Offline Conversion Tracking: When Your Website Tracking Is Fine but You Still Can’t Tell What’s Working

In part one, I looked at the issues I see most often inside the Google Ads account itself: nothing tracked, the wrong things tracked, every lead counted the same, and everything set to primary.

But even with everything set up just right, you might still be left asking: which of these leads actually turned into something valuable for the business?

Here’s why: (at the time of writing) Google Ads can typically track only what happens on your website or on Google’s platform, like ad phone calls or lead forms. So when someone fills in a form, books a call, or signs up for a free trial, Google Ads marks that as a conversion. But what happens next whether the lead was any good, if the sales team managed to get in touch, or if they actually became a customer happens outside of Google Ads’ view.

This is where things can get a bit murky, because there isn’t a one-size-fits-all way to join everything up. Every business has its own setup, different CRMs, phone systems, websites, and booking tools. But the core principle is the same for everyone: find a way to get what happens after the lead comes in back into Google Ads. That’s where offline conversion tracking comes in. It’s about bringing those leads and outcomes that happen away from your website back into your account.

Instead of a step-by-step setup guide, I’d like to walk you through the three points where things most often break down in real businesses. If any of these ring a bell, they’re definitely worth checking in your own setup.

1. The journey breaks where one system hands over to another

Phone calls

In part one, I talked about tracking phone calls. Here, the challenge goes a bit deeper. The real issue is when a lead comes in over the phone, away from your website, where your usual tracking can’t follow.

A good example is a garage door company I worked with many years ago. They wanted as many leads as possible and operated across multiple UK locations. There wasn’t a dedicated team answering the phones; the people taking calls had plenty of other duties too. Details were often taken down on a notepad, and with several people answering the phone, it was easy for something to be missed.

Even when callers were asked where they’d heard about the company, many would just say “Google”. Was that a Google ad, or did they find them organically? Most people don’t know the difference.

We could see the ads were running and the phones were ringing. What we didn’t have was a reliable way to grade and connect the two and feed that information back into the system.

There are two main options for closing that gap: Google’s own call tracking, or a third-party call tracking tool such as Mediahawk. Calls made directly from your ads, where someone taps the number in the ad, are tracked by Google without any code on your website. For calls made after someone lands on your site, both options work in a similar way: a bit of code on your website swaps your phone number for a dynamic tracking number, depending on where the visitor came from. So your main office number stays the same, but the number someone sees changes if they’ve clicked an ad.

Both options have their pros and cons. For example, Google’s own tracking doesn’t cost extra, but it only covers calls from Google Ads. If you’re also getting calls from leaflets, mailshots, email marketing, social media, or organic traffic, a tool like Mediahawk can help you track all those sources too.

At one point, the garage door company ran a limited-time offer: buy a garage door and get a matching front door in the same colour. They only promoted it through a leaflet drop, and by using a dedicated tracking number on the leaflet, they could see exactly how many calls it brought in. If phone calls are a big part of how your leads come in, having that kind of visibility is really valuable.

Multiple websites

Another common place where the journey breaks is when it moves between websites.

A lot of the time, people land on a marketing website or a specific landing page, which might sit on a subdomain, for example, a set of HubSpot landing pages. Then, when they click to book a discovery call, they’re taken to a booking portal on the main website, or to a tool like Calendly. In other setups, there’s a front-end marketing website, and the free trial sign-up happens on a separate domain where the actual software sits.

The main challenge is keeping track of the user as they move from one site to another. When the journey crosses into a different domain, you need cross-domain tracking to keep everything joined up. Third-party tools can make this trickier, since they don’t always let you add your own tracking.

When these aren’t joined up, you lose the information and, with it, the insight. Running Google Ads is a bit like testing a hypothesis. You put your best thinking into the keywords, the ads, and the targeting, but you don’t know whether it’s working until the data comes back. Without that data, it’s hard to tell whether the account is heading in the right direction or if changes are needed.

Recently, I worked with a geospatial data company whose journey ran across several websites and Calendly, and they had no conversion tracking set up. It took working with their web developer, Google Tag Manager, Calendly and a lot of testing to link everything up. Before, they couldn’t see anything. Now, as far as consent allows, they can see a joined-up journey from the ad through to the booking.

Consent

Consent is a big topic, and there’s a lot more to it than I can cover here. But one key thing to know: in the UK and EU, you shouldn’t enable your tracking cookies until someone accepts them (see analogy below). If someone declines, those cookies aren’t set, so you’ll always be missing some conversions in your reports.

Implementing Consent Mode helps you stay compliant, and there are two versions. The simplest way to picture it is as a door. Basic Consent Mode is like the door being fully shut: Google’s tags don’t load at all until someone says yes. Advanced Consent Mode is like the door being open just a crack: the tags load, and if someone declines, only limited signals without cookies are sent. This helps Google estimate the conversions it can’t see. Some legal advisers are more cautious about Advanced, since it sends some data even when someone declines. There are pros and cons to both, so it’s worth getting advice on which is best for your business.

