Genauer Titel der Studie
Stories Meta EMEA
Jahr der Veröffentlichung bzw. des Studien-Starts
2020
Gattung
Online/Digital
Kontaktperson für Rückfragen
Torsten Müller-Klockmann
E-Mail-Adresse der Kontaktperson
torstenm@fb.com
Urheber / Vermarkter
1. Zielsetzung der Studie (inkl. Forschungsfragen)
Evaluating the efficiency and effectiveness of campaigns booked across Facebook/Instagram Feed and Facebook/Instagram Stories in comparison to campaigns booked in Facebook/Instagram Feed only.
2. Untersuchte / gemessene KPIs (bitte mit kurzer Erläuterung)
Lift and Cost per impacted person for
- Ad Recall
- Brand Awareness
- Campaign Awareness
- Top Of Mind Awareness
3. In der Studie untersuchte Medien (intramedial / intermedial)
Facebook/Instagram
4. Studien-Konzept (Beschreibung in wenigen Sätzen)
We used a multi cell brand lift tests for 23 Facebook/Instagram campaigns across EMEA to find out if adding stories to feed placements would generate similar or even better brand lift than feed as a standalone buy and to see if we could gain any efficiency benefits. For the test we compared two cells:
Cell A: Facebook & Instagram feed ads only
In test Cell B we added the Facebook & Instagram stories ads to this mix.
We made sure that all other test details were the same across cells, - such as budget, audience, campaign duration and buying objective to make sure we can compare apples to apples.
In both cells we ran brand KPI pollings and compared the results between test & control groups to evaluate both brand lift (effectiveness) and cost per impacted person (efficiency)
5. Auftraggeber
Facebook Inc
6. Institut(e) / Subinstitute
Facebook Inc
7. Eingesetzte Methoden ( Multi-Methoden-Ansatz oder Single-Ansatz)
Facebook Brand Lift Studies: Experimental studies using a RCT approach.
8. Art der Veröffentlichung (Broschüre, Tabellen-Band, zählbarer Datensatz)
Presentation
9. Grundgesamtheit
Test groups: People that have been reached by the 23 different campaigns (usually broad age&gender targeting)
Control groups: People that would have been reached if our lift solution would not have suppressed ad delivery as they were in a randomly drawn control group.
The 23 campaigns ran in 6 different countries within EMEA, out of which 12 originated out of Germany (7 out of the UK, and 1 in each of the following countries: Netherlands, Italy, South Africa and United Arab Emirates).
Advertisers came from different verticals: 14 x CPG, 2x Autos, 2x Telco, 2x Ecomm, Entertainment, Financial Services, Gaming– 1 each
10. Erhebungsmethode (inkl. Technische Messung)
The Facebook lift approach is a Randomized Controlled Trial approach.
For each of the 23 campaigns the lift solution created a randomized holdout group for the campaign (5-10% of all Facebook/Instagram users).
During any campaign delivery, for every user who visits Facebook or Instagram, the algorithm decides whether this users should see the ad, according to multiple criteria as target group, location, preferences, bid in the auction, etc.
If the algorithm decides that this user should see the ad, the last step is checking whether she or he is in the holdout group for the campaign. If the latter is the case, this user won't see the ad, but get classified into the control group for this test.
If the user is not in the holdout group, the ad will be delivered and the user gets classified into the test group.
This approach ensures, that users in test & control group have identical characteristics, not only regarding demographic attributes but actually all criteria our algorithm considers for ad delivery. The only difference between the groups is ad delivery happening or not.
In this study there were two test and control groups as we randomly split all Facebook and Instagram users in two equally sized, disjunct cells in which test & control groups were created.
In Cell A the algorithm could deliver Facebook & Instagram feed ads only
In Cell B the algorithm could deliver Facebook & Instagram feed and Facebook and Instagram Stories Ads.
We made sure that all other test details were the same across cells, - such as budget, audience, campaign duration and buying objective to make sure we can compare apples to apples, in particular fair efficiency comparisons due to the same budget per cell.
To measure brand impact, we poll 500 persons per cell, group and question 4-48hours after exposure (or would-have-been-exposure for the control group) when they are returning to their feed, with regards to ad recall, brand awareness, campaign awareness or top of mind awareness.
The difference between test & control group shows to which degree campaign exposure can lift each KPI.
Budget per cell / (lift in percentage points x the number of reached people per cell) is the cost per impacted person.
11. Feldzeiten / Untersuchungszeitraum
Campaigns ran for several weeks each in the period from April 2018 to September 2019.
12. Fallzahlen (ungewichtet) / Wellen
Each campaign polling had 3 questions with at least 500 polls per cell and question each in test and control group each, so a sample of 2x3x2x500=6,000 participants per campaign (138,000 in total).
13. Stichproben-Auswahl (z.B. Quote, Random, Einsatz von Access-Panels)
Randomized Controlled trial.
14. Ggf. Rekrutierung / Incentivierung der Probanden / Interview-Dauer
Polls appeared in Facebook/Instagram Feed, one questions per user, no incentives.
15. Verwendete Datenquellen (z.B. Werbeaufwendungen von Nielsen Media Research)
None.
16. Messung / Berechnung der Kontakte bzw. Kontaktwahrscheinlichkeiten
Measured.
17. Eingesetzte Analysetechniken
Comparison of test vs. control groups.
18. Gewichtung (vor oder nach Analyse, Quelle der Gewichtungsvorgaben)
none
19. Ggf. eingesetzte Fusionstechniken
none
20. Verwendung von Benchmarks (bitte Erläuterung dazu, z. B. inter- oder intramedial, nach Branchen, nach Werbeformen etc.; Quelle von externen Benchmarks, Art der Berechnung)
none
Zurück