The New Search ROI: Measuring More than Conversion
Our moderator is Jeff Ferguson, SES Advisory Board & Senior Director, Online Marketing, Local.com. Our speakers are:
- Thomas Bindl, Founder & CEO, Refined Labs GmbH
- James Colborn, Director, Microsoft Advertising, Microsoft
- Aiko Yoshikawa, Sr. Product Manager, Convergence, Yahoo! Advertising Products Group
- Leigh McMillan, Senior Vice President & General Manager, Marchex Call Analytics
- Niraj Shah, Product Manager, Marin Software
Jeff welcomes the crowd after a nice outdoor lunch. He congratulates us for making it to these last-day sessions. Word. Jeff can remember a time when the biggest metric was hits. Then it was click-through rates. It finally got to a point where you could determine revenue — what a revelation! That was information you couldn’t get from other channels. As programs became more complex, the realization occurred that credit might be attributed to other channels.
The question became, how do you determine attribution? The first man in and the last man in is probably too simple. And what if you’ve got more than one thing to do on your site? All of these things need to be assigned to the right place, lest you cut something that’s not working. That’s the heart of the matter that will be discussed today.
Tom is up first. He’s going to focus on real data for better decisions. Even if you’re able to see the full ROI and channels involved, it’s still hard to believe all the data. In general, cookies get lost. How many days does a third-party cookie survive? After 30 days there is a loss of 21 percent of cookie tracking data. That was determined by setting cookies to users they could identify later, over three sites, covering different topics in different countries. That’s a lot of data that’s going to be wrong.
High AOV may result in less accurate tracking. AOV in travel industry increases over time. The data must be tracked from the first interest to the conversion. A possible customer journey might involve multiple channels that impact the conversion. Display, search and affiliate can all factor in. A quarter of all tracking data is wrong. 25.3 percent of conversions need 2+ clicks.
Attribution management: Why the last click shouldn’t get all the credit
- Try to understand cross-channel effects.
- The brand works better with generic terms.
- Define a value for all clicks that lead to a conversion.
- The longer the sales cycle, the more important attribution management is.
- In doubt credit the last click the most, but not everything.
Even the best technology needs help. There are ways to increase tracking accuracy:
- Store click IDs on your site and return on conversion
- Pass on order/client ID with every conversion
- Local shared objects (Flash cookies — often too slow for redirects)
- Ask your users for channel of origin
- Unique coupon codes
Online and offline rely on each other. How can you actually measure offline conversions?
- Coupon codes are often only generated online, but you can do it vice versa as well. A coupon code given in the store that buys accessories online will be connected to their original purchase.
- Discount IDs in ads and landing pages together with phone numbers.
- Check against prior behavior. A lot of people always take the same path to a site.
- Ask for original/influential channels. Ask the consumer — it actually works!
Jeff asks the panel if putting all consumers in one funnel is a detrimental way of doing business. Tom says that a lot of companies are putting money in channels and not paying attention to what’s happening. But with the technology to actually measure the results is foolish. You get more efficiency to all online marketing channels.
Niraj is next.
What are multiple conversion events?
- Visitor actions that indicate engagement
- May be directly revenue generating
- Maybe a sign of interest, future revenue
- Examples: store locator, sign up for newsletter, offline conversions (click to call), cross sells on thank you page, stages of a sales funnel (application submitted, approved application > registration)
Why use multiple conversion events?
- Track conversion events across paid search program.
- Analytics and reporting deliver insight to every level.
- Which keywords drive new users vs. existing users?
- Which creative phrases drive store locator but not checkouts
- Which ad groups have high abandonment rate?
- Assign value to conversion events separately.
- Give credit to keywords driving quality visits
- Tune bid setting to meet ever changing business goals
- Examples: month-end blitz to sign up new customers
- Monetization for cross sells changes
- Financial markets necessitate shift to affluent clients (vs. mainstream clients)
- When lag from click/revenue event is long (like 30 days)
- Option 1: Wait for latent revenue to trickle in and update bids accordingly
- Problem: bidding is reacting to changes 30 days late
- Option 2: Monetize intermediate events that have less lag.
- Advantage: bidding reacts to changes immediately.
How to choose multiple conversion events
- Match conversions events with business metrics and goals.
- Attribute values to events for bidding.
