A fact that surprises many business owners: in most industries, fewer than 5% of visits to a website convert into anything (a purchase, a contact, a request) on the first visit. The rest, the vast majority, leave without doing anything, and if nothing is done about it, they are lost forever. Remarketing (showing ads specifically to people who already visited the site, on other sites or apps they visit afterward) is the tool designed exactly to recover part of that lost opportunity, and Google Signals is the technical piece that makes it possible within Google's ecosystem.
What Google Signals actually is
Google Signals is a Google Analytics 4 feature that, when enabled, links a user's behavioural data on your website with their Google account (if that person has ad personalisation enabled on their account), which unlocks features like demographic and interest reports, cross-device tracking (someone who visits from mobile and later buys from a computer is recognised as the same person), and above all, the creation of remarketing audiences that can be used directly in Google Ads.
Why it is often misconfigured or switched off
It is common to find Analytics accounts where Google Signals was never enabled (often because whoever set up the account did not know it existed or how important it was), or where it was enabled once and never checked afterward to confirm it still worked correctly after some later change to the property settings. Without this feature active, you silently lose the ability to build remarketing audiences based on real on-site behaviour, one of the best-performing forms of segmentation in paid advertising.
How this translates into actual campaigns
Once Google Signals is active and the corresponding audiences have been created in Analytics, those audiences can be used in Google Ads for several different strategies: showing ads to whoever visited a specific product page but did not buy, excluding people who are already customers from acquisition campaigns (so as not to waste budget showing generic ads to someone who already knows the business), or creating specific campaigns with a different message for whoever abandoned the shopping cart versus whoever just visited the homepage without interacting with anything.
Segmenting remarketing by behaviour, not just "visited the website"
The most common mistake in remarketing is treating everyone who visited the website as a single homogeneous group. In reality, someone who reached the checkout page and abandoned right before confirming has a completely different purchase intent from someone who came in, looked at the homepage for five seconds and left. Segmenting audiences by depth of interaction (pages viewed, time on site, steps completed in a purchase process) lets you tailor the message and the offer to each level of intent, instead of showing the same generic ad to everyone equally.
Frequency and fatigue: remarketing can also be done wrong by overdoing it
Showing the same ad to the same person too many times in a short period produces the opposite of the intended effect: instead of building a positive memory of the brand, it creates rejection, and can even damage the overall perception of the business. Limiting impression frequency (a setting available on most ad platforms) and rotating the ad creative with some regularity helps avoid this ad fatigue effect.
The time window: how long it makes sense to keep following someone
It does not make the same sense to remarket to someone for 24 hours as for 540 days (the usual maximum allowed by platforms). The optimal duration depends on the typical decision cycle for the product or service: for a low-cost impulse purchase, a short window (7-14 days) is usually enough and more cost-effective; for a considered, high-value decision (a renovation, an investment, a medical treatment), a window of several months can still be relevant, because that person's real decision process also lasts that long.
Privacy and consent: the foundation all of this rests on
Google Signals and remarketing in general depend on the user having given consent for the use of cookies and data for advertising purposes, something privacy regulation explicitly requires in most cases. Without that consent, these features simply do not collect data on that person, which reduces (sometimes considerably, depending on the cookie banner's acceptance rate) the real size of the audiences available for remarketing.
An example of a well-segmented remarketing campaign
A custom furniture online shop had a single generic remarketing campaign targeting "everyone who visited the site in the last 30 days", with mediocre performance. By segmenting that audience into three groups based on depth of interaction (people who only saw the homepage, people who browsed product categories, and people who got as far as configuring a piece of furniture in the online simulator without completing the purchase), and creating a different ad and offer for each group (the third group, closest to buying, received a message directly reminding them of the piece of furniture they had configured, with a small limited-time discount), the cost per conversion for the most qualified group dropped to less than half compared to the previous generic campaign, while total remarketing spend barely changed.
Combining remarketing with smart exclusion
Besides deciding who to show remarketing to, it is equally important to decide who to actively exclude: someone who has just completed a purchase of a product that is not usually repeated in the short term (a custom piece of furniture, say) should not keep seeing generic acquisition ads for that same product, because it creates the impression of a business that does not know its own customers. Setting up those exclusions, just as important as setting up the remarketing audiences themselves, often falls outside the initial setup process and only gets reviewed once someone complains about seeing ads for something they have already bought.
The cost of not reviewing audiences regularly
A remarketing audience set up a year ago and never reviewed may have accumulated people with no real remaining connection to the business (they bought a while back and are not going to repeat, or they visited the site for a one-off reason no longer relevant), which dilutes the campaign's effectiveness without anyone noticing the exact cause. Reviewing and pruning audiences every few months, adjusting time windows to the business's real decision cycle, keeps remarketing focused on people who genuinely have a reasonable chance of converting.
Step by step: enabling Google Signals and creating the first remarketing audience
For a business that has not yet enabled this feature, the process is simple and free. First, go into the Google Analytics 4 property settings and enable "Google Signals data collection", accepting the corresponding terms. Second, wait at least 24-48 hours for the associated signals to start being generated. Third, in the Analytics audiences section, create a new audience based on a specific behaviour (for example, "visited a product page but did not purchase in the last 14 days"), instead of directly using the generic "all users" audience. Fourth, link that Analytics property to the corresponding Google Ads account, if not already linked. Fifth, in Google Ads, create a remarketing campaign using that audience imported from Analytics, with a message and offer specifically designed for that level of interaction. Sixth, let the campaign run for at least two or three weeks before evaluating results, because remarketing audiences need time to build up enough size. This entire process, start to finish, can be set up in a single afternoon of work, with no additional cost beyond the ad budget already being spent.
Frequently asked questions
Does enabling Google Signals slow down the site or affect privacy differently than normal Analytics?
It does not affect site speed, and in terms of privacy, it follows the same consent rules as any other Google advertising feature: it only collects data from users who have given explicit consent for personalised advertising.
How long does it take to generate a useful remarketing audience after enabling Google Signals?
It depends on the site's traffic volume, but it can take anywhere from a few days (on high-traffic sites) to several weeks (on lower-traffic sites) to build up the minimum audience size needed for ad platforms to use it effectively.
Do I need to be spending on advertising for it to be worth enabling Google Signals?
Although the main benefit shows up in paid campaigns, enabling Google Signals costs nothing and also brings reporting improvements (demographics, cross-device tracking) useful even without actively investing in remarketing, so it is worth enabling regardless.
Does remarketing only work on Google's display network, or also on search?
It works across several formats: display (visual ads on other websites), search (remarketing lists for search ads, which adjust bids or ads when someone from that audience searches on Google again) and YouTube, among others, each with its own configuration particulars.
What if a customer has already bought from me, should I keep showing them remarketing?
It depends on the goal: to avoid unnecessary spend on repeat acquisition, it is worth excluding already-converted customers from acquisition campaigns, but it can make sense to include them in specific loyalty or cross-sell campaigns with a different message than initial acquisition.
How do I know if my remarketing is generating results and not just spend?
By comparing the cost per conversion of remarketing campaigns against general acquisition campaigns; since these audiences already know the business, it is common (though not guaranteed) for remarketing's cost per conversion to be noticeably lower, which confirms it is adding value.