Determining the value of Facebook for your business– mistakes you’re probably making

A Rice University study last week claims that Facebook is an effective marketing tool because of these stats, comparing the Facebook fans versus their general customer base:

  • Made 36 percent more visits to DG’s stores each month.
  • Spent 45 percent more of their eating-out dollars at DG.
  • Spent 33 percent more at DG’s stores.
  • Had 14 percent higher emotional attachment to the DG brand.
  • Had 41 percent greater psychological loyalty toward DG.

At first, this sounds great– these Facebook users are more loyal and spend more money with this restaurant chain.  But as you learn in first semester college statistics– correlation is not causation.  The folks who fanned your page are likely already your strongest supporters– it wasn’t because they fanned your page that they all of a sudden started spending more money at your restaurants.  

Consider the fact that the percentage of people who visit the emergency room and die is higher than the percentage of random people who don’t go to hospitals, yet still die.  Clearly, there is a correlation between going to the hospital and dying, but that doesn’t mean if you’re sick or injured that you should avoid the hospital.   People who go to the hospital are already sick, so it’s not fair to compare them against the normal healthy population.

In the same way, of course the first fans to your Facebook page are not representative of your general customer base.  It’s your friends, coworkers, and folks closer to you.  OF COURSE these fans are strong supporters, especially since this particular study had a tiny sample size.

Does that mean you shouldn’t do Facebook or that fans are of low value?  Quite the opposite.  It means that after your initial wave of fans (which are your most loyal supporters), then you begin to attract the mainstream fans of your product or service.  These are the people who may have heard of you, but aren’t raving fans yet.  And it’s the viral loop of friends telling friends who then tell other friends– all the while spreading via the wall and word of mouth– that you get MASSIVE ROI.

Facebook is great for demand creation– to help spread word about your brand through folks who already love you.  Demand collection is showing ads to people who are about to buy.  That’s where social media and search work well together.  The AIDA funnel (Awareness, Interest, Desire, Action) shows that social media owns the first 3 phases and that search owns the last.  You have to generate awareness of your brand and also be there when they are ready to finally buy via Google.  

Thus, your Facebook ads are not “better” than paid search, nor is your Facebook Fan Page a better opportunity than your regular website.  You might as well argue that the steering wheel in a car is more important than the gas pedal.  You need both– it’s not an either or situation where they work together.

Stay tuned tomorrow for when we quantify the value of  your Facebook presence, starting first with determining “The Gap”– the difference between your offline brand strength and your online presence.

Ranking in Google News is easy– even for competitive terms

What’s the secret?  Just submit to about 20 free press release sites.  If you want to pay for PR Wire or PR Web, that’s fine.  Might save you some time. While it’s great to be able to rank for Facebook advertising and local online marketing, the truth is that these news results don’t provide much traffic.  You might be sitting on the first page of Google News for a day before you slide down and are replaced by something more recent.  

Perhaps with Google Caffeine and the greater emphasis on the real-time web, news-like results will become more prominent.  Imagine Google copying Facebook, where the search results page is essentially the Facebook News Feed.

How to use Google Conversion Optimizer in just 3 minutes

New to Conversion Optimizer or perhaps puzzled on how to use it properly?  Give me 3 minutes right now I’ll tell you what you need to know– this should save you a ton of time with bid optimization and, more importantly, increase your PPC profits.

First off, Google’s Campaign Optimizer works at the campaign level.  Find it in your campaign settings here (the tab in green):

Then scroll down a bit further until you get to bidding options and choose “Focus on conversions”.  A big yellow box opens up:

If you don’t see that option available to you, it means that you don’t have enough conversions for the system to be able to optimize.  Google has lowered the stated number of conversions needed to activate this option, though I believe that you should have at least 50 conversion per month to make it even worthwhile. 

You’ll notice that the recommended CPA target is far higher than the actual CPA you’ve observed.  Go ahead and accept that default or perhaps go 10% less.  Don’t be alarmed, this is not Google attempting to get you to spend more.  Remember that the max CPA is not your actual CPA– somewhat like your max CPC is not your actual CPC.  There is one key difference in that your actual CPAs may sometimes be HIGHER than your target.  This often happens the first few days while their system is learning how to optimize bids. Don’t freak out.

