How to Strengthen Your Sales Funnel with Remarketing: Retargeting Step by Step

Since 2012 I have opened hundreds of ad accounts, and the same remarketing setup greets me in most of them: one list called "All visitors", one banner, and a membership duration pushed to the maximum. That is not a strategy. Instead, it is a reminder that follows people for a year and a half. In this guide I explain how I build remarketing as a layered system that plugs the leaks in a sales funnel, with the platform rules and sources you need to do it yourself.

How do you strengthen a sales funnel with remarketing?

Remarketing strengthens a sales funnel by grouping people who visited your site but did not convert into lists based on the funnel stage they reached, then showing each list a message that matches that stage. The working rule is a five part chain: funnel layer, list rule, message, duration and measurement. Skip one link and the chain breaks.

In practice this means a blog reader, a product page visitor and a cart abandoner never see the same ad. For example, the reader gets a deeper piece of content, the product visitor gets a comparison, and the abandoner gets a reason to finish. Therefore the funnel stops leaking at several points at once rather than at one.

The rest of this article walks through that chain. First I clear up the terminology, then I show the infrastructure, the list rules per stage, the durations, the platform differences, the budget split and the measurement traps I see most often.

What is remarketing, and is it different from retargeting?

Remarketing and retargeting describe the same principle: reaching people again after a first interaction with your brand. The platforms simply use different labels. Google Ads calls the lists data segments, Meta calls them custom audiences, and LinkedIn calls the feature Matched Audiences. However, the mechanics do not change from one name to the next.

Some marketers reserve "retargeting" for cookie or pixel based ads and "remarketing" for email follow ups. That distinction made sense a decade ago. Today every major platform mixes site behaviour, uploaded customer data and engagement signals in the same audience tool, so I use the two words interchangeably in this guide.

What matters more than the label is the source of the audience. There are three sources you will work with:

• Site behaviour collected by a tag or pixel, such as page views and add to cart events.
• Customer records you upload, such as hashed emails and phone numbers.
• Platform engagement, such as video views, lead form opens or profile visits.

Awareness content feeds the first source. If your blog does not attract readers, your top of funnel lists stay empty, which is why remarketing and SEO consulting belong in the same plan rather than in separate budgets.

Why does a sales funnel leak without remarketing?

Because most people leave before they buy, and without a follow up channel you pay to acquire them once and never speak to them again. The Baymard Institute compiles 50 separate studies on this topic, and its average cart abandonment rate stands at 70.22 percent, last updated on 22 September 2025.

The reasons behind that number matter, because each reason points to a specific remarketing message. Excluding people who were only browsing, Baymard lists the main causes as follows:

• Extra costs such as shipping, tax and fees were too high: 40 percent.
• Delivery was too slow: 20 percent.
• Shoppers did not trust the site with their card details: 19 percent.
• The site forced them to create an account: 18 percent.
• The checkout process was too long or complicated: 17 percent.

Notice that none of these reasons is "the price of the product". In other words, a discount ad answers a question nobody asked. A shipping transparency ad, a delivery time ad or a guest checkout ad answers the real objection. That is the core of what I do in e-commerce consulting projects: read the leak, then write the message.

Which infrastructure do you need for remarketing?

You need one tag per platform, working conversion events, and the audience sharing links between your analytics and your ad accounts. Without the tag nothing gets recorded; without the events you cannot exclude buyers; without the sharing your lists never reach the campaign.

On Google, a visitor joins a data segment within seconds of opening a page that carries the Google tag, and each new visit refreshes the timestamp. According to the Google Ads help centre, the default membership duration is 30 days, the maximum is 540 days, and a list needs at least 100 active users in the last 30 days before it can serve ads on Display, Search or YouTube.

GA4 adds a second layer. A standard property allows up to 100 audiences, each with a membership duration of up to 540 days. When the Google Ads link and personalised advertising are on, GA4 shares those audiences automatically and prefills them with up to 30 days of historical data where available. Consequently, a fresh remarketing campaign does not start from zero.

Also configure Consent Mode from day one. Otherwise European visitors who decline cookies silently drop out of your lists and your measurement.

How do you split remarketing lists by funnel stage?

Split them by the deepest page type a person reached. That single rule produces four clean layers: awareness, consideration, decision and customer. Each layer has its own list rule, its own duration, its own message and its own exclusions. The table below is the template I start every project with.

Note that the durations are starting ranges from my field experience, not guarantees. Moreover, the exclusion column is the part most accounts forget. If a cart abandoner also sits in the awareness list, they see two competing messages and you pay twice. Therefore every lower layer excludes the layers above it, and every layer excludes converters.

