Keynote Roundtable: Why Does Search Get The Credit For Everything
I hope everyone had fun at the Google Dance last night (just try and take my light-up Google ring away from me. I’ll kill you.), but now it’s time to talk search again. It’s also time for a keynote roundtable. This is where smart people come to unleash search insight and talk over one another and Lisa’s finger’s fall off as she tries to keep up and decipher it all. I wonder who I pissed off to keep getting assigned the morning sessions while Susan and Virginia sleep? Oh yeah. I pissed off Susan. I hate her.
Kevin Ryan (SES, SEW) and Bill Hunt (Global Strategies International) are moderating Randy Peterson (Procter and Gamble), Mikel Chertudi (Omniture) and Sharon Gallacher (Neo@Ogilvy). Let’s do it.
Bill says when we look at the media spending mix we see lot of things going on. 96 percent of the budget goes to things other than search. For many large companies, this is the media mix. For many dotcoms, the media mix is exactly the opposite.
Bill says he gets to sit in a lot of meetings. He hears all sorts of silo’d marketing tactics. He hears that search is driving $250 million in revenue. Print is generating $100 million. Direct is generating $20 million. Then, at the end of the year, the CFO jumps up and says we only sold $100 million so someone is double counting their numbers. He wants to know who it is. It’s a matter of attribution.
Randy: When it comes to attribution, you have to think about it not just in terms of sales value or direct response, but also in terms of branding. You have to think of the brand value that’s created by search. When you build the brand there’s value associated with that.
Sharon: The branding aspect is very, very important. If you can make a 5 percent increase in branding, then that’s a big win.
Mikkel: In addition to the sales and orders and lifetime value, it’s important to realize that when you take credit for sales, you also have to take credit for the cost.
Bill talks about the “last click” concept.
Mikkel: There’s a display team responsible for all the banners. A user clicks on that banner ad and then they click on the email that comes in their inbox. Then they go and do a search in Google on the keyword [auto insurance]. They finally click through and fill out a lead generation form for a free quote. Many companies are just giving credit for that last action. What should be happening is that they should be measuring and accrediting all the actions that came before. What started it, what came between, and where did it end? You want to create a baseline on all of those. You can’t ignore the other channels.
Kevin: Are people understanding that search is reactive in a lot of respects and we have to have to figure out how to tie it back?
Mikkel: We see that there’s a lot of brand being driven by TV and display advertising. We see search gets credit for everything. There’s also another issue – cookies and cookie settings. Sometimes cookies expire in 7 days or 90 days or never. If you click on a banner with a cookie expiration date set for 30 days and then you click on an email link 31 days later, the display ad never gets credit.
Randy adds that comScore has found that 30 percent of people delete their cookies once a month.
Bill: What are some of the things that can go wrong if you don’t attribute properly?
Mikkel: He worked in display advertising about 8 years ago for the Web’s 2nd largest publisher site. They had two separate installations of an ad server, which caused them to double count everything. They were a direct response ROI-based business and found that they were trying to tie their cost-per-customer down a very specific number. When he would buy ads on a CPM basis they found that they had overspent to the tune of about 20 percent aka millions of dollars! That’s because they were over-assigning credit to the search channel and to the display channel.
Bill: This whole idea of attribution leads to this unhealthy tension between search and traditional advertising folk. People look at search as the Marsha Brady of marketing. It can lead to turf warfare and unhealthy tension. Are you guys seeing this?
Randy: It’s still early on for us. Search is still somewhat in its infancy. They attributed too much value to search and got unhealthy tension. They went back and assigned conversion goals to activities happening on the Web site so they could associate values. They’re building a model to do that and that has helped a lot with the brands that are spending big money on search.
Bill: Talk about this idea of misguided investment.
Sharon: It can take you a long time to ever find out that you’ve invested poorly, and that’s scary. If you’re off and you keep being off, you just keep building up.
Kevin: Is there a lot of resentment internally?
Randy: There’s still some of that, yeah. The strength of personalities in large companies makes things happen. The way around that is through data. Data beats personality.
Sharon: We’re seeing that in the rise of display and search. The TV budget was 3x the search budget for a long time. Having the data just obviates that argument.
Mikkel: If you have the data you can outmaneuver the strongest will in the office. Hopefully all of us in the room have a bit of a competitive advantage because we could manipulate that data. We’re not doing that, of course.
Bill: We want to talk about some of these metrics. We may be measuring the wrong things. Attributing results to the wrong things. What are you measuring? What are your pre-metrics (impressions, clicks, CPC) and post-metrics (orders, sales, lifetime value)?
Randy: You do a bunch of marketing activity and bring some value. Some of the metrics we think about are anything we can collect from analytics. That’s the broad category. It includes clicks, data we can pull from the engines, ad serving data, etc. We also look at all offline media. You have to put all that together. In the end, you also have to look at sales, searches on an ecommerce site, a major brand value, etc.
Mikkel: He likes Randy’s buckets.
Bill: We had a client who found that about 65 percent of their traffic was coming from organic, and 40 percent of that number was engaging with the live person execution they had on the site. They ran a report and decided that they should decrease the amount of search. They couldn’t tie back any sales from the traffic from search. Many people overwrite the conversion source. If someone comes in from search and then talks to a sales person, the sale’s person’s code gets put in for the conversion, not search. Can you talk about that?
Mikkel: This is a really important issue. You don’t want to overwrite objectives. Your objective is to bring in visitors to your Web site. Your site objective is to drive conversions. The remarket objective is to get repeat orders and cross sell. It’s critical not to overwrite because you’re overwriting your costs for those as well. You can’t do full blown ROI analysis that way.
Randy: There’s a quote that says all models are wrong, some are useful. In your case, the model was wrong. What you did was made the model more correct and then it became useful for you. That’s a good way to think about this.
Mikkel: The takeaway is don’t overwrite. Keep them all in the database.
Bill: Most attribute success to the last action and all decisions are based on these metrics. Is this wrong?
[A few people in the audience say it's wrong. A few more say it's something we have to live with. And everyone else is just hung over from the Google party.]
Bill: Is it possible to set up a fair and balance program that gives credit where credit is due? He says it is.
Mikkel: Sort campaigns by an array and look at which was the first campaign that brought this account into your database. What was the last one? What is the secret sauce that worked to get that campaign into your database? Once they start to repeat purchase, you want to track that independently as well. What was the last campaign before the repeat purchase? Store and measure all the data.
Bill: If we solve the technology problem, can we really change the hearts and minds of fellow marketers?
Sharon says if we solve the tech problem, we’ll solve the attribution problem somewhat. We’ll never get it solved 100 percent but we can find a better balance. There are different measurements we can use. They’re not all the same thing. Maybe we just have to daisy chain those pieces of research and make assumptions.
Randy: He agrees with Sharon. Data is so important. It’s friendly in this situation. It brings the discussion down to something that’s less emotional. That’s how you win over the hearts and minds of marketers.
Mikkel: Marketers find peace in the numbers. He can justify his job in the numbers. It’s important to be data-driven marketers on the metrics that matter.
Tips for Success
- Cover the basics: Do you know where this data is?
- Define your goals effectively: What are you trying to manage.
- Define the value of the action
- Know your bounce rate: We hear about the 5 percent click rate. What happened to the other 95 percent?
- Look for tools or indicators that you have in other media to help indicate impact and success.
- Create an attribution model, using data available.
- Get comfortable with not having a perfect answer.
- Standardize tracking and marketing applications on one system.
- Measure all views: 1st, last and shared.
- Don’t overwrite metrics and objectives: Track them separately.