Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Frameworks › The Measurement Plan Framework

Measurement Plan for Awareness

A measurement plan for awareness: reach and frequency definitions, brand-lift studies, branded search and holdouts in place of click-based conversions.
Objective guideintermediate7 min readUpdated

Why the conversion-led plan misreads awareness

The generic plan works downwards until each goal is proved by something a tracking tool records per user. Awareness goals resist that. The change the campaign is buying, a person now recognising the brand or recalling it when the category comes up, happens in memory and shows up in behaviour weeks or months later, mixed with everything else that reaches that person. Per-user tracking also thins out at exactly the point awareness media works: on connected TV, in apps, across devices and in walled platforms that do not share identities.

So the plan has to accept two kinds of evidence. Delivery evidence says who was reached, how often and in what conditions. Effect evidence says whether reached people think differently from comparable people who were not. Neither comes from the analytics property, and both have to be arranged before the money is spent. The brand awareness and lift guide explains aided, unaided and top-of-mind awareness; the plan picks which of them the objective is judged on.

Decisions that change for awareness

Plan componentIn the generic planIn an awareness planDecision it forces
KPIA conversion event per goalReach at an effective frequency, and lift in a survey measureWhich survey question is the headline, and what frequency counts
Tracking specificationEvents and parametersDelivery logs, verification data, survey sample, search trendsWhich vendor supplies each source, and when it is pulled
Source of truthAnalytics propertyPlatform delivery, a verification vendor and a study providerHow reach is deduplicated, and what is admitted as unknown
Baseline and targetLast period's KPIA pre-flight wave and an unexposed controlWhether a control can be built at all, and how
Lagging signalsRarely separatedBranded search, direct visits and share of searchHow long after the flight they are read
Data qualityTag QAViewability and attention thresholds per buyWhat an impression must meet to count

Choose KPIs from reach, frequency and lift

Pick one effect KPI, usually lift in aided or unaided awareness among the target audience, and one delivery KPI, usually reach at or above a stated frequency. Adding more effect measures, such as consideration and favourability, is tempting but each needs its own sample size. The trade-off is between a single clean answer and a broader picture with wider uncertainty. Choose awareness lift alone for a brand few people know, and add consideration only when awareness is already high and the question has moved on.

Treat deduplicated reach as a known limitation

Each platform reports its own unique reach, and adding them together counts people who saw the campaign in several places more than once. Some vendors model the overlap; none observe it directly across every platform. The plan should state the method used, the vendor and the assumption, and label the combined figure as an estimate. The guide to reach and frequency across platforms covers the identity and modelling choices. Choose a single method and keep it for the whole flight, because a mid-flight switch will look like a reach jump. The reach and frequency calculator helps check that the plan's targets are consistent with the impressions bought.

Book the lift study before the flight

Lift is a comparison, so it needs a group that could have been exposed but was not. Platform studies build that group automatically if switched on before delivery; independent surveys need a holdout region or an audience held back. This is Incrementality applied to memory rather than sales, and the incrementality testing guide covers holdout and geo designs in depth. The cost is media withheld from the control and budget for the survey; the alternative is a before-and-after reading that cannot separate the campaign from the season. Choose a geo holdout when the campaign runs across several platforms, and platform studies when most budget sits in one.

Track branded search and direct traffic as lagging signals

Branded search, direct visits and the brand's share of category searches tend to move after awareness does, and they are cheap to read. They are also noisy. The plan names them as corroborating signals, defines the comparison (same weeks a year earlier, holdout regions against exposed ones) and sets a read date after the flight ends rather than during it. Choose to report them beside the lift result, never instead of it. The share of voice guide explains how Share of Voice and share of search relate to longer-term market share. Market share itself lags, because it records what buyers have already bought, so it belongs in the plan's longer-term reading rather than at the flight's read date.

Put viewability and attention thresholds in the spec

An impression bought on a premium video placement and one served below the fold are not the same unit, yet both arrive in delivery reports as one impression. The plan states the viewability standard each buy must meet, the verification vendor, and any attention or completion measure used to compare placements. The viewability and attention guide covers definitions and vendors, and the YouTube ad formats guide explains how skippable, bumper and non-skippable formats are counted. Stricter thresholds raise effective cost per impression (the CPM calculator makes that visible) but make reach comparable; choose them.

