Quer saber como prever se uma campanha irá funcionar ou não para seu negócio?
Este webinar ensina como conectar o Ad Tech ao Marketing Tech, ajudando a tomar decisões estratégicas para sua compra de mídia, em tempo real!
Aprenda a conectar sua estratégia de mídia aos seus KPIs de negócio e como reportar e reagir às tendências que podem afetar o seu negócio.
Palestrante: Fernando Takahashi - Estrategista de marketing digital na Adobe responsável por toda a área de Data Driven Marketing e Acquisition Acceleration.
Bio: É formado em Marketing pela Anhembi Morumbi e tem mais de 18 anos de experiência em tecnologias voltadas para o marketing online e offline. Já passou por empresas como Y&R, Grey, Ogilvy, DraftFCB, além de anunciantes como o Pão de Açucar.
13. Integração Ad Tech e Mar Tech: Analytics
Others
N/ADiferença
Integrado
Dados de
Analytics
Comportamento no
site e engajamento
Não-
Integrado
14. Colete dados de forma integrada através de todo o
funil
OthersEverybod
y Else
Others
Disponivel através de
integrações com
ferramentas de
analytics
Buscas/Keywords
Impressões
Clicks
Conversão
Comportamento pós-click
• Compra
• Engajamento: tempo gasto/re-visitas
• Visualizações de página/produto
• Bounce Rate (Taxa de rejeição)
• Outras métricas de engajamento
”Feeds” das Search
Engines
Tracking de conversão
Não disponíveis
sem uma integração
15. 15
Saiba como seus indicadores de performance trabalham juntos
Resultado
por impressão
CTR Taxa de rejeição
(Bounce Rate)
Taxa de Conversão
Impressões Clicks Engajamento Conversão
16. 16
Este tipo de integração também permitirá que os analistas aprofundem as análises, correlacionando outras dimensões
com os resultados de campanha. Este tipo de análise permitirá um conhecimento muito melhor do funcionamento
das keywords, campanhas e até auxiliar na distribuição de verba de acordo com os resultados obtidos.
Número da Visita
Fidelidade do cliente
(Novo, Retornante,
Consumidor leal)
Línguas
Fuso horário
Geo-Segmentação (País, Estado/ Província
/ Região, Cidade, CEP, DMA)
Navegador
Sistema Operacional
Resolução do Monitor
Tipo de Conexão Operadora CelularHorário do dia
Dia da Semana
Acima estão somente alguns exemplos. De acordo com seu objetivo de negócio, outras métricas podem ser definidas.
Saiba como seus indicadores de performance trabalham juntos
20. 20
• Search Engine e Site
• Estrutura de Campanha
• KW Pesquisada
• Keyword por Search Query
• Keyword por Busca Interna
Adobe.com Insights
21. 21
• Search Engine e Site
• Estrutura de Campanha
• KW Pesquisada
• Insights Modificadores de
Bid• Campanha por hora
• Campanha por
demográficos
• Contas por Geo
Adobe.com Insights
22. 22
• Search Engine e Site
• Estrutura de Campanha
• KW Pesquisada
• Insights Modificadores de Bid
• Produtos
• Campanha por Produto
Adobe.com Insights
23. 23
• Search Engine e Site
• Estrutura de Campanha
• KW Pesquisada
• Insights Modificadores de
Bid
• Produtos
• Retenção do visitante• Keyword por Tempo Gasto
• Keyword por No. da Visita
Adobe.com Insights
28. • Desequilibrado
• Somente Manual = Perda de oportunidades. Estresse.
• Somente Automação = Otimização exagerada.
• Equilibrado
• Manual para as anomalias = Responsivo sem estressar
• Regras para as prioridades baixas = Foco para campanhas de maior
importância
• Portfolios = Automação do media mix.
Duas cabeças pensam melhor do que uma.
Tempo $$$$Tempo $$$$Tempo $$$$
Manual
Volume
Auto
Eficiência
29. Tracking
Dados das Search Engines
Dados de comportamento no site vindo de
Analytics
Conversões
Clicks
Impressão
Engajamento no site
• Visualizações
pg/prod
• Tempo Gasto
• Taxa de rejeição
• Micro-conversões
Clicks
Ad Set/KW+MT
Comportamento
no site
Conversões
4 4
• Média 5
Visualizações
por visita
• 10% Bounce
• Múltiplas visitas
• Tempo médio
gasto no site:
4:10
• Média de 1.5
visualizações por
visita
• 75% Bounce
• Sem visitas
adicionais
• Tempo médio
gasto no site:
0:20
0 01
Não disponíveis
Keyword 1 Keyword 2
+28% Bid -31% Bid
Automação das modificações de Bid
Automação ajuda a alavancar oportunidades
Q: Como vocês comprariam estas duas kw?