Change the settings for each campaign and you’ve now saved yourself a ton of wasted bids, avoided needing to spend money on fancy bid optimization software, and increased your profit with a lot less effort.

Here are some additional tips:

  • Separate your search and content campaigns– Yes, you’ve heard this before.  However, you might not realize another reason is that view-through conversions are not taken into account for Conversion Optimizer.  If you haven’t seen the power of a view-through conversion on a banner ad, it’s a conversion that happened where the user saw the ad, but didn’t click on it.  For every direct conversion (last click attribution), you may see an additional 10 to 20 view through conversions, which means your CPA could be 1/10th of what is being reported. Because Conversion Optimizer doesn’t take into account view-through conversions, you’ll want to set the CPA targets a bit higher on content campaigns that have banner ads.
  • When you use Conversion Optimizer, your dayparting goes out the window.  See screenshot below, where there is gray text that says “ad scheduling bid modifications are unavailable with focus on conversions”.  I think that it’s silly to not be allowed both, since perhaps your conversion is based on a call– and your call center is open only during certain hours.  Perhaps Google’s ad server, as sophisticated as it is, can’t handle this yet:


We’re told by the Google folks that Conversion Optimizer does take into account day of week into the bidding.  However, we’ve not seen that to be true.  For example, if you run online dating campaigns, you know that your conversions are higher between 5 pm and 10 pm local time, as well as on Friday and Saturday nights.  We have not noticed Google’s Conversion Optimizer adjust to bid higher during those peak conversion periods.  Instead, we see the CPAs follow a “sine wave” curve week after week.

As for whether Google can optimize based on placement on the Content Network, we are not sure, but would love to hear your thoughts.

A final tip for you:  Don’t let Conversion Optimizer cause you to be lazy.  In the past, you could literally upload a few hundred thousand keywords in bulk— not organized in any way and with a list of just garbage terms– and then count on Conversion Optimizer to automatically sort through them to find the terms that converted.  That used to be quite effective.  Nowadays, if you do that, you’ll likely get hit with Quality Score penalties, even if you have an aged account.

An example of where you want to group terms properly is on brand bidding.  Your own name will convert better than generic terms, largely because it’s actually navigation as opposed to true search (for people who can’t use the address bar and use the search box instead to get around the web).  Your branded terms on PPC are also stealing from your organic traffic, which also causes you to overstate the CPA if you’re going off the default last-click attribution that Google uses.  The answer isn’t to avoid brand bidding, but to put brand terms into their own campaign and not even have them on a CPA target at all.  Your max CPC on brand terms should be based on the cannibalization rate of organic search- more on how to do that another day.

Are our 3 minutes up?  I hope you got some value out of this article.

4 steps to make a KILLER Infographic and drive natural inbound links

Gather some interesting statistics and then make a chart out of it– and you’ve got yourself an Infographic.  Examples are showing a map of the world and income for each country displayed by green bars.  Or perhaps it’s the price of gasoline over the last 2 years graphed against milk prices to show some interesting trend.  The goal of an InfoGraphic is to visually stimulate people with statistics and get them to tell their friends or blog about it.  A few days ago, I saw an InfoGraphic showing the percentage of times people tweet after having sex. Certainly that drew some attention, although I’m not sure how accurate their methodology is.  It’s not as if you can set up hidden cameras in bedrooms across the world to measure this.  The percentage is 36% in that study, by the way, if you’re wondering.

