Bow City Storage
ROAS increase of 2,900% in addition to a 33% CTR increase.
GA4, server-side conversion tracking, consent that actually works, and closed deals fed back from your CRM. Attribution that makes every other channel measurable and defensible.
ROAS increase of 2,900% in addition to a 33% CTR increase.
Harmony Resorts reported a 6% conversion rate and an increase in ROAS of 74%.
A 41% increase in revenue along with an efficient 16% increase in ROAS.
From zero to $420K in online sales within 10 months.
Attribution built for the post-iOS 14.5 tracking gap. GA4, server-side conversion tracking, and closed-deal data from your CRM for accounts that have outgrown last-click reporting.
That means a written tracking plan, clean events, consent that changes what actually fires, closed deals fed back from your CRM, and reporting built around a decision rather than a metric count.
The audit usually finds something worth fixing immediately. After the rebuild, the new setup needs a period of clean data under its own definitions before anyone makes decisions on it. That waiting period is the part clients find hardest, and it is not skippable.
Six pieces of work, in the order they have to happen. Skip the first and the rest inherit its problems.
Almost every measurement problem we’re called about is a naming problem wearing a technical disguise. We write down what counts as a conversion, what each event is called, and which properties it carries, before touching a tag manager. Then we remove the duplicate tags most accounts have accumulated from three previous agencies.
Browser-based tracking now misses a meaningful share of activity, and the gap isn’t evenly distributed. Ad blockers and privacy defaults cut hardest into some audiences, which biases the data rather than just shrinking it. Server-side collection recovers part of that. It is not a loophole and we would not sell it as one.
A consent banner that doesn’t change what fires is a legal risk pretending to be compliance. We wire consent state through to the tags so declining actually stops collection. Canadian and European requirements differ. We build the mechanics properly; your lawyer defines the obligations.
If a human closes the sale, website conversions are a proxy and often a misleading one. We push closed-won outcomes back into the ad platforms so bidding optimizes toward revenue instead of form fills. Expect the campaign rankings to shift, because the cheapest leads are frequently the worst.
Every model is a simplification, and anyone selling certainty is overselling. Each one flatters a different part of the journey, so we treat them as lenses and settle disputes with blended acquisition cost.
A dashboard with forty metrics is a way of avoiding a decision. Each report answers a specific question for a specific person: what to change this week, and what the last quarter produced. If nobody opens it, it is not reporting, it is decoration.
Four stages. The first one usually produces the most valuable finding of the whole engagement.
How wrong is it now? We check what is firing, what is double-counted, whether consent is honored, and how far platform-reported numbers sit from your actual revenue.
Plan, then implement. A written tracking plan, consistent event naming, duplicate tags removed, server-side collection where it earns its cost, and consent wired through so it does something.
Close the loop. Offline conversions fed back from the CRM, reporting built around decisions rather than metrics, and platform numbers reconciled against blended acquisition cost.
Keep it honest. Tracking rots as sites change and platforms update. Quarterly we re-validate, and where volume allows we run holdout tests to check what the models are claiming.
This is the one service where the deliverable and the metrics are the same thing.
The gap between what the ad platforms claim and what you banked. The single most useful number we produce. Once it is quantified, every channel decision improves, and the gap is almost always larger than expected.
What proportion of known sales the analytics actually captured, reconciled against your own order or CRM records. If you’re only seeing seventy percent, every downstream calculation carries that error.
Total marketing spend divided by customers won. The figure that cannot be double-counted. When per-channel numbers disagree with each other, this is what settles it.
Which channels produce leads that actually become customers. This is where cheap leads get exposed, and it regularly inverts the ranking of channels compared to cost per lead.
What the spend added that would not have happened anyway. The only method that answers the question attribution models approximate. It costs some volume to run, which is the honest trade-off.
Does anyone open it, and did it change a decision? Unfashionable, and it decides whether any of this was worth paying for.
Foundational, if not exciting. The fair question is whether now is the moment for it.