Decision points in a worked awareness plan

Illustrative example — not a real result

A measurement plan for an invented helmet brand, Halyard Helmets

Halyard sells cycling helmets through sports retailers. Its unaided awareness among adult cyclists is about 6% in a pre-flight survey, against an aided awareness of 21%. It plans a six-week national video flight of £420,000 across connected TV, YouTube and an online video network. The plan, condensed:

GoalKPISourceBaselineTargetRead
Cyclists recognise the brandAided awareness lift, exposed versus controlSurvey of 2,400 cyclists, two holdout regions21%+5 points in exposed regionsTwo weeks after flight
Enough of them see it often enoughReach at 3+ exposures among adults 25 to 54Platform reach plus modelled overlap, vendor statedn/a35% of target audienceWeekly
Impressions are comparableViewable impressions, video completionVerification vendorn/a70% viewable, 60% completionWeekly
Interest followsBranded search in exposed against holdout regionsSearch trend dataSame weeks last yearExposed regions up 15% more than holdoutFour and eight weeks after
Category presence growsShare of search among helmet brandsSearch trend data9%11%Quarterly

The holdout regions cost about 8% of national reach. Conversion events from the website are kept as a secondary report only, and online helmet sales from exposed regions are compared with holdout regions as a check rather than a target.

The plan made one awkward decision visible early: without the two holdout regions, the survey could only compare before and after, and spring is when helmet interest rises anyway. Losing a little reach bought a result the board could believe.

Measuring the plan: KPIs and cadence

During the flight the plan reads delivery weekly: reach and frequency against target, viewable share, completion rate and cost per thousand viewable impressions. Effect measures are read once, after the flight, when the survey closes; reading lift mid-flight on a partial sample invites stopping early on noise. Lagging signals are read at set dates after the flight and compared against holdout regions and against the same period in earlier years. Over a longer horizon, regular brand-tracking waves show whether awareness holds or decays, and the tutorial on building a quarterly brand health report carries those waves forward beside share of search, delivery and share of voice once the campaign is over. The media planning framework sets the frequency the reach KPI is measured against; the plan only has to record which one was chosen.

Where an awareness plan misleads

Survey lift depends on the question wording and the sample, and small samples can produce impressive-looking lifts that vanish in the next wave. Modelled reach is a modeller's estimate of overlap, and different vendors disagree. Branded search rises with retail promotions and news coverage as easily as with advertising. And the plan measures what the campaign changed in the short run; the longer-term effect on sales, the reason for running awareness at all, needs evidence the plan can only point towards.

Common mistakes

  • Judging awareness on clicks. The buyer will optimise towards people who were already close to buying.
  • Adding platform reach together. The total counts the same person several times.
  • Booking the survey after launch. There is then no clean control to compare against.
  • Reading lift weekly. Partial samples swing; read once, on the full sample.
  • Counting every impression as equal. Without thresholds a cheap placement looks like efficient reach.
  • Treating branded search as proof. It corroborates a survey result; it does not replace one.

Frequently asked questions

Can an awareness campaign have a conversion KPI at all?

It can carry one as a secondary signal, but not as the measure it is judged on. Conversions from people reached by a broad video flight are few, delayed and mostly credited to whatever the person clicked last. A plan that makes them the headline will push the media buyer towards narrow, cheap audiences who were going to buy anyway, which defeats the objective.

When should the brand-lift study be set up?

Before the first impression is served. A lift study needs a control group that was eligible but not exposed, and the platforms and survey vendors build that group from the start of delivery. Adding a survey halfway through gives a sample of exposed people with nothing clean to compare them against.

Is branded search a reliable measure of awareness?

It is a useful lagging signal rather than a measure. It responds to many things besides the campaign, including seasonality, news and retail activity, and it only counts people curious enough to search. Read it against the same weeks in earlier years and against regions the campaign did not reach, and treat a rise as corroboration of a survey result rather than proof on its own.

Terms used on this page

Tools to try

Frameworks

Where this page is used

Also related

Sources

  1. Digital Marketing and Measurement Model — Avinash Kaushik
  2. Institute of Practitioners in Advertising — IPA

Related