.webp)
OnlineNot sure which channel fits your business?View all services →
Three inputs. Instant KPIs. No spreadsheets.Open a calculator →
OnlineAI bidding optimises to whatever signals you feed it. Here's the 2026 tracking stack - GA4 key events, Enhanced Conversions, offline import, Consent Mode v2 - done right.
Here is an uncomfortable truth about modern paid media: you are no longer really optimising your campaigns. The machine is. Smart Bidding, AI Max and Performance Max all run on the same basic logic — they pull whatever levers they need (bids, placements, audiences, creative combinations) to produce more of the conversions you have told them to value. Which means the single most important lever you still control is not a bid strategy or a budget. It is the quality of the conversion data you feed the algorithm.
Get that data clean, complete and revenue-weighted, and the AI works in your favour. Get it wrong, and it will confidently and efficiently spend your budget chasing the wrong outcomes. "Garbage in, garbage out" used to be a tidy proverb. In 2026 it is the difference between accounts that compound and accounts that quietly bleed. This is where the real competitive advantage now sits — and most accounts are losing performance to poor data long before tactics ever come into it.
When bidding was manual, a messy tracking setup was an inconvenience. You could eyeball the numbers, apply judgement and correct for what you knew was off. AI bidding removes that human buffer. The algorithm cannot tell the difference between a lead worth £50 and a lead worth £50,000 unless you tell it. It cannot know a form-fill turned into a closed deal three weeks later. It only sees the signals in the account, and it treats every one of them as gospel.
So two agencies can run near-identical strategies on the same budget and get very different results — not because one wrote cleverer scripts, but because one is feeding the machine real, revenue-shaped signals and the other is feeding it raw form-fills. The tactics have largely been commoditised by automation. The data has not. That is why, for B2B advertisers especially, first-party conversion data has become the genuine moat: it is hard to copy, specific to your business, and it is what the AI is actually competing on.
A robust 2026 setup is not one tool. It is a stack, where each layer feeds the next and closes a gap the previous layer left open. Here is how the pieces fit together.
Key events (GA4's term for what used to be called conversions) are the foundation. The common failure is treating every event as equal and firing them off a thank-you page URL alone. Do the boring work instead: define a small number of meaningful key events, make sure each one fires once per genuine action, and — crucially for lead gen — pass a currency and value parameter with your generate_lead event so the signal carries worth, not just a count. GA4's data-driven attribution and 30/90-day lookback windows only distribute credit sensibly when the underlying events are clean and consistently defined. If you are hazy on any of these terms, the marketing glossary has plain-English definitions of key events, enhanced measurement and data-driven attribution.
One 2026 wrinkle worth flagging: Google has continued to tighten how GA4 counts events and hand data to Google Ads, including recalibrating attribution and adjusting default lookback windows. The visible conversion count can look unchanged while match quality quietly degrades underneath. Audit periodically rather than assuming last year's setup still holds.
Browser-based tracking loses conversions to cookie restrictions, cross-device journeys and privacy controls. Enhanced Conversions plug much of that gap. When someone converts, your tag captures first-party data they have already given you — typically a hashed email address or phone number — and sends it securely to Google to match the conversion back to an ad click that the cookie alone would have missed. It is one of the highest-return, lowest-effort improvements you can make, and it directly improves the signal Smart Bidding learns from. Turn it on, and verify it is actually matching rather than just switched on.
This is the layer that separates serious B2B accounts from the pack, and it is where most performance is left on the table.
In a typical B2B funnel, the on-site "conversion" is a form-fill. But not all form-fills are equal: some are tyre-kickers, some are students, some are your competitors — and a small number are six-figure deals. If you optimise to raw form-fills, the AI will cheerfully find you more of the cheap, plentiful, low-quality ones, because that is literally what you asked for. It has no idea which leads your sales team actually closed.
Offline conversion import fixes this by closing the loop back to the CRM. When a lead is captured, you store the Google click ID (GCLID) — or the corresponding click identifier — against that record. As the lead moves through your pipeline, your CRM knows what happened: became an MQL, turned into qualified pipeline, closed at £48,000. You then import those outcomes back into Google Ads against the original click, with their real values. Now the algorithm optimises toward qualified pipeline and closed-won revenue instead of raw enquiries.
The mechanics in 2026 have consolidated around Enhanced Conversions for Leads, which upgrades the older GCLID-only import by also matching on hashed first-party data (email, phone) captured from your lead forms. Uploads increasingly run through Google Ads Data Manager rather than the legacy API path — Google has been migrating these functions to its Data Manager API through mid-2026 — so check which route your setup uses before you build anything. Whichever mechanism you choose, the principle is the same and it is the highest-leverage change most B2B accounts can make:
Stop optimising to the lead. Start optimising to what the lead became.
To value those outcomes correctly you need to know your real economics — true customer lifetime value, blended close rates and target CPA. If those numbers are fuzzy, the calculators are a quick way to pin down the LTV and CPA figures you will feed back into bidding.
None of the above works if consent breaks your measurement. For advertisers reaching UK and EEA users, Consent Mode v2 is effectively non-negotiable for using Google's advertising and remarketing features. It passes the user's consent choices to Google's tags. When a user declines cookies, properly configured Consent Mode allows Google to model the conversions it can no longer directly observe, using aggregated, privacy-safe patterns — so a consent gap becomes a modelled estimate rather than a silent hole in your data. Implemented badly, you either lose conversions wholesale or record them without a legal basis. Get the implementation right and get it reviewed.
Server-side GTM moves tag firing from the user's browser to a server you control. The practical benefits: better data durability against browser restrictions and ad blockers, more control over exactly what data leaves your environment, improved page performance, and a single clean pipeline for enriching and distributing your first-party data to platforms. It is not mandatory for every account, but for data-serious B2B advertisers it is fast becoming the backbone that makes Enhanced Conversions, consent handling and offline imports reliable rather than fragile. Underpinning all of it is first-party data hygiene: consistent identifiers, clean CRM records and a durable way to tie a click to a customer to a pound of revenue.
Most under-performing accounts are not broken in dramatic ways. They leak through a handful of predictable faults:
Run this against any account before you touch a single bid strategy. If you cannot tick every box, fix the data first — it will do more than any tactical change.
AI bidding has quietly changed the job. The advertisers winning in 2026 are rarely the ones with the cleverest tactics — they are the ones feeding the machine the cleanest, most complete, most revenue-shaped conversion data. Everyone has access to the same automation. Not everyone has the data discipline to make it work in their favour. That gap is your opportunity.
If you would like a clear-eyed view of where your conversion data is leaking and what it is costing you, a free growth plan is a sensible place to start. We will map your tracking stack against the checklist above and show you what to fix first.
Luke leads strategy across every Behaviour Digital account — building data-driven advertising systems for ambitious brands. When he's not auditing ad accounts, he's writing about what he found in them.
Get a free, tailored growth plan — we'll run our framework on your accounts and show you exactly what we'd do.
Get your free growth plan →