Google Ads Remarketing: How to Win Back the 97% Who Left

Google Ads remarketing exists because of a number every advertiser has seen and most have stopped reacting to: the overwhelming majority of first-time visitors leave without converting. Depending on the industry, somewhere between 95% and 98% of the traffic you paid for goes away, and for most accounts nothing whatsoever happens next.

Remarketing is the mechanism for the "next." Done well, it's typically the cheapest cost per conversion in the entire account, because you're advertising to people who have already demonstrated interest rather than to strangers. Done badly — one undifferentiated "all visitors" audience, no frequency cap, the same banner for ninety days — it burns budget and makes your brand feel like it's following people around, which it is.

What Remarketing Is and Where It Runs

Remarketing shows ads to people who have previously interacted with your site, app, or content. Google Ads supports it across several surfaces, and they behave very differently.

Surface	What it does	Typical use
Display remarketing	Banner ads across the Google Display Network	Broad reach, brand recall, cart/quote abandonment
RLSA (search remarketing lists)	Adjusts bids or targeting on search when a past visitor searches again	Highest intent — they're actively searching again
YouTube remarketing	Video ads to past visitors and channel engagers	Consideration, higher-ticket products
Dynamic remarketing	Shows the specific products or services someone viewed	Ecommerce and multi-SKU catalogues
Customer Match	Targets uploaded first-party email lists	Reactivation, upsell, exclusion of existing customers

Most small accounts run only display remarketing, which is usually the weakest of the five. RLSA is generally the highest-return remarketing available to a service business, because it targets the moment someone returns to search — a signal of genuine active intent that display can't match.

Segmentation: The Difference Between Working and Not

The single biggest determinant of remarketing performance is audience segmentation. An "all visitors, 30 days" list treats someone who bounced off your homepage in four seconds identically to someone who spent eleven minutes on your pricing page and started a form. Those are not the same person and should not see the same ad at the same bid.

Build audiences by depth of engagement:

1. Bounced / shallow — one page, under 15 seconds. Usually not worth remarketing to at all. Exclude them and save the money.
2. Engaged browsers — two or more pages, or over 60 seconds. Worth low-bid, brand-level messaging.
3. High-intent page viewers — visited pricing, service detail, or contact pages. Worth substantially higher bids and specific messaging.
4. Started but didn't finish — began a form, started a quote, added to cart. This is the highest-value list in most accounts. Bid aggressively; these people were minutes away.
5. Converted — past leads or customers. Either exclude them from acquisition campaigns entirely, or target them separately with genuinely different offers.

That last one deserves emphasis. Failing to exclude converters is the most common and most expensive remarketing mistake. If you don't exclude them, you spend weeks advertising a free consultation to someone who already booked one, which is both wasteful and slightly insulting.

Membership Duration Should Match Your Sales Cycle

Google's default membership duration is 30 days, which is wrong for most businesses in one direction or the other.

• Impulse or emergency services (emergency plumbing, urgent legal): 3–14 days. After that, they've solved the problem elsewhere.
• Considered services (home renovation, professional services, B2B under $10k): 30–90 days.
• Long-cycle B2B or high-ticket (enterprise software, commercial construction): 180–540 days.

Set duration against how long your buying decision actually takes, which you can measure from your CRM rather than guess.

Running remarketing with a single all-visitors audience? Get a Free Growth Audit →

Frequency: The Line Between Presence and Harassment

There is no universally correct frequency cap, but there is a reliable pattern: response rises for the first several impressions, plateaus, then turns negative as annoyance sets in. Running uncapped display remarketing means a small number of people see your banner hundreds of times, which produces no additional conversions and real brand damage.

Practical starting points:

• Display remarketing: 3–5 impressions per user per day, 15–25 per month.
• YouTube remarketing: 2–3 per week. Video fatigue sets in much faster.
• RLSA: no cap needed. It only triggers when the user searches, so frequency is self-limiting by their own behaviour.

Also rotate creative. The same banner for ninety days is the fastest route to banner blindness, and creative fatigue is measurable — watch click-through rate by ad over time and refresh when it decays.

Setting It Up Properly

The mechanics, in the order they should be done:

1. Get the tagging right. The Google tag on every page, with remarketing enabled, and event tracking for the actions that define your audiences — form starts, pricing page views, quote initiations. If your events are wrong, every audience built on them is wrong.

2. Build audiences before you need them. Lists take time to accumulate the minimum size Google requires before they can serve — 1,000 users for display, 1,000 for search in most cases. Create them now even if you won't run campaigns for a month.