How many days should a remarketing list last?

As long as the buying decision takes, and not a day longer. Google defaults to 30 days and allows 540; Meta caps website custom audiences at 180 days. However, those are ceilings, not recommendations. The right number depends on price and decision time.

My starting ranges, based on field experience and not guaranteed, look like this:

• Cart or checkout abandoners: 7 to 14 days, because intent fades fast.
• Product or service page visitors: 30 days, matching the platform default.
• Blog and guide readers: 60 to 90 days, because they are still learning.
• B2B quote requests: 90 to 180 days, because committees decide slowly.

In short, a simple decision tree helps. Low price and impulse purchase: short window. High price and several decision makers: long window. Seasonal product: window equal to the season plus two weeks. In addition, I often keep a second list of the same rule with a longer window, so I can test a soft reminder after the intense phase ends.

Can small websites run remarketing?

Yes, if they respect the audience thresholds and build lists wide enough to cross them. On Google you need 100 active users in the last 30 days per list. For uploaded customer lists, Google's Customer Match documentation sets the threshold at 100 active users for lists uploaded or refreshed after 1 February 2024, while older lists still follow the 1,000 matched users rule.

LinkedIn is stricter, with 300 member accounts before an audience becomes usable. So a site with 2,000 monthly visitors cannot run a seven day cart list on LinkedIn, but it can run a 30 day all visitors list on Google.

When a list stays too small, I use three levers, in this order:

1. Widen the rule, for example from "cart page" to "any product page".
2. Lengthen the window, for example from 14 days to 30 days.
3. Combine sources in a GA4 audience, for example site visitors plus YouTube viewers plus uploaded customers.

Still, do not force it. If a small site can only fill one list, one well written list beats four empty ones. Grow traffic first and add layers as the numbers allow.

Which message should cart abandoners see?

A message that answers the reason they left, in the order Baymard measured. Since extra costs top the list at 40 percent, the first creative should show the total cost including shipping. Discounts come last, if at all, because a discount trains visitors to abandon on purpose.

In addition, sequence matters as much as content. Days one to three: the objection answer. Then, from day four to day seven: social proof and reviews. Finally, days eight to fourteen: a soft reminder or a small incentive. After that, the person moves back to the consideration list and the pressure drops.

One honest caveat: no ad fixes a broken checkout. If the form asks for eleven fields, remarketing sends people back to the same wall. In those cases I fix the page first, which is often a web design job rather than an advertising job.

How do you run remarketing with a customer list?

You upload hashed emails, phone numbers or addresses, the platform matches them to logged in users, and you target or exclude that group. Google calls this Customer Match. For a list to stay eligible, at least 100 members must have been added or updated within the last 540 days.

Also, customer lists unlock the bottom layer of the funnel, the one most advertisers ignore. Three uses pay for themselves quickly:

• Cross sell: buyers of product A see product B after a sensible interval.
• Reorder: consumable buyers see a reminder shortly before the product runs out.
• Exclusion: current customers stop seeing acquisition ads meant for strangers.

Mix identifiers when you can. A file with email and phone matches more people than email alone. In addition, refresh the file on a schedule, weekly for e-commerce and monthly for services, so the 540 day rule never catches you off guard.

There is also a legal side. Under KVKK in Turkey and GDPR in Europe, uploading customer data for advertising needs explicit consent that covers this purpose. I ask for the consent text before I ask for the file. If the consent is missing, the list waits.

How does remarketing on the search network work?

Remarketing lists for search ads, usually shortened to RLSA, let you treat a known visitor differently when they search on Google. You can raise bids for them inside an existing campaign, or you can build a separate campaign that only shows to list members, often with broader keywords and bolder copy.

Before you touch a bid adjustment, know the baseline. The LocaliQ and WordStream search advertising benchmarks, covering Google Ads and Microsoft Ads across more than 20 industries with data updated on 1 June 2026, put the all industry averages at a 6.64 percent click through rate, a $5.42 cost per click, an 8.18 percent conversion rate and a $66.69 cost per lead.

Those figures describe search overall, not remarketing specifically, and I do not claim a separate remarketing rate. Instead, use them as the reference line. If your returning visitors convert well above your own account average, a higher bid for them is justified. If they do not, the list is telling you something about the landing page, not the bid.

My preferred setup is the separate campaign. It keeps budgets and reporting clean, which matters when I manage accounts under my Google Ads management service and need to show a client exactly what returning visitors cost and return.

