Careers
Part of Reading campaign measurement with a sceptical eye
How to assess campaign measurement benchmarks
Assess campaign measurement benchmarks by separating delivery, attention, response, and incrementality, then weighing local economics and uncertainty.
What to take away
- A benchmark must match the metric and measurement method before it informs a target.
- Attention, delivery, outcomes, and incrementality belong to different layers.
- Local economics and uncertainty should govern action, even when a peer number looks attractive.
- Set three thresholds before reading an outside numberan operational guardrail, an economic break-even point, and an evidence threshold.
- An external percentile supplies context only after those three are set.
Campaign measurement benchmarks can expose an unusual result, but they are not plug-in targets. A number loses meaning when its population, medium, design, period, or business model differs from the campaign being judged.
In US paid social prospecting over a four-week flight, typical click-through rates run about 0.1 percent to 1 percent. Typical display viewability runs about 50 to 70 percent. Placement, format, and audience set where a campaign lands. Break-even economics, not the peer number, decides whether the result was good enough. Without a named publisher, collection period, and sample, treat these ranges as orientation only, not decision benchmarks.
For a hypothetical four-week US retail test, the guardrail is a minimum viewable completion rate. The break-even point is incremental contribution per media dollar, set by margin and repeat rate. The evidence threshold is the smallest lift the planned sample can detect. A peer median enters only after all three are fixed.
Comparable sources, by layer:
- DeliveryNielsen and Comscore reach and frequency data, plus platform delivery reports from Google Ads and Meta Ads Manager.
- AttentionDoubleVerify and Integral Ad Science attention and viewability reports.
- Responsethe WordStream (LocaliQ) Google Ads benchmarks and the HubSpot marketing benchmarks.
- Market contexteMarketer and Insider Intelligence aggregates and US Bureau of Labor Statistics price and wage series.
Compare the benchmark layer
Possible reference
- Delivery
- Valid reach, frequency, completion, or viewability
- Attention
- Qualified attention measure under a disclosed method
- Response
- Click, visit, inquiry, or sale rate
- Effect
- Experimental absolute or relative lift
- Economics
- Incremental contribution or payback
Do not infer
- Delivery
- Customer attention or effect
- Attention
- Business outcome
- Response
- Incrementality
- Effect
- Universal repeatability
- Economics
- Suitability under different margins
IAB and MRC's attention measurement guidelines address definitions, methods, reporting, and limitations for attention measurement. The document helps audit comparability; it does not make attention a substitute for a campaign's business outcome. The same layer discipline applied to standing metrics sits in marketing KPIs benchmarks.
Benchmark Layers and Limits
Layer
- Delivery
- Reach, frequency, viewability
- Attention
- Qualified attention measure
- Response
- Click, visit, inquiry, sale
- Effect
- Experimental lift
- Economics
- Incremental contribution, payback
Possible reference
- Delivery
- Attention or effect
- Attention
- Business outcome
- Response
- Incrementality
- Effect
- Universal repeatability
- Economics
- Suitability across margins
Do not infer
- Delivery
- Attention
- Response
- Effect
- Economics
Treat incrementality as a design-specific result
IAB and IAB Europe's commerce media incrementality guidelines address incremental measurement in a defined commerce-media setting. A result still depends on assignment, outcome, window, identity, interference, cost, and uncertainty in the actual study.
Incrementality Result Checks
- Match numerator, denominator, outcome, maturity
- Check audience eligibility and market selection
- Compare channel and campaign objective
- Inspect assignment, exclusions, contamination, power
- Use an interval, not only a point estimate
- Rebuild value with local prices and costs
If the study is too imprecise to distinguish the break-even effect, more reporting decimals will not solve the decision.
Keep a dated benchmark register. Record the publisher, exact page, collection period, sample, market, metric contract, method, adjustments, and permitted use. The register and the three thresholds both appear on the campaign measurement checklist. When a material field is missing, lower the weight of the comparison and never let it trigger a budget change on its own.
Rebuild the number from its source
Recalculate the figure from the cited source or governed dataset, then record any mismatch. Keep market context separate from an internal target. A benchmark becomes useful only after the local team explains why the comparison group and method fit the decision.
For campaign measurement benchmarks, the GAO evaluation design guide explains how evaluation questions, evidence needs, and design choices fit together. The guide is written for federal program evaluation. Use its design discipline as a check on the method, not as proof that a marketing result is causal or transferable.
The NIST experimental design selection guidance starts design choice with the objective and practical constraints. That supports a clear boundary between descriptive benchmark reporting and a controlled effect estimate. It does not supply an advertising benchmark or make an observational comparison causal.
Reading a peer number as a target is one of the campaign measurement mistakes that survives every reporting redesign.
Review cadence and early refresh
Revisit a benchmark annually and re-check fast-moving channels every quarter. Refresh early when a platform changes reporting, when privacy rules shift, when a market such as Canada is added, or when media prices move sharply. A trigger is two consecutive periods outside the evidence threshold.
Set the next review date and name the change that would trigger an earlier check.
Common questions
What is a good campaign lift benchmark?
There is no universal good lift. Judge the estimate against its design, uncertainty, local break-even effect, capacity, and strategic alternative.
Can attention scores predict sales?
They may be tested as leading signals, but the relationship must be validated in the intended campaign and population.
Should benchmarks change in 2027?
Review them when methods, platforms, privacy rules, markets, prices, or customer behavior change. Preserve the source date and prior rationale.