30. Mesmo ”Bid” ($ 0.72), Performance diferente
Posição MédiaCPC MédioCTRClicksCustoBid
O aumento de performance pode ser alcançado se ”bidarmos” cada
keyword baseado em seu valor para o negócio.
* measured from Sep 1, 2015-Sep 30, 2015
Google Main acct, active keywords only
32. Objetivos Ponderados
Campanha de Branding
Impressões / Clicks
Profundidade da navegação: 20%
Tempo gasto no site: 30%
Assistiu a um vídeo
ou baixou um material
ou se inscreveu em um formulário: 50%
35. Otimização em todos os níveis
Portfolio
Engine
Account
Campaign
Ad Group
Keyword
Bid
Unit
36. Automação de Bids
36
First PPC
Platforms
Enhanced
Campaigns
AdWords
Real-Time
Bidding
Shopping
Campaigns
Complexity
37. DIMENSIONAL
PORTFOLIOREGRAS
Níveis de maturidade na automação de bids
KW1 KW2
Desktop
Mobile
SaturdayFridayThursdayWednesdayTuesdayMondaySunday
SaturdayFridayThursdayWednesdayTuesdayMondaySunday
Desktop
Mobile
SaturdayFridayThursdayWednesdayTuesdayMondaySunday
SaturdayFridayThursdayWednesdayTuesdayMondaySunday
KW1 KW2 KW1 KW2
PORTFOLIO
38. DIMENSIONAL
PORTFOLIOREGRAS PORTFOLIOObjetivo: Gerar o maior número possível de conversões a ≤ $20 CPA
$20 CPA
25 Conversions
KW1
DESKTOP
KW2
MOBILE
KW2
DESKTOP
KW1
MOBILE
$20 CPA
30 Conversions
$20.61 CPA
33 Conversions
$15.33 CPA
21 Conversions
KW1 KW2
Níveis de maturidade na automação de bids
42. 42
Estimativas de economia de tempo
Search
Tempo gasto
antes (horas)
% de economia de
tempo
Tempo gasto depois
(horas)
Otimização Manual de Bids 15.00 80% 3.00
Criação de novas campanhas (estrutura, palavras-chave, negativações) 10.00 30% 7.00
Composição de relatórios 5.00 70% 1.50
Otimização de Landing Page (Testes A/B) 3.75 30% 2.63
Testes de Criativo (headlines, ETA, etc) 3.00 35% 1.95
Expansão de palavras-chave 2.00 25% 1.50
Monitoramento de Search Queries, Expansões 2.00 25% 1.50
Descoberta de novas palavras-chave 1.25 25% .94
Total 42.00 52% 20.01
43. Foco de tempo das análises
Tarefas Horas gastas
Criação de Campanha 25.00
Bid Manual 20.00
Relatórios 11.00
Planejamento 5.00
Expansão 5.25
Testes/Otimização 16.75
Tarefa Horas gastas
Criação de Campanha 25.00
Bid Manual 20.00
Relatórios 11.00
Planejamento 5.00
Expansão 5.25
Testes/Otimização 16.75
Baixo Valor
Alto Valor
67%
33%
Task Hours Spent
Criação de Campanhas 25.00
Manual Bidding 20.00
Relatórios 11.00
Planning 5.00
Expansão 5.25
Testing/Optimization 16.75
Tarefa Horas gastas
Criação de Campanha 25.00
Bid Manual 20.00
Relatórios 11.00
Planejamento 5.00
Expansão 5.25
Testes/Otimização 16.75
Baixo Valor
Alto Valor
22%
78%
Sem ferramentas Com ferramentas
45. Adobe.com | Melhorias na previsão de resultados
0.00%
20.00%
40.00%
60.00%
80.00%
100.00%
120.00%
140.00%
160.00%
180.00%
1 2 3 4 5 6 7 8 9 10 11 12
Accuracy
(100%isthebest)
Portfolio
old model new model
92.03%
Accurate82.26%
Accurate
46. Adobe.com | Melhorias no RPC e no ROI
0
1
2
3
4
5
6
7
8
rpc roi
Portfolio APre
Post
rpc roi
Portfolio B
+66.6%
+70.4%
+19.5%
+5.7%
47. Os Insights do gestor
de negócio são
expandidos
Começo Meio Fim
Solução e
Próximos Passos
Compartilhe as
recomendações e discuta
os próximos passos
Elevando os
Insights
Compartilhe
descobertas que
revelam análises
mais profundas do
problema ou
oportunidade
Momento ”A-ha”
Apresente a maior
descoberta ou Insight-
chave
Set-up
Contexto da
situação atual,
personagem(n
s) e o gancho
Gustav Freytag
(1816-1895)
Entregando suas análises
48. Construindo sua história com dados
Momento ”A-ha”
Pense neste momento
primeiro. Ele é o
"clímax" da sua
história. Tudo deve
convergir para ele.
