• Have any questions?
  • info.zbook.org@gmail.com


4d ago
5.17 MB
7 Pages
Last View : 3d ago
Last Download : n/a
Upload by : Jewel Payne

REPORT:PODCAST ADS PERFORM,HOST READS OUTPERFORMBy Paul RiismandelIntroductionSince the dawn of podcast advertising there hasbeen a long-held belief that ads read by podcasthosts work exceptionally well. This comes not justfrom those of us who sell podcast ads; we’ve heardit over and over again from the brands that haverealized class-leading returns on their spend, fromthe trailblazers who first invested in podcasts to theleading brands of today.A leading reason why we believe host-reads excel –and why so many brands choose them – is listeners’trusted relationship with hosts. For them a podcasthost is not just a personality, but a dependableinfluencer.To do that we need to focus on those who webelieve are most influenceable: a hosts’ fans andlisteners. And that’s just what we did.MethodMidroll is in a unique position to conduct thisresearch because of our industry-leading first-partydemographic survey, Podsurvey. Tens of thousandsof Midroll podcast listeners have opted-in to participate in podcast research, and we tapped into thisresource to recruit a pool of respondents who arevery likely to be listeners to one of the shows on ourroster.Though our confidence is high, we wanted toput this hypothesis to the test. That’s why Midrollcommissioned a brand lift survey with Signal HillInsights to tackle this very question: do host-readpodcast ads outperform announcer read ads?692 podcast listeners chose to complete oursurvey, and 412 of this group told us that they’relisteners to the show we tested, allowing us toisolate results from that group, gauging the impactof that relationship.The Influencer EffectIn an online survey, one group of respondents – theexposed cell – heard two clips of the podcastunder analysis, each with one ad from a differentbrand.For a number of very practical reasons (see the sidebar discussion), most brand lift studies in podcasting work with general population panels of podcastlisteners. The folks who participate in the study arepodcast listeners – usually ones who like the genreof the podcasts being tested – but not necessarilyregular listeners or fans of that podcast or podcasts.While this method very reliably lets us measurepodcast ad effectiveness, we’re less able to isolatethe influencer effect tied to listeners’ relationshipwith hosts.stitcher.com midroll.com/insightsHalf of the exposed cell heard announcer-readads, and half heard host-read ads.Another group – the control cell – heard the sameclips, but without those ads.In order to help balance out effects of the brandsthemselves, we chose ads from four, each fromdifferent verticals. Two brands were classified asestablished brands, one in the health and beautycategory, and the other in the consumer web appcategory. The two classified as emerging brandswere in the home care and food and beveragecategories.1

Top-Line Results:All Podcast Ads Work!The good news is that all the ads we tested, announcer-read and host-read, generated a healthyamount of lifts. Confirming what we’ve seen in thedozens of brand lift studies we’ve conducted, as aplatform, podcast ads just work.However, host-read ads work even better.Let’s get into it.After hearing the podcast audio respondents completed a survey which included questions on recalland eight standard brand health measures.Lift is measured by comparing the results of theexposed cell against the control cell. A positivedifference, such as greater Total Awareness in theexposed cell, counts as a lift.Results were all tested for statistical significance at90% confidence level.For this study we measured lift two ways. We measured lift for each of the announcer-reads and hostreads over control, so we could see the core performance of each type of reads. We also measured thelift of the host-read over the announcer read, so wecould see the difference between them.Hosts Read Ads thatListeners RememberFirst, an important and core measure is unaidedrecall, which tells us how well listeners rememberthe brand they heard about. Unaided means thatit’s a fill-in-the-blank, tell us what you remember –they’re not prompted with any names.For the emerging brands we saw 59% more recall. After hearing the host-reads 27% of listeners could recall the brand, compared to just 17%who heard the announcer reads.Host-reads outperformed announcer-reads for allfour brands, and the difference was statisticallysignificant for three of them. 58% of the listenersexposed to the host-reads for the establishedbrands correctly recalled the brand name,compared to 35% of those who heard theannouncer-reads. That’s a lift of 23 points, or66% more recall.stitcher.com midroll.com/insightsYou’ll note that those recall numbers are lower thanfor the established brands, but that lines up withour experience. By definition, established brandsare more well-known, and come to mind more easily. To put this in perspective, only 6% of our controlgroup – who heard no ad – said they were familiarwith these emerging brands.Importantly, emerging brands are building theirawareness. These results show that host-reads dothat much better, creating a big boost. We’ll drilldown on that top-of-funnel story shortly.2