The audit alone usually quantifies the gap between what the platforms claim and what you banked, which changes budget decisions immediately. Sequenced before paid search and Meta, it means the same spend optimizes toward customers instead of form fills.
Present in most accounts we inherit.
Newsletter signups counted alongside sales-qualified inquiries. Smart bidding then buys whichever is cheapest, which is exactly what you told it to do.
Two analytics installs and three conversion tags inherited over the years, inflating everything by an unknown amount. Nobody removed the old ones because nobody was sure what depended on them.
The banner appears, the tags fire regardless. It is a compliance exposure with the appearance of diligence, and it is remarkably common.
Each platform counts generously and uses its own attribution window. Add the reports together and they will claim more revenue than the business made.
Without offline conversion data, bidding optimizes toward whoever fills in forms most readily. Those are frequently your least valuable inquiries.
Forty metrics, no recommendation, no owner. It looks like rigor and it substitutes for deciding anything.
Your CRM holds the outcome. Measurement carries it back to the platforms. The platforms then buy customers instead of form fills.
Whether a deal closed and what it was worth. Without it, everything upstream optimizes toward form fills and calls the result revenue.
Consistent events, honored consent, and outcomes fed back to the platforms. This is the plumbing, and it is invisible until it is wrong.
Bidding algorithms optimize toward whatever you told them to value. Fix the definitions and the same budget buys better customers without any change in strategy.
We normally sequence this deliberately: measurement first, then Google Ads and Meta Ads managed against blended numbers, with CRM work where the outcome data is the weak link. Getting this order wrong is why so much media budget optimizes confidently toward the wrong thing.
Because they answer different questions with different rules. Each platform counts conversions it believes it influenced, using its own window, including view-through and modeled conversions. Your analytics is usually stricter and last-click biased. Neither is lying and neither is complete.
Events go to a server you control, which forwards them on, recovering some data lost to ad blockers and privacy defaults. It is worth it when media spend is large enough that a systematic undercount costs real money. It is not a way around consent, and it has genuine setup and hosting costs.
A banner that appears but doesn’t change which tags fire provides the appearance of compliance without the substance, and it is the most common setup we find. We implement the mechanics properly. For your specific obligations, take legal advice, not ours.
None exclusively. Last click undervalues everything early, first click undervalues everything late, and data-driven models need volume most accounts don’t have. We use them as lenses and cross-check against blended acquisition cost. Where volume allows, a holdout measures genuine incremental lift.
Blended acquisition cost over time alongside revenue is the version that survives a finance conversation, because it cannot be double-counted. Beyond that, the strongest evidence is a holdout: turn a channel off in a region and measure what changes.
For most businesses the platform matters less than the tracking plan behind it. The failure mode is treating a migration as a technical task and carrying over inconsistent naming, which is how you end up with a modern tool full of unusable data.
Sending closed-won outcomes from your CRM back into the ad platforms, so bidding optimizes toward customers instead of inquiries. It matters most where a sales team is involved and lead quality varies, and it routinely reverses which campaigns look best.
One to two weeks for the audit, and usually a month to have the rebuilt setup live. Then four to eight weeks of clean data before making decisions on it, because comparisons to history aren’t valid until the new definitions have their own period.
You do. Analytics properties, tag manager containers, server-side infrastructure, and documentation sit in your accounts, and the tracking plan is a document you keep. An agency that keeps your tracking configuration in its own accounts has made your entire reporting history contingent on the relationship continuing.
One more we get often: can you just build us a dashboard? We can, and we would rather ask what decision it is meant to support first. A dashboard built on unreliable events is a faster way to be confidently wrong, and most dashboard requests we receive are really requests for trustworthy data.
Attribution only matters when it is wired to the channels it measures. Here is what it pairs with.
Five businesses, in their own words, on working with Consultus Digital.
The audit takes one to two weeks and usually changes a budget decision the same week. Contact us today to book it and learn which of your numbers you can trust.