How do you separate remarketing in Display, YouTube and Demand Gen?

By never mixing prospecting and remarketing in one campaign. The two audiences behave differently, cost differently and need different creative. So each platform gets a pair: one campaign for strangers, one for known visitors, with mutual exclusions between them.

The exclusion is the part that saves money. First, the prospecting campaign excludes every remarketing list, so it keeps hunting for new people. Second, the remarketing campaign excludes converters, so it stops paying for people who already bought. Without this pair of exclusions, the algorithm quietly spends the prospecting budget on your easiest audience and reports a great return that you would have earned anyway.

Next, format choice follows the funnel layer:

• Display suits the consideration layer, with product feeds and comparison messages.
• YouTube suits the awareness layer, with short explainers for blog readers.
• Demand Gen suits the decision layer, because it combines YouTube, Discover and Gmail with strong visual formats.

Also check where your lists actually serve. A list with 100 active users can serve on all three networks, but a list that just crossed the line will deliver slowly. Give it two weeks before you judge the results.

How do you set up retargeting on Meta and Instagram?

Install the Meta Pixel, fire the standard events, and build website custom audiences from those events. Meta's business help centre confirms that these audiences come from Pixel events, that retention runs up to 180 days, and that the audience keeps updating with people who meet the chosen rule.

The three events that matter most for a funnel are ViewContent, AddToCart and Purchase. From them you can build the same layers described earlier:

1. ViewContent in the last 30 days, excluding AddToCart: consideration layer.
2. AddToCart in the last 14 days, excluding Purchase: decision layer.
3. Purchase in the last 180 days: customer layer for cross sell and for exclusion.

In practice, Meta's strength is creative reach on Instagram, so I use it for the visual objection answers: delivery time, packaging, real reviews. Its weakness is measurement after browser restrictions, which is why the Conversions API belongs next to the Pixel. Otherwise the AddToCart list undercounts and the Purchase exclusion leaks, and you end up chasing customers who already paid.

How do you run B2B remarketing on LinkedIn?

You install the LinkedIn Insight Tag, wait for at least 300 member accounts to enter an audience, and pick a lookback window between 30 and 365 days. Those are the platform rules for website retargeting through Matched Audiences, and the 300 threshold is why many B2B sites need a longer window than they would use on Google.

The layers translate well to B2B, with different page types:

• Awareness: visitors to your company page or a thought leadership article.
• Consideration: visitors to a service or pricing page.
• Decision: visitors who opened a quote or demo form without submitting it.
• Engagement: people who watched a video ad past the halfway mark.

Because cost per click on LinkedIn runs high, I keep remarketing there narrow and content led. A case study, a short document ad or a webinar invite performs better than a "contact us" banner. Then the actual conversion often happens on Google search or by direct visit, which is one more reason to measure across channels rather than inside one platform.

How do you set frequency caps without annoying people?

Set a cap per person per week, rotate creatives, and exclude converters within hours rather than days. I deliberately avoid quoting one magic number, because the right cap depends on the layer. A cart abandoner tolerates more contact in three days than a blog reader tolerates in three weeks.

Three principles keep the experience respectful:

• Shorter lists get higher frequency; longer lists get lower frequency.
• Every list carries at least three creatives so the same banner never repeats all week.
• Purchase and lead events feed the exclusion list as close to real time as the platform allows.

Then watch two signals in the reports. When the click through rate falls week after week while impressions stay flat, fatigue has set in. When negative feedback on Meta rises, the cap is too loose. In both cases, lower the cap or refresh the creative before you touch the budget. Above all, remember that the goal is to be useful at the right moment, not to be everywhere.

How should you split the remarketing budget across the funnel?

Give the biggest share to the layer closest to the sale, then work upward. As a labelled example calculation with a monthly budget of 30,000 TRY: 45 percent to the decision layer, 30 percent to consideration, 15 percent to customers and 10 percent to awareness. These proportions are a starting point, not a rule.

Here is a second labelled example to show why the decision layer earns its share. Assume 10,000 monthly visitors, 1,000 of whom add to cart. At a 70 percent abandonment rate, 700 people leave. Put them in a 14 day list and assume a hypothetical 3 percent return rate: that is 21 recovered orders. With an average basket of 1,500 TRY, revenue reaches 31,500 TRY. If that list costs 4,000 TRY to run, the return on ad spend is 7.9.

