Experiential Marketing Measurement: Experience the Payback

Experiential marketing takes place away from consumers’ habitat, making linkage to retail sales difficult. Ken Madden and Silas Fisher of Ogilvy New York outline some possible measurement solutions, including use of predictive analytics.

As marketing plans become more diverse and expand to more regularly include experiential marketing, the pressure to hold the channel financially accountable is rising. However, along with this pressure comes a slew of questions and speculations as to how the channel can be accurately measured.

Unlike more traditional channels – such as TV, print, radio and online – which marketers are used to seeing measured in standard and industry-accepted ways, experiential marketing has very few measurement norms. This puts marketers in a challenging position: balancing an increasing demand for the channel with undefined ways to ensure accountability.

In part, the measurement challenge stems from the definition of ‘experiential’ itself, which can cover a wide array of marketing campaigns. Depending on the agency or client, experiential can include lifestyle events, on-premise events, product demos, sampling efforts, mobile tours, merchant visits, retail training programmes, PR stunts, and others. These are all very different activities and should not be grouped together – nor measured – as one marketing tactic.

Furthering the challenge is the difficulty of linking experiential activities directly to sales at retail. By design, many experiential programmes aim to participate in the lives of the target consumer away from retail. Event marketing and mobile tours, for instance, intentionally engage with consumers in unexpected places, such as concerts, sporting events, tradeshows, wine festivals, commuter hubs… even on the street. Not only do these engagements take place away from retail, they often happen far away from the market or postal code in which the consumer normally shops – which makes tracing any eventual sales even more complex.

So, how can these issues be resolved? Let's start by understanding the areas in which traditional analysis techniques are often failing experiential and then examine some alternative best practices that can be applied.

One common method for measuring the impact of marketing activities is the marketing mix analysis. This is especially common in the consumer packaged goods category, but can also be deployed across other industries.

These analyses use inputs that are specific to a marketing tactic such as geographic coverage, impression estimates/actuals, time periods and dollars spent in an effort to model those inputs against retail sales dollars. Generally speaking, these analytics models are most successful at attributing sales lift to marketing tactics that:

  • Have a wide geographic coverage and a high volume of impressions or engagements. Examples of these tactics are TV,
    print, radio, FSI and digital, whose coverage and reach typically yield tens of millions of consumer impressions. For these
    tactics, marketing mix analysis is an effective approach to measuring sales impact.
  • Occur at a specific set of retail stores and are ‘continually’ executing, meaning the activity is not active for a subset of
    store hours or days. Examples of this are Catalina offers – instant redeemable coupons and on-shelf messaging – as well
    as typical ‘trade’ vehicles, such as displays, features and price promotions.

So, by design, a marketing mix approach is not set up to measure experiential marketing very well. Experiential marketing typically occurs on a smaller scale than mass media marketing channels. This is an inherent consequence in the channel, as it does not simply broadcast a one-way message, but seeks to provide a two-way dialogue, hands-on experience and a more intimate engagement between the brand and consumers. This leads to high quality consumer engagements but a lower number of impressions compared to mass media. And since marketing mix models use impressions and do not account for the‘quality’ of the engagement, experiential marketing programmes have less influence in the results than do wide-reaching media channels.

Also, experiential marketing programmes do not continually execute in the way other media do. For example, an in-store display can be executing for three weeks, seven days a week. But an in-store demo or sampling representative may only be in-store for three or four hours, once a week. Similarly, a radio spot may be running in a market for one month but a mobile tour stop in the same market may only last one or two days. These scenarios pose challenges in marketing mix analysis when trying to attribute retail sales numbers that are often aggregated at high levels (ie at the weekly level) and have difficulty focusing on the exact times of the experiential events.

Clearly, experiential marketing is in need of a measurement model that is as unique and flexible as the channel itself. The first step towards building an effective measurement model is to understand the components that make a particular experiential marketing programme unique and valuable. Very often, those components are things such as personal connection with the brand, an interactive experiences deeper engagement, and buzz or word-of-mouth value. Rather than trying to measure the volume of impressions, which is a somewhat subjective measurement, a plan should be developed to capture the number of consumer engagements, the length of those engagements, and the information exchanged during the engagements.

To use a simplified example, we conducted a series of mall sampling events during the holiday season. We intercepted 120,000 potential consumers. The engagement lasted, on average, 20 seconds for 30,000 consumers, 40 seconds for 40,000 and 60 seconds for 50,000. We then calculated an equivalent number of impressions for, say, a 15-second TV spot using the length and quality of the engagement. In this example, 120,000 event engagements is equivalent to approximately 1, 125,000 15-second TV impressions.

DETAILED DATA

While basic engagement measures can be captured with many, if not all, consumers at experiential events, more detailed data is needed if real insights are to be generated. This is where it is critical to start setting up one-to-one links between the consumer and the experiential event. Random sampling among consumers has been a very effective approach to generating these links. With each programmes survey is administered to gather basic demographic information, brand awareness, current usage, experience ratings, purchase intent, and so on. The link is then made by gaining opt-in for post-event contact.

These links, while valuable, involve more work during, and sometimes after, the event. That is where random sampling helps. By understanding the potential size of the consumer population that will be exposed to the experiential programme, analytics teams can inform execution teams of the number needed for accurate representation.

