Beyond last click: Understanding your consumers’ online path to purchase Insights and learnings from Google’s clickstream research

Contents • Executive Summary • Introduction • Methodology • Online Research Duration • Purchase Paths • The Role of Search • Brand vs. Generic Searches • Conclusions • Next Steps

Executive Summary This paper summarises the key findings and insights from Google’s research into the important area of clickstream attribution. It will give you the facts to answer key questions such as: how long is the typical path to purchase, how many sites will be visited on average and how does search contribute to consumer decision making? We hope that this paper will encourage marketing leaders to move beyond a ‘last click wins’ model and take a fact-based approach to understanding their consumers’ path to purchase. Key findings include: • Research journeys are long and can last a month or more in some categories. • One in three conversions occur 30 days after the online research began. • 70% of purchasers use search at some point in their research journey. • Search paths to purchase often include both brand and generic terms and 48% of purchasers switch between terms at some point in their journey. • The way shoppers research and purchase different products online does not neatly follow traditional marketing funnel models. • The amount of research conducted online is not clearly linked with the average order value of the products in question.

Introduction According to the Connected Kingdom Report1, the Internet contributed £100 billion to the UK economy in 2009, and last year 62% of UK consumers bought goods or services online. The online advertising market continues to grow and the latest PwC/IAB figures show it is now worth over £4 billion in the UK2. The sheer number of people engaged in online research in some categories is staggering. For example, 60% of the UK internet population researched travel products online in a 3 month period3, and half of all travel in the UK is booked online1. In the apparel sector, where the majority of sales still go through retail stores, there were over 22 million people researching apparel online in the UK in the 3 month period we analysed2. In a market where £1 in every £4 of marketing budgets is spent online, understanding shoppers’ online journeys will help ensure online marketing activity and advertising measurement practices reflect real consumer behaviour patterns. To this end, Google and Nielsen partnered to conduct clickstream analyses of online research and purchase behaviour in multiple sectors in the UK. The objectives of these clickstream analyses was to provide specific insights into: • • • • •

What consumers research online How long they spend researching Which sites they visit How they navigate between sites What they search for

1. Boston Consulting Group / Google - Connected Kingdom report, October 2010. 2. IAB / PwC as spend figures for 2010. 3. Custom Google / Nielsen clickstream research, 2010.

Methodology The clickstream analyses commissioned by Google are based on data from Nielsen’s NetView online consumer panel – the source of UKOM data. The first step in a study is to identify a sub-set of NetView panellists by defining certain behaviours we are interested in, for example researching travel online, and then examine all other category related activity in a given period. Then we isolate a time period of analysis – usually a three month period. This length was chosen as previous research in this area indicated it was long enough to capture entire journey periods but short enough to isolate the relevant activity. Given the elongated nature of the house purchasing process in the UK, the property study was run over a longer 6 month time period. Once a category and time period has been chosen we then work with Nielsen to classify the sites and search terms into relevant sub-categories. For a given category (such as car insurance or travel) we identify all the relevant sites that fall under that category. Then different types of sites within a category, e.g. aggregator or advice sites, are sub-classified. We also analyse website URL structures to identify consumers who performed an action of interest, e.g. placing an item in a basket, filling in an application, or buying a flight. Finally, all category-relevant search activity is identified and classified as either branded (e.g. “Vodafone”, “Nokia”), generic (e.g. “phone contract’’, “mobile phone”), or a combination of both. Once we have classified sites and search terms, we segment the audience into mutually exclusive groups: researchers, estimators, and purchasers. Researchers are people who visit content-related pages but don’t get a quote or transact. Estimators are people who place items in baskets or get quotes online but do not make a purchase. Purchasers are people who transact or perform another conversion event, as defined in figure 1. For the majority of this paper we focus on purchasers.

As we are using Nielsen’s opted-in consumer panel all data protection laws have been met and no personally identifiable information has been used in these studies. Figure 1: Studies included in this paper UK Study

Sample size*

Definition of purchase

Travel

15,453

Online travel purchase

Apparel

12,831

Online apparel purchase

Mobile Phones

6,713

Online mobile phone purchase

Energy

4,729

Online utilities purchase

Car Insurance

3,796

Online insurance purchase

Loans

1,700

Online loan purchase

3,722

Download (PDF, schedule etc), Action (email friend etc) or contact pages

Property

* Sample size refers to any user who visited content or searched on terms related to that category in the defined period.