- Enables SEM program to easily adapt to business goals
- Cross sells, offline conversions, ensure keywords get credit
- New vs. existing customers
- Value new customers based on life time value: 2x initial order
- Value existing customers at 1/2 of order size
- High end vs. mainstream customers
- Value high-end customers at 100% of captured revenue
- Value mainstream customers at 75% of captured revenue
Conversion events aren’t always revenue events:
- Stages of a longer sales cycle
- Choose proxy value for intermediate stages
- Regularly review proxy value vs. actual revenues and adjust
- Store locator, newsletter sign up
- Correlate with offline and future conversations
- Qualify visits
- Useful for conversations with no explicit link online
Leigh is up next. She’s going to look at leveraging call tracking and analytics — the under utilized advertising intelligence. She asks how many people are with an agency. A few hands go up. How many use call tracking or are familiar with it? About 25 percent raise their hand.
Call tracking is analytics for calls rather than clicks or Web site activity. She’ll refer to it as call analytics. On average calls convert at ten times the rate of clicks. People get to talk through the service or there’s an urgent need.
Overall there’s 23 billion plus estimated calls driven from performance advertising and listings in the U.S. (search, print, yellow pages). That’s a robust source of data. Before, business had no cost-effective way to measure these calls — and the business that results from the calls. What campaign, what time of day, what keyword drove the highest value.
Now, a plethora of call data is available. It can be tracked by source, ad, campaign, keyword, time of day and call duration. With call duration, a baseline of calls helps you analyze and separate sales vs. service calls. Repeat calls, hang ups, missed calls, geo-location, recordings and caller info can also be collected.
Value Proposition of Call Analytics
- Prove ROI
- Capture offline conversions
- Gain advertising intelligence
- Improve sales staff performance
Call analytics measures the action most important to local and service-based businesses. One client saw that three times more conversions were counted because 66 percent of conversions were over the phone.
Who should use call analytics? Any service-based or local business and the agencies that work with them. Not only does the agency get credit but it also acts as intelligence to act on in the future.
- Home services
- Professional services
- Real estate
- Business service
How it works:
- Provision the tracking number.
- Place the number in an ad, site, landing page, etc. Any marketing where there’s a phone number.
- Leverage the data to optimize campaigns.
- Use local numbers whenever possible.
- Test and track what you really need.
- Call-optimize your landing pages.
- Listen to call recordings.
Evaluating call analytics providers:
- System uptime and infrastructure
- Ease of implementation
- Number of processes available
- [… I missed the rest!]
James will be our next presenter. He’ll look at the normal understanding of search and ROI and how you might also want to think about it, along with possible testing scenarios. How many people in the audience are buying search, display, content and image ads — a good number raised their hands for each one.
How do we think about search ROI today?
Consumer search > Click > Action = Desired ROI
At that action point, we’ve looked at multiple sources for getting the desired ROI. Interestingly, a lot of marketers stop their thinking process at the click. A lot of us also stop at the action point. But he’s going to focus on the first step. What actually prompted them to search for you in the first place?
How should we think about ROI?
Online advertising is accountable and measurable and provides some control. But the downfall is getting to the point of juggling a thousand plates on sticks at the same time. That’s where analytics and understanding come in. How many steps did it take to get to a conversion?
Is it worth trying to spend the right time trying to find the right mix?
Possible testing scenarios:
- Reality is that it’s far easier to do this with technology. But…
- If you don’t control a display budget, content/image ads are available
- Test medias together. You have the last-click data at hand
- Design campaigns to prompt cross-media connection
- Build a control campaign and test against it with multiple campaigns
Aiko will wrap up the presentations with her look at Yahoo! Conversions & Assist reports. Last click may not tell the full story. How do you value attribution across different marketing campaigns? There have been a lot of studies about display and search working best together. Yahoo’s Sponsored Search reports provide into the full story. Conversions = “last click”: prior to conversions. Assists = metric which gives credit to ads beyond the last clicked ads.
What are assists? They’re marketing activities that lead to conversions. Yahoo tracks up to 45 days before the conversion. Assists aid decision making. In the past, a campaign might be discontinued because it’s relatively high CPA. But high assists number would suggest manufacturer should maintain budget for campaign. The assumption was that there was causation involved, not just correlation. This kind of data is available to all Panama advertisers.