Worth looking into

• Are calls being tracked, and from which sources?
• Does the journey move to a different website or tool at any point?
• If it’s a third-party tool, can you add your own tracking, or can it pass information back?
• Is Consent Mode set up, and which version?

2. Nothing comes back from the CRM

The website knows someone filled in a form. Only the CRM knows what happened next.

Take a software company I regularly work with that offers free trials. Google Ads drives free trial sign-ups for their software, and every time someone signs up, a conversion fires. That’s fine as a starting point. But we know a certain percentage of those sign-ups will be junk, a certain percentage won’t be the right fit, and only some will be the ideal customer.

Google Ads can’t tell the difference. To the account, every sign-up looks the same.

Junk leads

Junk is a good example of why this matters. With another software client, we were counting free trial sign-ups, but the sales team couldn’t get hold of a lot of them. When we looked more closely, there was a clear pattern: people signing up with a personal email address rather than a business one were much harder to reach and far less likely to go anywhere.

On the front end, every one of those counted as a conversion. Only the follow-up revealed the difference.

I worked with the web developer to make changes to the free trial form, including adding a question asking people which category they’d put themselves in. Only allowing business email addresses would have been another option, but that’s a business decision.

Feeding offline conversions back

This is where it steps up a level. Once a sign-up reaches the CRM, the sales team scores it, and it makes its way through the pipeline. Some will become marketing qualified leads, some won’t. By connecting the CRM to Google Ads, you can feed those qualified leads back into the account, so you’re effectively telling the system: these are the ones we want.

Traditionally, this has worked by capturing the Google Click ID (GCLID) when someone converts and passing it back with the date and time of the qualified lead. Google now recommends enhanced conversions for leads, which matches leads using hashed email addresses or phone numbers instead.

It can sound easy on the surface, and some of the guidance out there makes it sound as simple as one, two, three. In reality, it often isn’t. Some CRMs, like HubSpot, connect directly. Others, like Pipedrive, may need a tool such as Zapier in between. And some businesses don’t use a CRM at all and rely on Excel spreadsheets, which brings its own challenges with keeping information up to date.

There are a lot of moving parts. You’re relying on the sales team to keep records updated, and on everyone being on the ball. In some setups, there’s one person close to everything and it’s fairly simple. In a busy business, it can be really difficult. Getting buy-in can be hard too, because it’s a piece of work and the payoff isn’t always obvious at the start.

A practical note on access

In practice, the technical connection is often the easy part. The harder part can be access. To send CRM data into Google Ads, whoever sets it up usually needs more than view-only access to the CRM, and with some CRMs, including HubSpot, that can sometimes mean a paid seat or a higher plan. I’ve had client setups where connecting HubSpot for audiences worked fine, but sending conversions back needed a level of access that came with an extra cost.

That’s not a reason to avoid it, and it’s sensible for businesses to be careful about who can access customer data. But it’s worth checking early who owns the CRM, what access level is needed, and whether there’s a cost, so it can be agreed and budgeted for upfront rather than becoming a blocker halfway through.

The payoff

In part one, I talked about giving different types of leads different values. Say each free trial is worth £200. That’s useful, but when someone signs up, every trial looks the same, so they all get the same value. Quality only becomes visible later, in the CRM. Feeding CRM data back adds that quality: it separates the junk and the poor fit from the ideal customer.

Once that’s going back in, the system can start to learn which leads are the right ones to find more of. It can then bid and adjust to bring in more of the right people.

It’s not a magic bullet, though. Keywords, positioning and the offer still matter. I recently worked with a client where, after some deep work on who their ideal customer really was, the whole positioning needed to change, and that affected everything. Feeding data back in is one important part of the picture, not the whole of it.

The same data can also be used to build audiences. With another software company that had a churn problem, we used their CRM data to run a dedicated campaign for customers flagged as at risk of cancelling, then looked at whether those customers stayed on past the point they’d been expected to leave. Audiences are a topic in their own right, but it shows the data you bring back can work harder than just reporting.

Choices to make once it’s working

It’s worth thinking about what the account then optimises for. If the free trial and the qualified lead are both set to primary, the system counts both, which takes you straight back to the issue I covered in part one. Usually, the qualified lead becomes the main signal and the sign-up becomes secondary, or the values are adjusted to reflect the difference.

That switch is best made mindfully. Changing what the bidding optimises for, especially to a conversion that arrives some time later and in smaller numbers, can temporarily unsettle performance. I usually bring the qualified lead in as secondary first, check that the volume and reliability are in a good place and only then make it the primary signal. Think of it like steering a large ship: small, deliberate course corrections work better here.

There’s a trade-off here too. The further down the pipeline you go, the stronger the signal, but the fewer the conversions. A sales-qualified lead one the sales team has spoken to and agrees is a real opportunity tells the system more than a marketing-qualified lead, but there may not be enough of them for it to learn from. Finding the right balance is part of the work, and it’s something I’ll come back to in point 3.