The value in using a panel-based analysis, compared with website analytics or ad server tracking data, is that we get a more complete picture of the online journey across the whole category ecosystem, rather than just through the lens of an individual advertiser’s touch points with their shoppers.

Online Research Duration 1 in 3 conversions takes place 30 days after the research began One of the first things we learned from the clickstream research is that consumers typically spend much longer researching purchases online than might be conventionally assumed from examining ad server or analytics path data. People shopping for clothes or shoes online spend, on average, almost a month shopping around and making up their mind before making a purchase. Travel, with all of its fantasy shopping potential and myriad of providers, aggregators and advice sites involves journeys with an average length of 24 days. Figure 2: Average journey length in days per purchase 30

27 24

25

23 20

20

15

12 9

10

5

0 Apparel

Travel

Mobile Phones

Base: All purchasers as defined in each category

Energy

Car Insurance

Loans

Though the average path to purchase for all verticals is under 1 month, we find that one in three conversions occur at least 30 days after the initial research began. In travel, more than one quarter of purchases occur after 30 days. This trend is repeated in retail where one third of transactions occur after 30 days. In contrast, when people research online for a gas or electricity provider about half (54%) of them convert within 24 hours, compared with only around 20% of within the day purchases in travel and apparel shopping. Figure 3: Cumulative percent of purchases for apparel, travel and energy

100% 90%

% of people who purchased

80% 70% 60% 50% 40% 30%

Apparel Purchasers Travel Purchasers Energy Purchasers

20% 10% 0

<24 1-5 6-9 10-14 15-19 20-24 25-29 30-39 40-49 50-59 60-69 70-79 80-89 Hours Days Days Days Days Days Days Days Days Days Days Days Days Time to purchase Base: All purchasers as defined in each category

Advertisers can determine what their own conversion journeys look like by using conversion tracking tools which allow them to make informed decisions on their cookie windows. Ideally such decisions would start to get category specific – the attribution window for different categories should reflect different customer journey lengths.

Figure 4: Average research time in hours per purchase 03:30

03:17

03:00 02:30 02:00

01:43 01:29

01:30

00:50

01:00

00:43 00:29

00:30 00:00 Apparel

Travel

Mobile Phones

Car Insurance

Energy

Loans

Base: All purchasers as defined in each category

More days in market usually adds up to more activity racked up online – in terms of both hours spent and number of sites visited, as illustrated in Figures 4 & 5. Figure 5: Site visits before a purchase Category

Average site visits per purchase

Number of different sites visited per purchase

Average number of visits to each site

Travel

21.6

9.4

2.3

Apparel

11.4

2.9

3.9

Mobile Phones

9.4

4.1

2.3

Energy

7.4

3.4

2.2

Car Insurance

5.6

3.9

1.4

Loans

4.2

2.8

1.5

Overall Averages

9.9

4.4

2.2

Base: All purchasers as defined in each category

Purchase Paths Journey paths can be very long and complex In order to bring these various metrics to life, it helps to look at actual user paths, and how people search and research on their way to a purchase. Figure 6: Travel path Quotation

Searches

DAY Sites Visited

Purchase Blue bay escape Blue bay escape

Corfu holidays 1 First Choice

2 Travelzoo African Safari Club

4 Travelzoo First Choice TUI Thomson

7 Bing maps & local Holiday-Truths.com TripAdvisor TUI Thomson Olympic Holidays

Searches

DAY Sites Visited

TripAdvisor Kallisto resort, corfu Kallisto resort, corfu 16

Kontokali bay resort & spa Kontokali bay resort & spa Kontokali bay resort & spa Bus tours in corfu 22

TripAdvisor First Choice Holiday-Truths.com Olympic Holidays TUI Thompson TUI Thomson Travel Republic Booking.com Google Maps

Palace mon repos 12 TripAdvisor TUI Thomson Sunshine.co.uk Bing maps & local

Corfu holidays 27 Travelzoo

28 First choice

Travel Library

Travelzoo

For example, above we have an individual’s online journey for a high involvement purchase: travel. This particular journey lasted 28 days, and included many generic keywords (generic in this context means travel related terms not including any brands for companies in the travel sector). The range of sites visited include aggregators such as TripAdvisor, as well as tour operators such as First Choice and Thomson, many of whom provided quotes – but First Choice gets the business in the end after all that shopping around. The final purchase is made after the individual searches once again for “corfu holidays”, which was the query that began their journey.

