How to measure effectiveness

The wrong question to ask of a beauty DOOH campaign is “what was the ROAS?” — that demands a last click from a channel built to move the upper funnel, and you’ll conclude it failed when you simply measured the wrong thing. The right question is “did it work?” — and that breaks into three layers an advertiser can actually measure: was the ad seen (attention), did it stick (recall and brand lift), and did it drive action (attribution). The discipline that separates a credible read from a flattering one is matching each layer to what you can genuinely prove at your scale — and being honest that no verified salon-specific attention, recall, lift or CPM number exists, so every effectiveness benchmark you cite is general OOH evidence, not a beauty-venue fact. This guide is how to measure each layer, what’s deterministic versus what needs a pilot, and why it all has to be set up before launch. (For the agency-side mechanics of the wrap report, see measuring & reporting to clients; this guide is the brand’s outcome view.)

1. The three layers of effectiveness

Stop treating “effectiveness” as one number and it becomes measurable. It stacks in three layers, and the IAB’s DOOH Measurement Guide (July 2025) standardizes the definitions across all of them — impressions and attention, then brand lift, then attribution via matched-market and synthetic-control testing (IAB — primary):

• Attention — was the ad actually seen? Viewability-adjusted exposure, dwell, eyes-on.
• Recall & brand metrics — did it move awareness, consideration, favourability or purchase intent? Measured exposed-vs-control.
• Attribution / outcomes — did it drive a measurable action: a search, a visit, a code redemption, a sale?

Pick the layer that matches your campaign objective (set when you planned it) and measure that — not all three badly, and not a last-click metric the channel can’t produce.

2. Attention — was it seen?

Attention is the newest layer and the one most prone to over-reading. The honest state:

• Vendor metrics are models, not currency. Lumen counts an ad “viewed” on a fixation over 100 ms; Adelaide’s “AU” is a machine-learning media-quality score. Both are real tools, but their lift claims are vendor case-aggregates — flag any attention number as a model, not a measured truth (Lumen, Adelaide — directional).
• The general OOH attention case is directionally strong but not in-venue. Per-exposure attention runs ~1–2 seconds with OOH often highest of measured media, and as little as ~1.5 seconds can encode memory with strong brand assets (Lumen/JCDecaux, Amplified Intelligence — directional; measured off-channel, not in salons).
• Standards now exist to hold vendors to. The IAB/MRC published Attention Measurement Guidelines (Nov 2025) — ask any attention vendor to map to them (IAB/MRC — primary).

The takeaway: you can report attention, but present it as a vendor model and never as a salon-specific eyes-on figure, because none has been measured.

3. Recall and brand lift — did it stick?

This is the heart of an upper-funnel campaign, and the method is well-defined: a brand-lift study surveys an exposed audience against a verified unexposed control and attributes the difference in awareness/recall/favourability/purchase-intent to the exposure (method — primary). Two honest constraints:

• Lift studies need scale and a clean control. With a handful of salon screens you rarely reach the exposed/control sample sizes needed for significance — so treat brand lift as a pilot to design carefully, not a deliverable you promise on every flight (see §6).
• Don’t import benchmark ranges as fact. “Typical lift of 5–15% awareness” circulates widely but we could find no primary source publishing it as a benchmark — use any range as illustrative, uncited (unverified — do not assert). Real datapoints exist as single vendor cases (e.g. a programmatic DOOH campaign reporting +19% awareness / +6% consideration — but that’s a specific beverage campaign in one market, not a beauty or category norm) (vendor case — directional).

And one figure to actively avoid: the widely-circulated “47% OOH vs 35% digital” recall stat does not appear in the primary OAAA/Solomon Partners release it’s attributed to — don’t cite it (refuted). What you can cite is the directional finding that OOH tends to produce high ad recall versus other channels, framed as a meta-analysis of an OOH-favourable body of studies (Solomon Partners/OAAA — directional).

