Enhanced conversions for leads and offline conversion import

Most lead gen accounts optimise for the wrong number. They count every form submission as a conversion and let Smart Bidding chase as many of those as it can — while Google has no idea which of those leads ever becomes a customer. Offline conversion import flips that around. It feeds the real value of a closed deal from your CRM back into Google Ads, so the bidding automation optimises for qualified leads that actually close rather than for form spam.

It sounds technical. Here’s why the default setup costs you money, how the import works, which stage should steer your bids, and why you may want to end up optimising for profit rather than revenue.

Storing the GCLID in your CRM: 4 steps

1. Read the GCLID on landing: With auto-tagging on, Google appends the click ID to the landing URL as the gclid parameter. Read it on the first page view and keep it only as long as your consent setup allows.
2. Send it with the form: Pass the GCLID as a hidden field so it stays attached to the lead after several page changes.
3. Save it in its own CRM text field: unchanged, together with the enquiry time and the lead ID. Status (qualified, won), time and value come later.
4. Report it back as an offline conversion: Import the stage the campaign should optimise for, then check that the rows were actually accepted.

1. Why optimising for form submissions costs you money

The default looks like this: someone fills in your form, that fires a conversion, and you set the campaign to Maximise conversions or a target CPA. What you’ve told Google is “get me as many form submissions as you can at this price”. And that’s exactly what the automation does — with full commitment.

The problem: Google only knows the click, not the close. To the bidding algorithm every form submission is worth the same. The serious prospect with a budget and the price-shopper pinging ten suppliers at once both count as “one conversion”. So does the bot filling your form with garbage.

In reality lead quality swings wildly — by search term, by ad, by time of day, by audience. A few patterns every sales team recognises:

• Leads from comparison and “free” searches enquire a lot and buy rarely.
• Certain keywords pull in students, job applicants and pure information-seekers, not buyers.
• Some audiences close at a high rate, others almost never.
• Form spam and bots inflate the conversion count without a cent of revenue.

When Google optimises for “form submitted”, the obvious thing happens: the automation scales exactly the sources that deliver the most submissions for the least money. A low lead price says nothing yet about the close rate. Your CPA in the interface falls, conversion counts climb, the report looks great — and nothing usable reaches sales.

The nasty part is the feedback loop. The more cheap submissions the automation collects, the lower the average CPA, and the harder it bids into precisely those sources. The error reinforces itself. You see a clean, falling curve in the account while your sales team spends more and more time phoning through hopeless enquiries. The metric everyone watches and the metric that carries your business drift apart completely. And since the form submission is the only truth Google has, the automation can’t behave any other way. It isn’t stupid, it’s just briefed wrong.

That’s the core of the misunderstanding. A form submission isn’t a conversion in any economic sense — it’s a statement of intent. Train the bidding automation on statements of intent and you get more statements of intent, dump a pile of unqualified enquiries on sales, and pour budget into the wrong search terms. Garbage in, garbage out, this time at lead level.

2. Offline conversion import: feeding the real deal value back to Google

The fix is to finally tell Google what happened after the click. That’s what offline conversion import (OCI) does. You report the qualified or closed deal as the success, not the form submission — with its real value attached. Smart Bidding then learns which clicks turn into customers and optimises for those.

The classic route runs on the GCLID, the Google Click ID. Four steps, simplified:

1. Capture the click. With auto-tagging switched on, the landing page URL can contain a GCLID. You read it out and write it into a hidden field on your form. On iOS and under certain consent setups the parameters wbraid and gbraid come into play instead of the GCLID — you need to check separately whether your specific integration supports them.
2. Store it in the CRM. On submission the GCLID (or wbraid/gbraid) lands as a field on the lead, next to name, contact and source.
3. Maintain quality. Your sales team does what it already does — disqualify the lead, send the quote, win the deal. All that matters is that the status and the deal value sit cleanly in the record.
4. Send it back. When a lead gets qualified or you win the deal, you upload the conversion with its GCLID, timestamp and value back to Google — by manual upload (CSV or a linked Google Sheet), through a supported API import, or automatically via a native CRM integration. For new API connections, check the current Data Manager API route; existing Google Ads API uploads may depend on transition rules.

From then on the bidding automation isn’t judging “submitted the form” but “became a customer and was worth X”. It shifts budget towards the campaigns, keywords and audiences that actually close, and pulls it out of the sources that only produce cheap submissions.