3. Separate campaigns, not shared ones. Remarketing should be its own campaign with its own budget, not a bolt-on to a prospecting campaign. Mixed campaigns make performance impossible to read, and remarketing's superior metrics will mask prospecting's problems.

4. Write messaging that acknowledges the return. A returning visitor doesn't need your value proposition explained from scratch. They need the specific objection that stopped them handled — price, trust, timing, or complexity. "Still comparing quotes? Here's what our pricing includes" outperforms a generic brand banner by a wide margin.

5. Send them somewhere appropriate. A high-intent remarketing click that lands on your homepage wastes the intent. Send pricing-page abandoners to pricing. Send form abandoners back to the form, pre-filled where possible.

6. Exclude your converters, and check the exclusion is applied. Every campaign, every time.

Privacy, Consent, and the Shrinking Cookie

Remarketing has been reshaped by privacy changes over the last several years, and pretending otherwise leads to bad forecasting. Third-party cookie deprecation, browser-level tracking prevention, and consent requirements in the EU and increasingly elsewhere all reduce addressable audience sizes.

Three practical consequences:

• First-party data is now the durable asset. Customer Match lists built from your own email database are unaffected by cookie loss. If you have a customer list and aren't using it for remarketing and for exclusions, that's the first gap to close.
• Consent Mode matters. Where you operate under consent requirements, implementing Consent Mode correctly preserves modelled conversions rather than simply losing the data.
• Audience sizes will look smaller than they used to. This is measurement reality, not a setup error. Compare against current benchmarks, not against numbers from 2020.

Think with Google publishes ongoing guidance on the first-party data transition, and Search Engine Journal tracks the platform-level changes as they land.

RLSA Deserves More Attention Than It Gets

Remarketing lists for search ads are consistently the highest-return remarketing available to service businesses, and consistently the least used — largely because they're less visible than banner ads and require an existing search campaign to attach to.

The mechanic: when someone on your remarketing list performs a search, you can bid differently, target keywords you wouldn't normally bid on, or show different ad copy.

Three ways to use it:

Bid adjustments on existing keywords. Your standard approach. Someone who has already visited your pricing page and is now searching your core commercial term is worth substantially more than a cold searcher. A +30% to +60% adjustment on high-intent lists is a common starting point.

Targeting broader terms you wouldn't otherwise buy. Generic head terms are usually too expensive and too unqualified to bid on cold. Restricted to past visitors only, they become viable — you're paying for a broad term but only reaching people who already know you.

Different ad copy for returning searchers. They don't need the introduction. They need the objection handled. Copy that acknowledges a return ("Still comparing? Here's what's included") outperforms your standard ad against the same query.

RLSA costs nothing to layer onto campaigns you're already running, requires no new creative, and is bounded by the user's own search behaviour so there's no frequency risk. If you implement one thing from this article, make it this.

Customer Match and First-Party Lists

Customer Match — uploading your own customer email list for targeting — is the most durable remarketing asset you have, because it doesn't depend on cookies that browsers are progressively eliminating.

Four uses worth building:

1. Exclusion. The highest-value use for most businesses. Upload your customer list and exclude it from every acquisition campaign. Stop paying to acquire people you already have.
2. Reactivation. Customers who haven't purchased in 12+ months, targeted with a specific reason to return. Costs a fraction of new acquisition.
3. Upsell and cross-sell. Customers of service A targeted for service B. High relevance, high conversion, low cost.
4. Lookalike-style expansion. Using your best-customer list as a seed for similar-audience targeting, where the platform supports it.

Two practical requirements. First, list hygiene — a stale, poorly formatted list matches badly and produces an audience too small to serve. Second, consent: uploading customer data for advertising has real legal requirements depending on your jurisdiction and how the data was collected. Check that your privacy policy and consent capture actually cover it before uploading anything.

Attribution: Judging Remarketing Honestly

Remarketing has a structural measurement problem that makes it look better than it is. You are advertising to people who already showed intent. A meaningful share of them were going to come back regardless. Last-click attribution hands remarketing full credit for those conversions, which is how accounts end up over-invested in retargeting and under-invested in the acquisition that created the audience in the first place.

Three ways to get an honest read:

Run a holdout test. Exclude a random share of your remarketing audience — 10–20% — and compare conversion rates between the exposed and held-out groups over a full sales cycle. The difference is your actual incremental lift. This is the only genuinely rigorous method and it's underused because it requires deliberately not advertising to some people.