Every number in that example is an assumption and none of it is guaranteed. Your own return depends on margin, basket size and how well the message matches the objection. To run the same arithmetic with your figures, use my ROAS calculator and change one variable at a time.

Then review the split monthly. If the customer layer outperforms, move budget there. If awareness lists stay tiny, pause them and fund content instead.

Are third-party cookies gone, and does remarketing suffer?

No, they are not gone in Chrome, and cookie based remarketing continues there. On 22 April 2025 Google announced on the Privacy Sandbox blog that it would not introduce a separate prompt for third-party cookies and would keep its current approach, with users managing cookie preferences in Chrome's Privacy and Security settings.

That said, my reading of the situation is that reach keeps shrinking. Safari and Firefox restrict cross site tracking, and a share of users decline consent. Chrome's decision removed a cliff, but it did not restore what other browsers already took away. Plan for a smaller, better identified audience rather than for the old universal one.

Google's own guidance points the same way. Its third-party cookie FAQ for advertisers recommends a multi pronged approach built on first party data, AI powered solutions and privacy preserving technologies, and notes that these solutions already contribute to campaign performance today. In practice that means three habits:

• Collect first party data through accounts, newsletters and purchases with proper consent.
• Move tagging server side so measurement survives browser limits.
• Use Consent Mode so modelled conversions fill part of the gap.

How do you measure remarketing results honestly?

By separating what the campaign caused from what would have happened anyway. Remarketing reaches people who already know you, so it always looks good in a last click report. The trap is view through conversions: a person sees a banner, ignores it, buys later through search, and the banner takes credit.

I use three checks in every account:

1. Report click conversions and view through conversions separately, and judge budget on clicks first.
2. Run a holdout when volume allows it, by excluding a random slice of the list and comparing purchase rates.
3. Tag every remarketing link with UTM parameters, so GA4 shows the channel split without relying on platform reports.

For the third check, keep a naming convention and stick to it. A campaign called "rem_cart_14d_meta" tells you the layer, the window and the platform at a glance. My UTM builder generates consistent links, and consistency is what makes the report readable six months later.

Finally, watch the funnel metrics, not only the campaign metrics. If cart recovery rises while total revenue stays flat, the campaign may be shifting sales rather than adding them.

What are the most common remarketing mistakes?

The same five errors appear in most accounts I audit, and each one has a quick fix.

1. One list, one ad. Fix: at least three layers with separate messages.
2. Converters stay in the lists. Fix: an exclusion list fed by the purchase or lead event.
3. Membership set to 540 days for everything. Fix: durations matched to decision time.
4. A discount on day one. Fix: objection answers first, incentives last.
5. Prospecting and remarketing in one campaign. Fix: separate campaigns with mutual exclusions.

In addition, two softer mistakes deserve a mention. The first is judging the campaign after three days, before the lists have filled and the algorithm has learned. The second is creative laziness, where the retargeting banner is the same as the acquisition banner with the word "back" added. However, none of these is hard to repair once you see it. The hard part is noticing, which is why a monthly audit of list sizes, exclusions and frequency belongs in your routine.

How do you set up remarketing step by step?

Follow the ten steps below in order. Each one depends on the previous, so resist the urge to jump to campaign creation.

1. Install the Google tag, the Meta Pixel and, for B2B, the LinkedIn Insight Tag, with Consent Mode active.
2. Define conversion events: purchase, lead, add to cart, quote started.
3. Build GA4 audiences for each funnel layer with the starting durations above.
4. Link GA4 to Google Ads and confirm the audiences appear in the shared library.
5. Create the layer lists on Meta and LinkedIn with matching rules.
6. Create exclusion lists from converters and from deeper layers.
7. Split campaigns: prospecting versus remarketing, with mutual exclusions.
8. Prepare at least three creatives per layer that answer a specific objection.
9. Set frequency caps by layer and schedule creative rotation.
10. Review list sizes, click conversions and view through conversions every week.

Also expect the first two weeks to be quiet while the lists cross their thresholds. After that, the decision layer usually shows results first, followed by consideration and then customers. Keep a changelog, because the value of this system compounds when you can see which adjustment moved which number.

When should you get professional help with remarketing?

When your lists exist but you cannot tell which layer earns money, or when the setup steps above sound reasonable and you still do not have the hours to run them weekly. Everything in this guide is doable in house. I work without an agency layer, follow this exact system, and track results through my own CRM around the clock. You can see the scope on the packages page, or send me a short note through the contact page with your platform, monthly traffic and current spend. I will reply with an honest view of whether remarketing is ready to help you yet.