Set-up
Procure
destacar o
impacto do
contexto nos
negócios
Elevando os
Insights
Compartilhe
descobertas que
ajudem a construir o
momento A-ha
Solução e
próximos
passos
Determine estratégias
de ação curtas e que
demonstrem
resultados mais
rapidamente, para
gerar tração, sempre
visando o objetivo
final.
49. Como esta estrutura se compara a outros modelos?
Passado Presente Futuro
Set-Up
Elevando
os Insights
Momento
”A-ha”
Solução e
Próximos
Passos
Problema Solução Benefício
O que? E daí? E agora?
Does anyone kNãow why Ciclistas ride like this? (Drafting)
When riding in a pace line, how much effort do you think the second Ciclista is giving? Third Ciclista?
Performance is Nãot only about playing your best, but playing the smartest. 29% energy savings over a 100-mile course adds up to a huge boost in performance.
Lead Started/Initiated Lead Completed Final Sale from Lead
Scottrade – customer score = credit score, trade volume, trade values, etc.
Overcome bid inertia by separating the
three hidden parts of every keyword report
Zoom in to each colored block and show some example data
WHAT TO LOOK FOR
The above screenshot of the performance Segmentos goes into more detail, but here is the general idea:
Reward the most efficient converting keywords with higher bids to trade some of that efficiency for more volume at a higher position
Punish the most inefficient converting keywords with lower bids to re-establish balance between volume and efficiency for the maximum profit
If micro-conversions were included, decrease keywords with the highest spend more slowly than you would with obviously wasteful keywords. (Investigate the conversion path to see if the site could be adjusted to lead more visitors to the final conversion)
Pull back aggressively on Sem Conversões keywords with the most spend. Even the best accounts have 5-10 keywords at the top of this segment that are responsible for the majority of waste. (This is where solutions like the Cross Visit Participation VISTA rule can help you validate whether these keywords are valuable "root" terms that are resulting in a large number of assists or conversions from a different allocation perspective or attribution model. However, until you can prove that theory with data avoid ego bidding and make sacrifices according to the data available to you).
Pause or significantly reduce the bids of keywords with an extremely high number of Impressões that have not led to any clicks. "Extremely high" will vary for the unique situation and environment of each account, but a 1% CTR is generally accepted as a lenient benchmark for any industry. If a keyword has thousands of Impressões in a top fold position without any clicks, its quality score should be improved with better ad alignment before bidding will make much of a difference. However, if a keyword has received a significant number of Impressões with no clicks at a low position, consider increasing the bid to give the keyword's ad a better chance of getting noticed so that more valuable metrics can be used to measure the keyword's conversion potential.
Clearly explain that the stars represent outliers
Each keyword segment should be optimized towards the most appropriate KPI
Then you can simply focus on the 80/20 concept and handle the top and bottom
Then let automation handle anything that is not an extreme
The obvious Segmentos are converting, non-coverting Custo, and clickless searches
Smooth out the Segmentos by falling back to multiple microconversios and time frames
Allocation: Orders (Last Click) Orders (First Click) Assists (Orders in the Middle)
Retail: Orders Cart Adds Product Views Long Term Orders
Because these are complex business decision that must be adapted to each unique situation, these are the types of things that should not be delegated to a machine.
ASSUMPTIONS
"Sem Conversões" has become increasingly difficult to define as many have begun preaching that "last click is dead". Although I agree that difficult questions must be asked to determine whether default attribution models should be accepted, real world experience has shown that a small number of keywords actually point marketers towards a different type of optimization when viewed from some other perspective like first, last, linear, weighted, etc.
The few keywords that do prove their worth outside of the last click model are definitely worth digging for. SearchCenter and SiteCatalyst can use the CVP (Cross Visit Participation) VISTA rule to provide users with very flexible attribution models. Once it is implemented, users simply add similar conversion metrics in different allocation perspectives with a single click just like they would for any other metric. Since the development of the CVP VISTA rule over a year ago, "Assists" have grown to become one of the most powerful micro-conversions available.
However, the majority of clients who still do not have the CVP VISTA rule implemented should let fear and paranoia prevent them from taking action on the valuable data already available to them. It is better to learn from action and mistakes than to wait passively and make excuses with unproven theories.