Once this sample is gathered at each event, the information can be attached to the event, and post-behavioural work can begin. Ultimately, by linking consumers to events, marketers can confidently estimate how many sales are being driven by their experiential programmes, rather than relying on the sales to show up in a marketing mix analysis, which can be like trying to find a needle in a haystack.

Beyond sales, links also allow marketers to see which events are driving different metrics, the types of consumers that respond best to their programmes and the buzz generated after the events.

By simply making the effort to capture engagement data and by adding consumer-to-event linkages, the framework of an experiential ROI model begins to take shape. For example, results from post-programme purchasing and word-of-mouth can be applied to total engagements since participants were randomly sampled. This gives a one-time conversion value and addresses short-term value.

However, a longer term value can be estimated as well. This usually requires the marriage of event data and client or thirdparty data, which includes loyalty rate, purchase cycle and profit margin. With all these inputs, a model can be built to estimate the number of repeat purchasers and their value to the brand.

IN-STORE EVENT MODELLING

For a large, global retailer, we were able to combine our own historical data on similar in-store event types with the client's reported shopping patterns, loyalty rates, purchase cycles and profit margins to build a planning model that compared the potential value of the in-store events executed in two very different ways. This allowed us to show, in real dollar terms, the most effective way to execute the upcoming programme. The model clearly showed that the deeper engagement would produce substantially more sales over the course of a year, more than making up for the increased cost. Having both short and long-term ROI estimates can be very useful when planning a mix of future experiential programmes.

There are a couple of considerations that affect how data is collected within the constraints of the experience, one of which is the concern that ‘research’ will taint the consumer experience. There is a real fear that consumers will walk away from an experience with a bad taste if they are tracked down and asked a series of questions. However, there are ways to subtly collect data and maintain an authentic experience. Random sampling drastically limits the number of people being exposed to research, sometimes needing as little as 3% to 5% of consumers.

Additionally, if the linkage work is done on the outskirts of the experience, it is also less likely to appear as a focal point to the consumer – rather, consumers merely participate in an exit poll of sorts, which is more familiar.

Even simply having the research administrators dressed in different coloured shirts or uniforms helps consumers to disassociate their experience from the information they are providing afterwards.

Finally, it is important to measure all programmes in a consistent manner. Invariably, different elements of a programme will require different approaches to measurement, but having a core set of metrics that can go across any programme allows for benchmark development and historical analysis.

At OgilvyAction, we have a database that can compare lifestyle events to retail events, one category to another, conversion rates to awareness rates, and so on. In total, there are over 500,000 consumers linked to events and more than 10 million consumer engagements on which to filter.

We have developed an app, Data Vane, to harness the power of these collected metrics. Data Vane has become not only a rich data repository, but also a useful resource to analyse and show programme effectiveness. By looking at past benchmarks, clients gain a better understanding of how their programme performed compared to aligned benchmarks – benchmarks across the same category, execution type, or even the same product in prior years. This adds a valuable layer of context to results.

In the end, what is measurement providing if there is no application to future direction? Simply knowing that a programme generated x amount of sales or did best at retailer y does not always help marketers to plan for the future and feel confident that new programmes will generate positive results.

To get to the next level, where past measurement is informing future planning, it is necessary to leverage predictive analytics. Predictive analytics uses large datasets that often take a long time to accumulate. This means that marketers need to be cognisant of where and how they are storing their data – and should always be thinking about how new data can fit into that schema.

With organised, consistent and plentiful data, marketers can begin tinkering with different inputs for a future programme and comparing predicted outputs. For example, some inputs may be estimated event attendance, number of representatives staffing the experience, length of the event and category of the brand. Put very simply, if the output you want to know is, say, conversion rate, then all the available inputs are assessed and plugged into the predictive analytics model as variable and, ultimately, a number is generated based on those inputs and past outputs.

The real value in this approach is that the learning and data from each programme executed ends up feeding back into the prediction process – making a stronger and stronger application over time. And because the process is iterative, you can start simple and build a stronger, more complex predictive model over time that leverages additional variables and generates more accurate predictions.

Once in working order, predictive analytics should play a role in planning, strategy, creative, and – as always – execution. It can help to optimise the market strategy, execution approach and product mix of an event or mobile tour. It can also warn against programmes that are likely to fail.

Getting to the predictive analytics stage takes several steps. First, you must recognise that to effectively measure experiential marketing, traditional approaches are not reliable. Second, you must build consistent measurement approaches and data storage across all programmes. And finally, you must leverage advanced analytics practices to fully maximise past learning and better plan for the future.


ABOUT THE AUTHOR
Ken Madden is executive vice-president, strategy and analytics, OgilvyAction. He leads the Insights & Research, Analytics and Marketing Technology groups.
ken.madden@ogilvy.com

ABOUT THE AUTHOR
Silas Fisher is group director, marketing analytics, OgilvyAction. He leads an analytics practice that is dedicated to developing innovative measurement tools and direct response models in experiential, event and shopper marketing.
silas.fisher@ogilvy.com

Reproduced with permission of Admap, the world’s primary source of strategies for effective advertising, marketing and research. To subscribe visit www.warc.com/admap.  © Copyright Admap.