Campaign intelligence
Most email reporting answers "what happened" with numbers nobody can act on, and half of them are inflated: Apple Mail Privacy Protection has been preloading tracking pixels since iOS 15 in 2021, which means open rates overstate reality. Our reporting layer reads every campaign and flow monthly, scores what changed and why, and hands you a ranked list of next moves.
What gets measured?
Placed-order rate and revenue per recipient, per flow and per campaign, because those are the numbers a CFO recognizes. Click rate as the behavioural signal, benchmarked against Klaviyo's averages: 1.69% for campaigns, 5.58% for flows. Complaint and unsubscribe trends tracked against Google's 0.1% guidance, since a revenue win that spends reputation is a loss on a delay.
The AI layer earns its keep on breadth. It sweeps every segment-by-campaign cell for anomalies, fatigue curves, and timing effects, the kind of pattern that hides when a human reviews forty sends in an afternoon. An analyst verifies each flag before it reaches you. The month the sweep found 71% of sends going to 180-day-inactive segments, that finding became a suppression project inside a week; thepet-supplies case study tells that story from the client side.
How do tests stay honest?
Every A/B test declares its detectable effect size before launch. If the sample can only resolve a 1.5-point difference, we say so, and a 0.4-point result gets reported as inconclusive rather than dressed up as a win. This discipline is why test-driven changes at Sendnexa tend to hold up: the coffee-subscription client's campaign click rate moved from 1.2% to 1.8% across 8 weeks of verified iterations, and stayed there.
Intelligence feeds the other services. Findings route intoflow changes or segment updates, and pricing for the combined program is on the pricing page.
Why not report on open rates?
Because opens stopped being trustworthy in September 2021, when Apple Mail Privacy Protection began preloading tracking pixels and inflating open counts. Opens still have diagnostic uses, deliverability trend-spotting among them, but as a success metric they mislead. We report placed-order rate, revenue per recipient, and click rate, which still reflect real behaviour.
What does "AI-assisted analysis" mean in practice?
Models do the sweep; people do the verdict. Each month the full campaign and flow dataset gets scanned for segment-level anomalies, timing effects, and fatigue signals a human eyeballing 40 sends would miss. Every flagged finding is then checked by an analyst before it reaches your report. Nothing auto-generated ships unreviewed.
How do you call A/B tests honestly?
With sample-size discipline. A subject-line test on 800 recipients per arm cannot resolve a half-point click difference, so we pre-compute the detectable effect size before any test launches and report "inconclusive" when that is the truth. The result is fewer declared winners, but the ones we do declare keep performing the following month.
What arrives in the monthly report?
One page of numbers, one page of decisions. Placed-order rate and revenue per recipient by flow and campaign, complaint and unsubscribe trends against the 0.1% guidance, test results with confidence noted, and a ranked list of next actions. Reviewed live in a monthly working session, 45 minutes, your team and ours.
See what a month of analysis surfaces
Bring last quarter's campaign data to a working session and we will run the first sweep with you watching.
Book a working session