One question decides the quality of the whole setup: which stage steers the bids? The qualified lead can be available earlier and more often than a close. The close says more economically, but it may only arrive after a long sales cycle. Set the stages up as separate actions and decide deliberately which of them counts for optimisation in each campaign. If you add both up as equally weighted primary goals without a value model, you can end up rating one prospect as an acquisition more than once.

Check the conversion window and the current upload deadlines of your import method before you map a long sales cycle. A late business outcome can’t be attributed to an earlier ad click indefinitely. A CRM connector’s lookback is also not the same thing as the permitted age of a conversion. Google’s guide to Enhanced Conversions for Leads describes the different import routes and their requirements.

Enhanced Conversions for Leads supplements the attribution with user-provided data such as email address or phone number. The required normalisation, hashing, tag configuration and the CRM integration’s support all have to fit together. Hashing does not make personal data anonymous. An existing, lawfully collected click identifier stays useful; the extended method is no reason to remove a working GCLID handover without checking. Google describes the setup in its tag guide.

Both need a sound consent and data-processing chain. A successful upload replaces neither the check of the consent behaviour nor the check of actual lead quality.

3. Revenue is good, profit is better — and what you need for it

Once you send back the deal value instead of just the submission, you’ve already made the biggest jump. Google then optimises for revenue instead of counts — in most accounts that’s the difference between “lots of leads, little business” and “fewer leads, more revenue”.

But revenue isn’t the finish line, because revenue isn’t profit. If your margin varies by deal type, optimising for raw revenue can pull you in the wrong direction. An everyday example:

• A big job with a high invoice total but a thin margin, because it’s stuffed with bought-in work and materials.
• A smaller job with a fat contribution margin, because it’s almost pure service.

Optimise for revenue and Google chases the large, low-margin jobs — even though the small, high-margin deal is worth more to your books. So the more honest value to send back is the contribution margin or profit per deal, not gross revenue. You don’t need it accurate to the cent; a rough margin per deal type or product category is usually enough to steer the bidding in the right direction. I run the same logic in e-commerce — how to optimise for real contribution margin instead of revenue there is in E-commerce: track profit, not revenue.

You don’t need enterprise infrastructure to make this work, but a few things have to be in place:

• Attribution with consent. Carry supported click identifiers through to the CRM and, where appropriate, supplement them with lawfully collected user details for Enhanced Conversions for Leads. Check the prerequisites of your specific integration and reflect consent correctly.
• A CRM with value and status fields. One field per lead for status (qualified, won, lost) and one for value — ideally contribution margin, at minimum the deal value.
• A clear definition of a good lead. What counts as qualified, what counts as closed? Without that definition you upload inconsistent signals and the automation learns noise.
• Clean base tracking. OCI sits on top of your tracking foundation. If conversions double-fire, consent is missing or the GCLID never arrives, you’re importing junk with a value label on it.
• Some patience. The bidding automation needs volume and time to learn on the new signal. Think in weeks, not days — and leave the form conversion running alongside as a secondary signal rather than switching it off immediately.

The honest reality check first: if your base tracking isn’t solid, fix that before anything else. OCI amplifies what’s there, good and bad.

The bottom line

Raw form numbers don’t show whether an enquiry is worth anything economically. A low lead price can be good, but it has to be judged together with the qualification rate and the closes. Offline conversion import can connect that feedback from sales with the ad account.

Three things to take away:

• Pick a robust target stage. Send qualified or won leads back and decide, based on data volume and sales cycle, which stage steers the bids. Avoid counting the same prospect as an acquisition several times without a thought-through value model.
• Treat Enhanced Conversions for Leads as a supplement. User details can support the attribution. Keep existing lawfully collected click identifiers, and check tag, CRM, import method and consent together; that doesn’t guarantee a simpler or better route across the board.
• Optimise for profit where margin varies. Send back contribution margin rather than just revenue, or Google will chase big low-margin jobs.

The foundation stays the same: clean tracking with consent, and a CRM that knows value and status. Whether OCI improves campaign steering then depends on signal quality, data volume and how the goals are used in the campaigns. Nobody can take the definition of which lead counts as a close in your business off your hands.

If you want a second pair of eyes on whether your setup can carry an offline conversion import, write to me via the contact form. I’ll tell you honestly what I see.