There are 4 steps to getting this done:

  1. Get the raw data-– some folks will do a survey, which is easy enough to do via Facebook and twitter.  Just do a web search and you’ll see a number of sites that allow you to create free or inexpensive polls. The plus– polls are easy and you can get data fast.  The cons– massive sampling bias, as you’re not getting a random sample, plus your sample size is likely too small to have statistical significance.  The best results are where you can scrape from a large dataset– but this may require you to spend money to get that data via a gnip, addtoit, or other service. Some are free and some require no programming.
  2. Crunch it– slice, dice, and manipulate the data.  Some Excel wizardry– or SQL queries if you have a larger data set and need to load it into a database– and you have your aggregates.  Group by keyword, geography, type of user, or other attribute.
  3. Make an image— Easiest and most common tactic is to do a map overlay. For example, look at the beer drinkers in America by state.  Or you can do something silly, such as The Onion’s mockery of MySpace’s privacy. Not a great designer?  Just find someone on rentacoder or odesk for $100, telling them what imagery to imitate.  If you’re doing an Infographic on how many cups of coffee Americans drink, broken out by state.
  4. Promote the heck out of it: If you’ve done the previous 3 steps right, you’ll go viral. Make sure that nobody can nail you on poor methodology– bad sampling, incorrect assumptions, or other flaws in your research. Blog about it, get your friends to Digg it, post it on your Facebook and Twitter.  A good headline here can make or break the result. 

If you generate enough controversy or have something hilarious and/or interesting, then watch this go viral– and the links to your site will start flowing.  Brent Csutoras, the best social media linkbuilder on the planet (in my opinion), told me that he can sometimes get tens of thousands of links from a single post.  That includes a smattering of PR7 and PR8 links– if you hit a home run. But perhaps a typical viral campaign will generate just a few hundred links– you never really know. 

Now compare those results against trying to buy links or reaching out to bloggers one at a time. Even if you could buy links without getting in trouble, what would the comparable cost be?  Matt Cutts, the Google spokesperson for SEO, says that this link building methodology is completely white hat and legitimate.

So what are you waiting for?  What interesting factoids and tidbits can you assemble for the website that you’re trying to promote?

Google’s fabulous Christmas Gift to BlitzMetrics

Sure, people are complaining about how Google gave $20MM to charities on our behalf, instead of giving us trinkets this year– whether they be a photo viewer, Flip video, or whatever.  But Google gave us some stuff that was far better than something you could just buy yourself at Best Buy. Look at these campaign features:
  • Sitelinks in PPC ads: I won’t cover the ins and outs of AdWords sitelinks, but suffice it to say that this feature increases your CTR.  You can specify up to 10 links, but how many they actually show is usually just 3 or 4.
  • Automatic matching: Scary, but Google will choose other keywords for you to show up on. Do you remember a couple years ago when Yahoo! just started tweaking people’s campaigns without telling them, causing advertisers to piss away money on irrelevant keywords– then get irate? Yahoo had the right idea, but was perhaps premature in the game of trying to simplify life for advertisers, as it was too big a step at once and didn’t actually improve performance.  The same naysayers of “expanded broad match” are complaining about Google’s move here. However, I don’t think it’s all that bad. 

Some considerations on Automatic Matching:

Yes, it encourages laziness. And will put in crappy words if your landing pages (which they crawl for keywords) stink.  If you have a great campaign already with tight ad groups and strong negatives, it’s unlikely they’ll be able to add much traffic. Otherwise, there is risk that your budget will get maxed out quickly.

You can still run keyword performance reports to determine exactly what keyword triggered your ad.  So in the same way that you run placement reports for content to weed out junky sites you’re showing ads on, you run a keyword performance report to see what new keywords you’re matching on, so you can then stick them in your campaigns or just negative them out. At the bottom of the ad group, they even break out the automatic match traffic, which is incremental to your regular traffic and doesn’t affect the rest of your bids, positions, or Quality Scores.

They do take into account location: Yes, so if you’ve got a franchised account with multiple locations, they will be smart enough to choose geo-modified keywords and pair them with the right ad and landing page. However, in Google’s semi-helpful FAQ on automatic matching, they caveat this by saying it works only when ALL the keywords in an ad groups are geo-modified.

You should try it-– but don’t just assume Google’s recommendations are going to automatically improve your campaigns.  It will certainly improve Google’s profitability, but not your conversions.  Unlike most tin foil hatters, I consider this feature– and anything else they roll out– good, if you watch carefully what you’re doing.  It’s effectively like the keyword opportunities tool, where they propose new keywords for you– except they’re automatically jammed into your campaigns.  Note that for this to work, every keyword in your ad group must be location-specific.

