The scenario
This playbook applies to a B2B SaaS company, say one selling workflow software to finance teams, that has grown through a product-led free trial and paid acquisition spread across Google Ads, LinkedIn, Meta and a handful of review sites. Each channel is run by a different person or vendor, each reports in its own platform, and added together those reports claim more signups than the product database records.
A typical trigger is a board question the company cannot answer: how long does it take to earn back the cost of a paying customer, and which channels are driving it? Axoria is engaged to run the performance marketing programme with that question as the brief.
The challenge
Three problems compound each other. Platform attribution double-counts conversions, so no channel’s cost per trial can be trusted. Trials are treated as equal even though a large share never activate, so the metric being optimised is disconnected from revenue. And there is no mechanism for deciding between channels: budgets are set annually and defended by whoever runs them. The goal is not to cut spend but to know which spend produces paying customers that would not have arrived anyway.
Research and analysis
We begin by rebuilding the measurement layer. Trial signups are joined to activation events and billing data in the warehouse, so every trial can be classified as qualified (reached the activation milestone) or not, and every paying customer traced to its first and last touch. GA4 is reconfigured with server-side tagging so the same event definitions feed every platform.
We then compare platform-reported conversions against warehouse-verified ones. The pattern in this scenario is consistent: paid social over-claims through view-through attribution, branded paid search captures demand that organic and direct would probably catch anyway, and review-site placements are under-credited because their traffic often converts days later through a direct visit.
Strategy
We propose a coordinated system rather than a set of channel plans, organised around three commitments.
The trade-off is that incrementality tests deliberately withhold spend from some regions for a period, which costs short-term volume. We make this explicit in the plan, because the alternative is continuing to allocate a large budget on numbers nobody trusts.
Execution
The first quarter goes on tracking, the warehouse join and the first two geo holdouts. Branded search is tested first because it is simplest to isolate; paid social follows, with matched market groups selected on historical signup patterns. While those tests run, non-brand search is restructured around qualified-trial imports so that Smart Bidding learns from the right outcome.
In parallel, the CRO track focuses on the gap between signup and activation. Session recordings usually show unqualified trials abandoning at the data-connection step, so the onboarding flow is tested with a guided starter dataset. Raising activation rate lowers CAC for every channel at once.
Optimization
The holdout results reshape the mix. Where branded search proves largely non-incremental, its spend is reduced to a defensive floor. Paid social usually proves more incremental than the raw view-through numbers suggest but less than the platform claims, so it is kept with a lower budget and a creative programme aimed at cold audiences rather than retargeting. Review-site placements and non-brand search absorb the freed budget. Reallocation happens in monthly steps with warehouse-reported CAC payback as the check on each move.
How results are measured
Every reported figure comes from the warehouse, not from ad platforms, and is compared against the prior year on a like-for-like budget basis. Platform-reported cost per trial is retired from the reporting pack once the warehouse model is live, so there is only one version of the truth in every review.
- CAC payback: months to recover blended acquisition cost from gross margin, calculated per cohort of paying customers. A programme on track shows payback shortening as spend moves away from non-incremental channels and activation rate rises.
- Qualified trials: trials that reach the activation milestone, and their share of all trials. Success means qualified trials growing on a flat total budget while the qualified share improves through onboarding tests.
- Blended efficiency: total paid spend divided by warehouse-verified paying customers. In a working programme this falls steadily, and the direction holds after each monthly reallocation rather than reversing.
What we have learned
- Measurement is the strategy. Until every channel reports against the same warehouse-verified outcome, budget allocation is negotiation, not optimisation. Our analytics and tracking work is usually the first step for a reason.
- Test incrementality where it is cheapest to be wrong. Starting with branded search builds confidence in the method before the harder paid-social test.
- Activation is an acquisition lever. Improving what happens after signup lowers CAC across every channel simultaneously, a pattern common to product-led SaaS.
- Move budget in steps. Monthly reallocation with a payback check avoids over-correcting on a single test result.