TIM WADDEL FEEDBACK ON ROUGH DRAFT
Keyword Segmentation similar to Audience Segmentation in Display
Those that read reviews, added something to the cart
Responding with retargeting based on that last action
Attribution / Assists
Through attribution, how can you more easily identify the varied roles of keywords based on their different position in the funnel
Balanced with avoiding paranoia. Dont let the assists prevent you from the obvious. Build awareness in the ES solution (who cares about "productized") In the meantime, use what you have. You dont have to boil the ocean. Guilty until proven innocent. Avoid ego bidding.
The segmetation slides made sense to me last year, how can we make that even more clear with real examples and visualization instead of words
The difference – what no one else can offer – is a complete picture of your conversion pipeline
- Having a fuller picture means you can more clearly see value and effectively chase qualified visitors, putting more in the top of your pipeline, and getting more out of it on the bottom line
What data is available to us and how can we collect them along the ad conversion funnel
For the offline data solution in analytics to work we need to configure the report suite for this option, and you need to load a feed into analytics which has the “hidden” transaction ID + the offline conversion metrics you want to associate, + the transaction date for those metrics. If your s_code has set the following:
S.transactionid=s.purchaseid
Then the transition ID you use will be the purchase ID – this simplifies things quite a bit. If you have Nãot done the above, then you’ll have to find the transaction ID for each conversion and I don’t kNãow exactly where you get that information. I think you can report out on that Nãow – but Nãot sure which report.
Clearly explain that the stars represent outliers
Each keyword segment should be optimized towards the most appropriate KPI
Then you can simply focus on the 80/20 concept and handle the top and bottom
Then let automation handle anything that is not an extreme
The obvious Segmentos are converting, non-coverting Custo, and clickless searches
Smooth out the Segmentos by falling back to multiple microconversios and time frames
Allocation: Orders (Last Click) Orders (First Click) Assists (Orders in the Middle)
Retail: Orders Cart Adds Product Views Long Term Orders
Because these are complex business decision that must be adapted to each unique situation, these are the types of things that should not be delegated to a machine.
Let’s say you’re flying from LA to NYC.
Unfortunately you board a Southwest flight (Missouri wrong airport reference) whose compass is off by one degree.
Where do you think you would land?
Accuracy is paramount in flight as well as in our models. If we project a return rate at a certain bid amount, we need to do it accurately.
With more than 110 product development resources, we are constantly inNãovating and releasing new updates to the product. We have releases every 2 weeks, which enables us to consistently beat our competition to market with new search engine API features. We were first to market with support of Google Enhanced Campaigns, first to market with support for Google Product Listing Ads (PLAs) and first to market with support for Yandex and the latest version of Yahoo Japan.
And because we’re profitable, unlike all of our competitors, we are continuing to grow our development resources to get even faster.
With more than 110 product development resources, we are constantly inNãovating and releasing new updates to the product. We have releases every 2 weeks, which enables us to consistently beat our competition to market with new search engine API features. We were first to market with support of Google Enhanced Campaigns, first to market with support for Google Product Listing Ads (PLAs) and first to market with support for Yandex and the latest version of Yahoo Japan.
And because we’re profitable, unlike all of our competitors, we are continuing to grow our development resources to get even faster.
Just Human / Manual = Stress and missed opportunities
Only Automation / Rules = Failure and over-optimization
With more than 110 product development resources, we are constantly inNãovating and releasing new updates to the product. We have releases every 2 weeks, which enables us to consistently beat our competition to market with new search engine API features. We were first to market with support of Google Enhanced Campaigns, first to market with support for Google Product Listing Ads (PLAs) and first to market with support for Yandex and the latest version of Yahoo Japan.
And because we’re profitable, unlike all of our competitors, we are continuing to grow our development resources to get even faster.
With more than 110 product development resources, we are constantly inNãovating and releasing new updates to the product. We have releases every 2 weeks, which enables us to consistently beat our competition to market with new search engine API features. We were first to market with support of Google Enhanced Campaigns, first to market with support for Google Product Listing Ads (PLAs) and first to market with support for Yandex and the latest version of Yahoo Japan.
And because we’re profitable, unlike all of our competitors, we are continuing to grow our development resources to get even faster.
Better forecasting meant better bidding, which delivered tangible results on the bottom line – improved RPC and ROI.
Let’s look at the same structure from a data perspective. First, you start with the set-up where you provide background on the current situation, characters, and the hook. The hook is the inciting incident where you reveal there’s a problem or opportunity. You then share additional findings that provide deeper insights into the problem or opportunity. Eventually, you reach the climax of your presentation where you share your main finding or insight, which should be monetized so that it catches the attention of your audience. How quickly you get to your aha moment will depend on your analysis – some may not have as many layers and won’t need as much build-up. Then we jump to the recommended solution and potential next steps. Through this process your audience’s knowledge is expanded.
Now let’s dig into how you can leverage this data storytelling arc.