So between the improvements in Conversion Optimizer, sitelinks (beta only), automatic matching (beta only), mobile ads (who knows), and new local options, Google has given us a ton of new gifts for Christmas.  I’m surprised so few people are talking about it– perhaps because Google themselves doesn’t talk about it, instead just quietly releasing new features into the wild.

Google AdWords Bid Simulator– observations

 The new Bid Simulator, which you’ll see in the keywords tab, forecasts how many clicks and impressions you’ll see at different bid levels.  It only shows the simulator for some of the keywords– not sure what logic is used to choose which ones.  It certainly isn’t search volume, since some of the lowest volume terms in our campaigns have the Bid Simulator icon.

Important to note that the Google AdWords Bid Simulator doesn’t predict the future— rather, it estimates what would have happened in the last week had everything else stayed the same except for your bid.  Google explains it here.

In this first screenshot, you see that we’d get nearly the same traffic at any bid price for this keyword.  Note that the estimates impressions is the same.  By bidding higher, we move to a better position.  We are currently bidding $3 a click to get 63 clicks, but if we drop our bids to $1.01 (a third the price), we get only 3 clicks less (a 5% reduction).  Thus, a 200% bid drop for only a 5% click volume drop– for you economics students out there, that’s significantly inelastic.

Why?  At some point, you’re already in first position, so bidding higher won’t matter.  Google’s AdWords bidding auction, as clearly explained by Hal Varian (Google’s Chief Economist and the author of my undergrad Econ textbooks) in this video, shows that our price is based upon an increment of the next highest ranked bidder and your Quality Score.  P1 = B2Q2/Q1.  In other words, the price you pay to be in position 1 (P1) is the AdRank of the advertiser in position 2 (Bid of Advertiser 2 x the Quality Score of Advertiser 2)– then divided by your Quality Score.

On high volume, highly competitive terms, you would expect to see a more gradual fall-off in this bid curve.  Normally, you’ll get hit with the double whammy of more clicks at a higher cost per click— if clicks and CPC are both increasing by 50%, then you’re hit with an overall cost increase of 1.5 squared, which is 225%.

What I think Google may have neglected to include in their Bid Simulator is the impact of sitelinks in position 1, which give a tremendous boost to position 1 advertisers.  This estimation would be hard to do, given that the feature is not in wide release– though BlitzMetrics is fortunate enough to have enough accounts that we have a few of them with this beta feature enabled.

Here is an example of a Bid Simulator shown when there is hardly any data— on a tail term with phrase match on.  Were there enough ad data, we’d be able to calculate the Incremental Cost per Click (ICC).  Don’t make fun– the ICC is the term that Google uses to describe this concept– namely, the additional price you pay for incremental clicks, measured by the change in cost divided by the change in clicks.  If you’re bidding up, your ICC is significantly higher than your average CPC, which averages in all lower click costs.

Overall, I find the AdWords Bid Simulator partially helpful.  Looking at average position I believe is nearly as good, since Google won’t tell you the bids and Quality Scores of the other advertisers anyway. It also doesn’t appear to take into account the effects of negative keywords, dayparting, geo-targeting, and other settings (at least based on their internal PowerPoint showing the actual data used versus estimated for Bid Simulator.

The next step for Google is to make recommendations on how to increase profit based on the simulation.  If I raise my bid to get more traffic, then I’m also decreasing my profit per click.  Google should tell me what bid makes the most of this trade-off.  Currently, all Google’s recommendations seem to be to increase bids, add keywords, and increase budget, so not sure if they’re going to do this any time soon.

However, it is true that if you use Conversion Optimizer, that you can’t use Bid Simulator and that using Conversion Optimizer is effectively maximizing profit if you know the right CPA target.  Love to hear anyone’s experiences here with Bid Simulator, ICC, Conversion Optimizer, and other estimation tools.