Overall navigation patterns in travel remain complex, right up until the moment of purchase. Figure 7 illustrates that the online shopper is still consulting a wide range of sources during the final session when a transaction happens. Figure 7: Audience movement in the travel market before a purchase 17%

8%

13%

13%

Air Travel

Ground Transport

15%

19%

16%

10%

Lodging

16%

11%

8% 9%

7%

8%

12% 10%

Travel

11%

26%

2%

OTA 7%

10%

4%

7% Other Search

Tour Operator

5%

3%

Start points in black Google Traffic Base: Purchasers in the Travel market. Note: Arrows represent proportion of users who go from one site type to another in the purchase session. (Audience movements under 7% not stated)

In contrast, we see that a typical online car insurance purchase is much shorter and more direct. In this example path (figure 8, below), there are only two active days in the category, and the individual obtains fewer quotes prior to engaging in a conversion event. In aggregate the navigation paths insurance shoppers take (figure 9) appear much simpler than in travel. Figure 8: Car insurance path Quotation

Searches

Purchase

Car insurance Car insurance

DAY Sites Visited

Go Compare Money Supermarket

1

15

Go Compare Money Supermarket

Go Compare Money Supermarket Saga

Figure 9: Audience movement in the car insurance market before a purchase 40% 27% Aggregators 26%

Start points in black Google Traffic

Direct Insurers 10%

25%

Base: Purchasers in the Car Insurance market. Note: Arrows represent proportion of users who go from one site type to another in the purchase session. (Audience movements under 7% not stated)

The Role of Search 70% of consumers who purchase use search Across the sectors studied, over 70% of consumers who purchased online performed at least one relevant search during the observation period. Individuals who buy online are 30% more likely, on average, to have performed a search than individuals whose online research did not end with a conversion. These individuals are also more likely (by 17%) to have clicked on a sponsored search ad. Figure 10: Percent of consumers using search in their purchase journey

100%

92% 86%

90%

76%

80%

66%

70%

66%

60%

60%

56%

50% 40% 30% 20% 10% 0 Travel

Property

Apparel

Mobile Phones

Energy

Car Insurance

Loans

Base: Searchers who purchased as defined in each category

This analysis only shows correlation between searching, clicking on sponsored links and making a purchase, not causation. However, this does indicate that people who are in market to buy are more likely to click on ads to navigate more directly to relevant shopping related content.

Figure 11: Average number of searches per purchase 25

20

20

15

10 7

6

5

5 3

3

Loans

Car Insurance

0 Travel

Mobile Phones

Apparel

Energy

Base: Searchers who purchased as defined in each category

Although apparel shoppers spend the longest number of days deciding to buy, travel shoppers are by far and away the most prolific searchers during the research process. In fact, 1 in 10 travel researchers actually searched more than 50 times during the 3 month observation period.

Brand vs. Generic Searches Generics play a key role in the research process In some categories, such as property or travel, using generic queries (e.g. “property for sale oxford”) is almost ubiquitous (see figure 12). In other sectors, although the volume of generic activity is still substantial, a significant proportion of shoppers have good knowledge of the sector participants from prior experience or extensive advertising and navigate solely through brand terms (e.g. “Vodafone”). It should be noted however that although the reach of generic keywords may not match the aggregate reach of brand terms, the reach of generic keywords will usually be much greater than the reach of any individual brand’s search terms, as their share of voice will only be a fraction of the branded search total. Figure 12: Brand vs. generic searchers Only Generic search terms

Brand and Generic search terms

Only Brand search terms

Property

Travel

Loans

Apparel

Energy Mobile Phones Car Insurance 0%

10%

20%

30%

40%

50%

Base: Searchers who purchased as defined in each category

60%

70%

80%

90%

100%

Just as people who convert online are more likely to use search and sponsored links than non-converters, they are also 19% more likely, on average, to search on generic terms than those who end up just looking and/or converting offline. Figure 13: Generic to branded paths 1st Search

2nd Search

3rd Search X - 19%

X - 20%

Generic 29%

G - 53%

G - 60% B - 21% X - 12%

B - 27%

G - 19% B - 69%

Search X - 8% X - 17%

Brand 71%

G - 14%

G - 56% B - 37% X - 16%

B - 69%

G - 11% B - 73%

X: No further search

Base: Purchasers in the ‘Apparel’ Market

It is often assumed that consumers start their online journey with generic terms, and use branded terms as they move closer to the purchase event. However, these clickstream analyses do not fully back up this belief. There is a tendency for more generics to appear earlier in the process, but many paths do not follow the expected pattern.

