Know which clicks turned into customers.

Five things. An audit that documents every tag, conversion action and imported goal and quantifies how far the platforms sit from your own sales data. A rebuild that separates primary conversions from secondary ones, removes duplicates and sets counting and attribution deliberately. Values, so bidding knows what each conversion is worth. Consent mode implemented and verified in every state. And feedback loops — server-side delivery plus offline and CRM imports — so the platforms learn from qualified pipeline. You finish with a measurement layer you own, documented well enough for your own team to maintain.

A few symptoms are close to diagnostic. Platform conversions that exceed the number of orders or leads your business actually received. A conversion rate that looks impossibly high on one campaign. Recorded actions arriving with no value, or all with the same value. A sudden step change in the data on the day of a site release. Sales telling you the leads are poor while the account reports a great cost per acquisition. Any of those is worth an audit, and the audit is cheap relative to the spend it protects — which is why we quote it as a fixed piece of work.

Yes, and Google publishes the number rather than leaving it to agencies to claim: advertisers bidding to conversion value who implement enhanced conversions see an average 8% incremental ROAS on Search, based on 99 conversion lift studies run globally between April 2024 and April 2025. The mechanism is simple — hashed first-party data from your site lets Google recover conversions that would otherwise be unobservable, so bidding sees more of what really happened. The catch is that it cannot rescue a broken setup: if your conversion actions are wrong, enhanced conversions makes the wrong data more complete. Fix the layer first.

More than most teams realise, in both directions. Google's own worked example shows an advertiser with a 50% consent rate seeing a 19% drop in observed sales, because unconsented users convert less often. Implemented properly, consent mode lets Google model that gap instead of simply losing it — Google reports Air France experiencing a 9% conversion uplift after rolling Consent Mode out across European markets. What matters is that the consent platform genuinely communicates with the tags and that every state is tested. We check granted, denied and partial rather than trusting a vendor's integration badge.

Yes, and for service businesses it is usually where the real conversions are. Calls are tracked with dynamic numbers so the source is known, counted only over a duration that indicates a genuine enquiry, and passed back as results with values. Offline sales come back through imports from your CRM: the qualified lead, the booked appointment, the signed job, each with a value. That is the difference between an account optimising for cheap enquiries and one optimising for revenue. We automate the feed so it does not depend on anyone remembering to upload a spreadsheet.

Because they answer different questions, and none of them is lying. Each ad platform reports conversions it believes it influenced, using its own attribution window and modelling, so adding platforms together always overstates — two of them will claim the same sale. GA4 attributes on a different model again and counts sessions rather than credited conversions. Your CRM knows what actually closed. The workable answer is to use each for its purpose, reconcile them on a schedule so the gaps are known quantities, and make budget decisions on blended efficiency plus your own sales data.

A little of their time, rarely much. Most of the work happens in tag management, the ad platforms and your CRM, which we can do with the right access. We need developer involvement when a data layer needs adding, when a checkout must expose order values or identifiers, or when a release has to be deployed — and we write the spec, test it in staging and stay on the call for the release. Teams with no developer availability are common and not a blocker; we simply sequence the work so the parts that need code come last.

Yes, but for a different job than the ad platforms. GA4 is where you understand behaviour: which pages and paths lead to conversions, where people drop out, how segments differ. It is not the right source for bidding, and importing GA4 goals into Google Ads alongside native conversion actions is one of the most common double-counting mistakes we find. So we configure GA4 properly, define its key events to match the same dictionary as everything else, and keep the bidding signal native. Our analytics team takes it further where you need warehouse-level reporting.

Monitoring, plus a habit. We alert on volume drops, tag errors and campaigns spending without results, so a break surfaces in hours rather than at month end. Every significant site release triggers a QA pass, because a template change is the most common cause of a silent failure. The setup is documented and conventional, so your own team can diagnose it too. And because we reconcile monthly against your orders or CRM, drift shows up as a number rather than as a surprise in a quarterly review.

Very often that is exactly what we are asked to do, and we approach it without drama. We document what exists, keep what is working, and replace only what the audit shows is wrong — usually duplicate conversion actions, missing values, secondary events treated as primary, and consent wired incorrectly. Then we reconcile against your own sales data so the improvement is measurable rather than a matter of opinion. If you want the same team to run the media afterwards, our paid search and paid social teams work from the same definitions; if not, we hand over documentation your current agency can use.

We price the audit and the rebuild as fixed-scope projects, so you approve a number rather than an open-ended retainer, and the audit comes first so the quote reflects what your setup actually needs. The scale is driven by how many platforms you buy, whether a server-side container is needed, how much developer work the data layer requires, and whether CRM feedback is in scope. Ongoing monitoring is an optional monthly fee covering alerting, the QA pass and monthly reconciliation. We do not charge for measurement as a percentage of media spend — the two are unrelated, and tying them creates the wrong incentive.

Google Ads and GA4, Google Tag Manager including server-side containers, Meta's pixel and Conversions API, TikTok Events API, LinkedIn, Microsoft Advertising, Pinterest and Snapchat. On the data side: HubSpot, Salesforce, Pipedrive, GoHighLevel and most common CRMs, call tracking platforms, Shopify, WooCommerce, Webflow and custom stacks. We also work with whichever consent management platform you already run rather than insisting on a switch. If you use something unusual, the pattern is the same: define events once, deliver them server-side where it matters, verify against your own data.