Hosts Drive SignificantBrand Health LiftFor brand health measures (see sidebar for a deeper dive on these measures), host-reads generatedan astounding 91% more statistically significantlifts than the announcer-reads.Now, keep in mind that the announcer-reads performed very well. Across a total of 32 possiblelifts (4 brands x 8 measures1), the announcer-reads generated 11 significant lifts. But thehost-reads generated 21. Next we’ll drill down onhow these lifts affect the sales funnel.Announcer readsperform,buthost reads outperform.Hosts Boost Awarenessand Top-Funnel MeasuresIn addition to driving 59% more unaided recall forour two emerging brands, the host-reads alsodrove greater lifts for the three top-of-funnelmeasures: top-of-mind Category Awareness;Brand Familiarity; and Total Awareness (everheard of the brand).We were confident in identifying our establishedbrands as such because we could see it in the data.With the control group – listeners who didn’t hearan ad – Total Awareness was 99% for the healthand beauty brand, and 81% for the web app. Withbrands that strong, awareness isn’t as critical asboosting measures further down the funnel.Even so, podcast ads still generated lift for established brands together in these top-funnelmeasures, and host-reads outperformedannouncer-reads.stitcher.com midroll.com/insights1 The eight brand measures are: Category awareness; Total Awareness; Familiarity; Affinity; Intent to Purchase;Intent to Look for Info; Intent to Recommend; Intent to Take Any Action.3

Hosts Entice CustomersCloser to PurchaseThe path from awareness to purchase is driven, inpart, by frequency, which is something more difficult to measure in a study limited to a single ad exposure. Therefore, when we see lifts in down-funnelKPIs in this study it’s a strong indicator of real-worldperformance.As we observed with the top-funnel metrics,consumers are just being introduced to emergingbrands, and moving them to action is even moredependent on frequency. This is where host-readads really take it to the next level.The host-reads generated lifts for emerging brandsin all four bottom-funnel metrics, while announcer-reads only boosted two: Intent to Seek Info andIntent to Take Action. Host-reads also exceededannouncer-reads in these measures. Recall thatannouncer-reads actually drove more Intent toSeek Info for the established brands, but we seethe inverse here.Looking first at the established brands, all adsgenerated lifts over control for three out of fourlower-funnel measures: Intent to Purchase, Intentto Seek Info and Intent to Take Any Action. In thefourth measure, Intent to Recommend, only hostreads generated lift.While announcer-reads did not drive Intent to Recommend, they did exceed the host-reads in Intentto Seek Info. Host-reads outperformed for the otherthree measures.That’s an interesting result, which could mean thatannouncer-reads were a little better in piquingcuriosity among listeners who are more on thefence, while host-reads did a better job of pushingthem over the fence, moving more listeners both toPurchase and Recommend.Across all forms of advertising purchase and recommendation intent are the most challenging metrics to drive. Moreover, additional exposure – beyond just the one in this study – is usually neededfor an emerging brand to move a customer to buy.So it’s not surprising that saw host-reads drivingrelatively small lifts for these measures, and announcer-reads resulted in a neutral impact.The takeaway here is that the host-reads enhancelisteners’ education about a newer brand, as well as piquingtheir curiosity to pursue more information.stitcher.com midroll.com/insights4