4. Attribution — did it drive action?

Attribution is where a niche in-venue campaign can actually shine, if you use the methods that work at small scale. They run from deterministic to study-grade (IAB, method — primary):

• Deterministic at any scale (your backbone): QR codes, unique URLs, promo/voucher codes, unique search keywords tied only to the DOOH flight. These need no panel and no control group — make them the spine of measurement for a small beauty buy.
• Search/online lift: the strongest verified evidence that OOH drives action online. OOH accounted for ~26% of offline-media search activations on only ~7% of ad spend (≈4× its weight), with ~46% of adults searching after seeing OOH, ~38% going to Facebook and ~25% to Instagram (Nielsen/OAAA Online Activation, March 2017, n=1,089 — primary; survey-recall, dated, offline-media denominator). A separate OOH study found ~90% noticed OOH in the past month and ~66% of smartphone users took an on-device action afterward (Nielsen/OAAA 2019 — primary).
• Footfall lift: exposed devices’ subsequent visits vs a matched unexposed control — a real method, but one that needs scale and a clean control, so it’s a pilot, not a default (method — primary).
• The multi-touch caveat: never 100%-attribute a sale to one touchpoint; consumers are touched many times, so the IAB pushes incrementality over last click (IAB — primary principle).

5. ROI — and its honest limits

You can place beauty DOOH in an ROI frame, but only as category evidence. OOH has been estimated to return around $5.97 in sales per ad dollar on average, and adding OOH to a media mix lifts overall returns (Benchmarketing/OAAA, Analytic Partners — directional; dated, US-average and Australian/vendor-commissioned respectively, not beauty-specific). These justify the channel; they do not predict your flight’s return. The deterministic methods in §4 are what tell you whether your campaign actually paid — the ROI studies tell you the medium can.

6. Set up measurement before launch

Measurement you design after the campaign is measurement you can’t trust. The pre-launch checklist (IAB, method — primary):

1. One primary KPI per campaign, tied to a single funnel layer (attention / brand lift / footfall / code redemptions / sales).
2. Establish the control group up front — matched unexposed devices or markets. It can’t be reconstructed afterward.
3. Issue unique codes / vanity URLs / QR per placement before the flight.
4. Run brand-lift one campaign at a time — overlapping flights contaminate the control.
5. Pre-register the expected effect size and the sample you’ll need, so a null result is interpretable rather than ambiguous.

7. What’s credible versus over-claimed for a niche beauty campaign

Draw the line explicitly:

• Credible at any scale: QR/unique-URL/promo-code/keyword tracking, on-site conversion counts — deterministic, defensible, small-scale-friendly.
• Pilot, not guarantee: footfall lift, brand-lift surveys, sales-lift test/control — they need scale and a clean control to be valid; design them as experiments.
• Model, not currency: any attention metric — report it as a vendor model.
• Category evidence, not your result: every OOH/DOOH effectiveness number — frame it as what the channel does, never as what your flight delivered.

The IAB 2025 guide is the “what good looks like” anchor: accredited methodologies, third-party verification, incrementality over last click (IAB — primary).

8. The measurement mistakes that mislead

• Demanding last-click ROI from a brand channel — a category error (§1, §4).
• Importing a precise salon attention/recall/lift number as fact — none exists (§2, §3).
• Citing “47% vs 35%” recall or “5–15% lift” as sourced benchmarks — they aren’t (§3).
• Single-touch attribution — never 100%-credit the screen (§4).
• Comparing vendor numbers that vary 20–30% by methodology as if interchangeable.
• Designing measurement after launch — too late for a control group (§6).

So — how do you measure beauty DOOH effectiveness?

By measuring the right layer the right way. Decide whether you’re proving attention, recall or action, and match the method to your scale: lean on deterministic QR/URL/promo tracking and search lift as the backbone a niche campaign can genuinely stand behind, and run footfall and brand-lift studies as pre-designed pilots, not promises. Cite OOH’s effectiveness as category evidence, treat attention metrics as models, and never present a borrowed number as a salon result — because there isn’t one. Set it all up before launch, with one KPI and a real control group. Do that, and you’ll know whether the campaign worked on the terms the channel can actually deliver — which is the only honest way to judge it.