A Short History
Retargeting has been around long enough to have generations. Knowing which one you are running tells you most of what you need to know about how much money you are leaving on the table.
Retargeting 1.0 is a single pixel and a single ad. You drop a tag on your site, and everyone who visits any page gets shown the same ad until the budget runs out. It works well enough to look like a success, which is exactly why most advertisers never move past it. The overwhelming majority of retargeting being run today is still version 1.0.
Retargeting 2.0 introduced segmentation. Instead of one audience you build several, based on what people did: visited the pricing page, watched the video, started a cart. Each segment gets a message that fits. A small minority of advertisers have made it this far, and the ones who did typically cut their wasted spend sharply the moment they stopped treating every visitor as identical.
Retargeting 3.0 is the version almost no one runs, and the one this article is about.
Where the Budget Leaks
Before the fix, it is worth being precise about what version 1.0 actually wastes, because "it works" hides a great deal of waste.
- It pays full price for low-intent traffic. Someone who bounced off your homepage in four seconds and someone who sat on your checkout page are worth wildly different amounts. Version 1.0 bids the same for both.
- It keeps paying after the sale. Without a customer exclusion, you spend for weeks showing "come back and buy" ads to people who already bought. That is not retargeting. It is an apology you are paying to deliver.
- It burns frequency. The same ad, over and over, to the same person. Past a handful of views, each additional impression does less and annoys more, and you are charged for every one.
- It has no finish line. A visitor from 89 days ago sits in the same pool as one from yesterday, despite being a fraction as likely to convert.
Estimates vary, but it is common for half of a naive retargeting budget to be spent on impressions that were never going to move the needle. The goal of Retargeting 3.0 is to claw that half back.
Retargeting 3.0: Laddering
Versions 1.0 and 2.0 both think in terms of audiences, a flat set of buckets. Retargeting 3.0 thinks in terms of a ladder.
A ladder is ordered. Each rung represents a deeper level of intent, and a prospect climbs from one rung to the next as they take more meaningful actions. The message, the budget, and the offer all change as they climb. Someone is never just "in the retargeting audience." They are on a specific rung, and your job is to give them the one nudge that moves them up.
"Stop asking who visited my site. Start asking how close this person is to buying, and what single next step gets them closer."
The shift sounds subtle. In practice it changes everything, because it forces every ad to have a job. An ad is no longer "remind them we exist." It is "move this person from rung two to rung three."
Building Your Ladder
Four rungs, four jobs, four budgets
Rung 1 — Visited: low intent, low budget, a soft brand or value message. Rung 2 — Engaged: read a post, watched the video, viewed pricing; warm them with proof and specifics. Rung 3 — High intent: started a cart or booked a call and bailed; spend the most here, with urgency and risk reversal. Rung 4 — Customer: exclude from the prospecting ladder entirely and move them to a separate retention and upsell track.
Two rules make the ladder work. First, exclusions matter as much as inclusions. Each rung must exclude every rung above it, or your high-intent message competes with your low-intent message for the same person. Second, budget follows intent. The instinct is to spend evenly. The money belongs disproportionately on rung 3, where a small nudge converts a near-certain buyer, and barely at all on rung 1.
Get 7 Proven Trainings Free
I've sold these trainings individually clients have paid me $10,000+ to implement them. Enter your email and the first one arrives immediately.
Setting It Up
The ladder is the strategy. Here is how to make it real without drowning in complexity:
- Run everything through a tag manager. Use Google Tag Manager rather than hard-coding pixels. You will be defining and revising audience triggers constantly, and you want that in one dashboard, not scattered through your site's code.
- Define rungs as events, not just pages. "Viewed pricing" is a page. "Watched 50% of the demo" and "added to cart" are events. The deeper rungs of a good ladder are almost always event-based.
- Build the exclusion audiences first. Set up "all customers" and each rung's exclusions before you launch a single ad, so no one is ever served two competing messages.
- Account for how tracking actually works now. Third-party cookies have largely collapsed and browser restrictions keep tightening. Lean on first-party data, the ad platforms' server-side conversion APIs, and shorter audience windows. The ladder strategy is unchanged. The plumbing underneath it has to be modern.
Measuring It Honestly
Retargeting is the most over-credited channel in marketing. It shows ads to people who were already likely to convert, then takes the credit when they do. Last-click attribution makes a 1.0 campaign look brilliant.
The only honest measure is incrementality: how many conversions happened because of the retargeting that would not have happened anyway. Run a holdout. Withhold retargeting from a random slice of your audience and compare their conversion rate to everyone else's. The gap is your real return. Everything above that gap is the campaign congratulating itself.
A Retargeting 3.0 ladder will often show lower reported revenue inside your ad platform than a 1.0 campaign did, because you stopped paying to advertise to guaranteed buyers. The incrementality test is what reveals that your real, profit-generating return went up. Measure the right thing, or the upgrade will look like a downgrade.
The retargeting ladder works on traffic that has already reached you. To put your pixel in front of qualified traffic that has not yet, on partner sites and partner pages, see pixel swapping.