ConclusionMidroll was one of the very first networks to testpodcast ad effectiveness rigorously with thirdparty brand lift studies. Whether read by hostsor announcers, the more than five dozen studieswe’ve conducted show time and again that podcast ads perform very well. Now, for the first time,we can demonstrate how the effect is magnifiedwhen the show’s host does the ad.From the top of the funnel to the bottom, bothemerging and established brands realize additional benefits when they opt for host-read ads.Does this mean that a brand shouldchoose only host-read ads?Not necessarily.An effective ad campaign is bigger than one adread, and no single study or method can captureevery aspect of performance over the life of a campaign.Both announcer-read and host-read ads havetheir place, and each can play an important role ina well-designed campaign. Moreover, there areshows where only host-reads or announcer-readsare on offer, and we would never advise counting apodcast out based on that factor alone.This study confirms that host-reads are highlyeffective for hitting KPIs, especially outperformingin lower-funnel measures. At the same time, announcer-reads offer the ability to scale a messagethat drives the awareness and familiarity that bringcustomers into the funnel in the first place–all withefficiency.To maximize frequency and scale along withclass-leading effectiveness, the smartest option isto choose both5stitcher.com midroll.com/insights

Additional ResourcesSidebar:Brand Health MeasuresThis study focused on two recall measures aswell as eight brand health measures that arecommon in lift research.Recall MeasuresRecall is only relevant for respondents whowere exposed to a given brand’s message,therefore it is not appropriate to measure lift.For Aided and Unaided Recall we only reportedthe percentage of respondents who were ableto successfully recall the brand, along with thedifferences in recall between those exposed tothe announcer-reads or host-reads.Unaided RecallThis measures how well respondents are able torecall the brands they heard advertised, withoutprompting, by filling in an open text box.Aided RecallThis measures how well respondents are ableto recall the brands they heard advertised, withprompting, by letting them select from a list ofbrands, which includes both the ones that wereincluded along with ones that were not.Brand Health MeasuresFor the brand health measures there’s an opportunity to find lift in each measure across thefour brands in the study, which is how we resultin having 32 total possible lifts. These are theeight measures:(1) Top-of-Mind Category AwarenessRespondents are asked what is the first brandthat comes to mind, unaided, in a given category, such as mattresses or meal kits.As with Aided Recall, for all of the followingmeasures respondents are presented with a listof brands to rate. This list includes both brandsheard in the podcast content along with onesthat were not heard.(2) Total AwarenessThis measure refers to the proportion of respondents who indicate they have heard of eachbrand.(3) FamiliarityThis measure indicates the proportion of respondents who report being “moderately,”“very” or “extremely” familiar with each brand.(4) AffinityThis measures respondents’ affect for a brandon a five-point scale using the following descriptors: love it; like it; neutral; dislike it; hate it.Intent to (5) Look for Info, (6) Recommend,(7) Purchase or (8) Take Any Action:For each listed brand respondents are asked“which of these actions do you plan to take inthe near future?” The actions are: look up information about their products; use or buy theirproducts; recommend them to a friend; none ofthe above (exclusive).6stitcher.com midroll.com/insights

Additional Resources (cont.)Sidebar:Why Wasn’t this Tested Before?As the medium has matured we’ve gotten betterat objectively measuring effectiveness usingmethods like attribution and brand lift studies.While these metrics have provided undeniableevidence that podcast ads – across all typesof creatives – are one the very top performingad platforms, getting a clear read on that hosteffect has been elusive.There are two reasons for this gap. First, mostclient-commissioned brand studies aren’tdesigned to measure this. Any given campaignmay have all host reads or all announcer reads;very few have a mix of creatives. And those fewthat do are unlikely to have both announcer andhost reads on the same show, a prerequisite toconduct a fair apples-to-apples test.Moreover, the vast majority of studies aren’tconducted with the listeners of a specific show.Though the overall podcast audience is growing by leaps and bounds, it’s still a challenge toassemble a sufficiently sized respondent poolof any given show’s audience, especially on adeadline and on a budget.If our working hypothesis is that the performance of host read ads is enhanced by the influencer effect, then we need to focus on thosewho we believe are most influenceable: a hosts’fans and listeners.7stitcher.com midroll.com/insights

how these lifts affect the sales funnel. Announcer reads perform,but host reads outperform. Hosts Boost Awareness and Top-Funnel Measures In addition to driving 59% more unaided recall for our two emerging brands, the host-reads also drove greater lifts for the three top-of-funnel measures: top-of-mind Category Awareness;