Shoppers often switch between generic and brand keyword searches at different points in their journey. For example, figure 13 documents first three searches of apparel shoppers. 14% of apparel shoppers who start with a branded search follow on with a generic search, however 27% of people who start their journey with a generic search, continue up with a branded search. On average across categories, 48% of people switch between brand and generic at some point on the way to a conversion (see Figure 14). Figure 14: Percent of consumers who use both brand and generic terms in their purchase journey 90% 80%

79%

70%

59%

60%

56%

50%

50%

48% Average 39%

40%

34%

30%

21%

20% 10% 0 Travel

Property

Loans

Apparel

Base: Searchers who purchased as defined in each category

Energy

Mobile Phones

Car Insurance

Conclusions Consumer behaviour is intensive, complex and highly personal To achieve the maximum number of profitable sales or conversions advertisers need to understand the full value of all consumer online touch points and in particular the role of ‘assist’ clicks in the consumer path to conversion. Traditional marketing funnel models are less useful in the online environment, where smart, savvy consumers are in control. The Nielsen clickstream studies show the extent to which people shop around, and research over multiple sessions, entering multiple searches across both branded and generic keywords. Evidence from the studies calls into question traditional assumptions about the role of branded and generic keywords in the funnel as they are often both used at all stages in the online journey. To maximise sales online advertisers must take advantage of all consumer touch points, not just those that drive business on a last click basis. Multi-click journeys matter, they drive sales volume and value: they account for nearly half all paid search conversions and they are worth more, on average an additional 8% in basket value4. Generic search activity tends to happen earlier in the consumer journey, and therefore is likely to be undervalued by last click attribution models.

4. td Search weighted attribution paper, 2010

Next Steps Refine your approach to search to address all stages of the customer journey. Use the following process to help you actively manage and bid effectively on both converting and assist (sometimes known as “pathway”) keywords: Test and iterate • Examine your own customer journeys and assess areas for potential misattribution. Use tools such as Google’s Search Funnels or DoubleClick Exposure to Conversion reporting to help. • Measure the sensitivity of keyword performance to multiple attribution models to expose potentially undervalued keywords and identify what needs further validation. • Run experiments or regression modelling to validate what really works. • Feed test results back into the attribution models or CPA targets. • Continue to optimise. • Repeat! The test and iterate cycle should be a regular routine.

© Copyright 2011 Google. All rights reserved. Google and the Google logo are registered trademarks of Google Inc.

Google Clickstream whitepaper Services

No information is available for this page.

2MB Sizes 7 Downloads 316 Views

Recommend Documents

DLP Whitepaper Services
A description for this result is not available because of this site's robots.txtLearn more

Cloud Whitepaper Services
Disclaimer. Introduction. 1. The Canada PIPEDA. 1.1 Google Cloud and the Canada PIPEDA. 2 . Security and Trusted Infrastructure. 2.1 Google data centre infrastructure redundancy. 2.2 Google data centre security. 2.3 Data in transit. 2.3.1 Between a c

Identity whitepaper Services
A description for this result is not available because of this site's robots.txtLearn more

Whitepaper - STeX
If any two coins are listed on STeX, you can trade one against the other ... with some predicting a possible increase to a $200 billion market cap by the end ... Bitcoin today - while data from other sources, such as Coinbase and ARK ...... KeyCAPTCH

Whitepaper - STeX EXCHANGE
explosive growth in recent years that no existing crypto exchange is capable of ... Have you tried to switch between coins, following trends in emerging tokens ? ... STeX's own cloud cluster is physically hosted in many countries to prevent the .....

G DATA Whitepaper Vorlage
Unlike iOS or Windows Phone,. Android is an open source operating system. Because of this freedom, numerous app stores run by third-party providers have.

WhitePaper ENG.pdf
Transparency and openness are primarily due to the fact that all parties can access the same. information (on decentralized sections of the network).

WhitePaper ENG.pdf
In addition, tokens are needed to pay respondents. for their time to participate in a particular study. Page 3 of 22. WhitePaper ENG.pdf. WhitePaper ENG.pdf.

Whitepaper English.pdf
Social networks are internet sites formed by communities of individuals. who share interests or activities, since they may share ... Latin America and Spain and Google + was launched in June 2011. We plan to launch the first decentralized social ...