Compare against a pre-remarketing baseline. Imperfect, since other things change, but a directional check.

Watch new-customer conversions specifically, not total conversions. If remarketing conversions are up but new customers aren't, you're paying to reach existing ones.

Expect the honest number to be lower than the reported one. That doesn't make remarketing a bad channel — it makes it a channel whose true value is smaller and whose correct budget is therefore smaller than the dashboard suggests.

A Worked Example

A GTA home renovation company was running a single display remarketing campaign — all visitors, 30-day membership, one banner set, no frequency cap, converters not excluded — at about $1,900 a month. Reported cost per conversion looked acceptable, but the sales team said most of the leads were people who had already booked.

That last detail was the tell. The campaign was largely re-serving ads to existing customers and taking credit for conversions that would have happened anyway.

The rebuild:

• Five audiences created by engagement depth, with converters excluded from every acquisition audience
• Membership duration extended from 30 to 120 days, matched to their real decision cycle measured from CRM data
• Bounced/shallow visitors excluded entirely — roughly 44% of the previous audience
• RLSA layered onto existing search campaigns with a +45% bid adjustment for the high-intent list
• Frequency capped at 4 per day and 20 per month; three creative sets in rotation
• Landing pages matched to audience — quote-form abandoners returned to a pre-filled form

Over the following quarter, at the same $1,900 monthly spend: impressions down 52%, genuinely new leads up 61%, and cost per new lead down from an inflated $180 to $71. The RLSA layer, which cost almost nothing to implement, produced the largest single share of the improvement.

The headline number that had looked fine before was fine because it was counting the wrong conversions. This is why remarketing performance should always be reviewed against new-customer conversions rather than total conversions — a point covered more fully in our guide to measuring digital marketing ROI.

Mistakes Worth Avoiding

• Not excluding converters. Wastes money and inflates reported performance.
• No frequency cap. Small number of users, enormous impression counts, brand damage.
• One audience for everyone. Guarantees you underbid on your best prospects and overbid on your worst.
• Remarketing to bounced traffic. Someone who left in four seconds is not a warm lead.
• Stale creative. Refresh at least quarterly, sooner if click-through is decaying.
• Attributing every remarketing conversion to remarketing. Many of those people were coming back anyway. Judge the channel on incremental lift, not last-click credit.
• Running remarketing before fixing the site. If the landing page is the reason they left, remarketing just pays to show it to them again. Fix conversion first.

Creative That Works for Returning Visitors

Remarketing creative is usually just the prospecting banner reused, which wastes the one advantage remarketing has: you know something about the person seeing it.

Four principles that consistently lift performance:

Match the message to the audience's depth. Someone who viewed a service page needs different messaging from someone who abandoned a quote form. The first needs a reason to consider; the second needs the specific friction removed — usually price uncertainty or a sense that the process will be complicated.

Handle the objection, don't repeat the pitch. They've already seen your pitch. The reason they left is the thing to address. If you don't know what that reason is, ask your sales team — they hear it daily.

Give a reason to act now. Not manufactured scarcity, which erodes trust, but a genuine one: seasonal timing, a booking window, a real limited availability. Absent urgency, a warm prospect stays warm indefinitely.

Refresh on a schedule, not on inspiration. Set a quarterly creative refresh in the calendar and watch click-through decay by ad to catch fatigue earlier. Three sets in rotation beats one set shown three times as often.

Sizing and format matter too — build the full complement of responsive display sizes rather than two, since inventory availability varies enormously by placement and a limited size set caps your reach for reasons that have nothing to do with your targeting.

Where It Fits

Remarketing is a multiplier on existing demand, not a source of it. It performs in proportion to the quality and volume of traffic feeding it, which means it should be built alongside acquisition rather than instead of it — a structural point our Google Ads management work treats as a prerequisite, and one that sits inside the broader logic of integrated marketing.

If your account structure or wasted spend problems are unresolved, fix those first: negative keywords and campaign structure both produce faster returns than a remarketing rebuild, and remarketing built on top of a leaky account just retargets the wrong people more efficiently.

Is your remarketing generating new customers, or just following your existing ones around?

We audit your audience segmentation, exclusions, frequency, and attribution, then show you what share of your reported remarketing conversions are genuinely incremental — and what it would take to fix the rest.

Get a Free Growth Audit →

---

Related reading: Google Ads Negative Keywords | Google Ads Campaign Structure | How to Measure Digital Marketing ROI