Two things that catch people out

The first is how often the data goes back. Sending qualified leads back once a month is far less useful than daily or weekly, because the system learns more slowly.

The second is timing in your reports. Imported conversions are recorded against the date of the original click, not the date they’re sent back. So recent weeks will always look weaker until the data catches up. If you’re reporting to a manager or a board, it’s worth knowing this before the numbers raise questions. Google Ads also has a column that shows conversions by the date they actually happened (by conversion time), which can give a clearer picture when you’re reporting on a recent period.

Worth looking into

• Which CRM do you use, does it connect to Google Ads directly, and how often is data sent back?
• What stages does a lead go through, and does everyone agree on what they mean?
• Once qualified leads are coming back, what is the account optimising for?
• Who owns the CRM, and what access is needed to connect it?
• How are junk leads being spotted and filtered out?

3. Outcomes land too far in the future for the account to use

Even with everything connected, there’s one more problem: time.

Google Ads can only connect a conversion back to an ad click within a limited window. For offline conversions fed back from a CRM, that’s up to 90 days from the click, and sometimes less depending on the method. If a lead becomes a customer after that, Google Ads can’t link the sale back to the ad, so it can’t learn from it.

For some businesses, that’s not an issue. For others, it rules out the sale entirely as something the account can learn from.

I’m currently working with a business selling complex, high-value services to large organisations. They’re often dealing with large buying groups. The person who makes the first enquiry usually has to bring lots of other people on board, and they may not be the one who holds the budget. We estimate it takes well over a year from first enquiry to becoming a customer.

There’s also a volume problem. Even if timing weren’t an issue, a business that only needs one or two very large sales a year wouldn’t produce enough conversions for the system to learn from.

So what do you use instead? You look for earlier steps that tell you you’re reaching the right people. For this client, that means engagement actions: signing up to a personalised newsletter, joining an email track, or registering for an educational webinar. It’s about bringing the right people into the fold, and giving them what they need to build the case internally, because it’s not a decision they can make on their own.

I give these actions different values depending on how meaningful they are, rather than just relying on primary and secondary. I’ve also worked with clients to build what we call a quality rate. It doesn’t sit in Google Ads; it’s our own measure, worked out with the sales team, that checks whether the right people are coming through at each stage: are the job titles right, is the company size right? That applies to leads, but also to webinar registrations and newsletter sign-ups.

That’s why the quality rate matters. Once an engagement action carries a value, there’s a risk the system finds lots of cheap sign-ups from the wrong people. The quality rate stops us rewarding volume over fit.

It’s still an ongoing, iterative process, and there’s no set formula. What works for one business won’t necessarily work for another.

The long-term outcomes still matter. They just belong in your reporting, your budget decisions, and the bigger commercial picture, rather than in the bidding. That only works if your CRM records where each lead came from in the first place, usually through UTM parameters on your ad links. If it doesn’t, a sale that closes over a year later can’t be traced back to Google Ads at all, even manually.

Worth looking into

• How long does it typically take for a lead to become a customer?
• Which earlier step tells you most about whether a lead is a good one?
• Does that step happen well within 90 days of the click?
• Does your CRM record which campaign or source each lead came from?
• How are you checking the quality of the people coming through, not just the numbers?

Where to start with offline conversion tracking

In my experience, there’s no single setup, but the principle stays the same: get what happens after the lead back into Google Ads.

If you can’t connect everything at once, start with the bare minimum: getting qualified lead data fed back in, with all the consent bases covered. That depends on leads being marked as qualified in the first place, which depends on whoever is working on the CRM side. But if you can get that process in place, you can build up from there.

As for how long it takes, it’s hard to say, because every system is different. In my experience, it’s rarely as quick or as easy as it’s made out to be. There are little things that trip you up along the way. Permissions, budgets, and people who can’t yet see the value can all get in the way. Google changes things regularly, and consent needs keeping an eye on too.

So if you’re finding it difficult, don’t be hard on yourself. It is technical. But you can get there with some work, and it’s very much an iterative process. I’ve been on a journey with it myself.

You probably won’t get it perfect, and that’s fine. Work with what you’ve got, and think a little outside the box. If you can’t get the data back one way, is there another way? The quality rate came about because we couldn’t get what we needed directly, so we found another route.

If you’d like a second pair of eyes on how your leads are being tracked, a Google Ads Power Hour is a great way to get started. Or if you’d like someone to help join up your tracking, your CRM and your account, and manage it for you, feel free to book a discovery call.

A note on how this post is written

In the spirit of transparency, I use AI tools, including Claude, Grammarly, and others, to help turn my own thinking, experience, and dictated notes into finished writing. The ideas, opinions, and experiences in this post are mine. AI helps with structure and editing, turning what's in my head into something readable and, I hope, useful for thinking about your own Google Ads campaigns.