from blasting the same clip ten times.
Quality beats volume when the wallet is involved.
What to stop doing this week
Stop stacking ten interests plus a lookalike plus a placement filter and calling it sharp targeting. Stop running retargeting to people who already bought. Stop reporting weekly with a new attribution rule every time the chart looks sad. Stop chasing the lowest CPM if it is not bringing people who convert. Stop showing the same ad for a month because you got busy. These are small fixes that pay in days, not quarters.
Remove the sand from the gears.
The short list that matters
Own your first party data. Use exclusions with discipline. Seed lookalike with quality, not volume. Buy context where intent lives. Let campaigns learn, then prune. Control recency and frequency. Segment by journey. Respect geo and time. Treat creative as targeting. Pick an attribution rule and run. That is the playbook that fits the tools we have today and the changes shipping this fall.
Aim small, learn fast, spend smarter.