Tag : big-data

post image

235: What do you do when data scares you?

I received a Facebook Message early on a Saturday morning. It was from an auctioneer who was getting angry comments on the Facebook ads I’d created for him. Rather than just not clicking on the ads, people complained that they wouldn’t click on an ad that didn’t show the location of the items for sale or the date of the auction. They’d click to comment and write—but not to go to the auction catalog.

The auctioneer was concerned he was losing buyers. He wasn’t. (He told me after the auction that he thinks he set new records for this auction’s equipment.) But it was a big auction for an enterprise-level seller. I’ll tell you why I didn’t think he would lose a single buyer in a minute; but first, you need to know that the date and location were in every ad I built for him. 

Sort of.

Concerning Facebook Data

Almost every Facebook ad I create for auctions uses variable data headlines and subheads. I feed Facebook up to five different headlines and include up to five supporting subheads for each ad. For the ads in question, the location was one of those headline options; and the date was in three of the five subhead options.

Facebook’s variable content tool tries all of the options on each of the audiences the ads target. (On this campaign, we had six different audiences in the first round of ads.) As prospects respond to some content more than others, Facebook adjusts the distribution of those options daily. In this way, the headlines and subheads that work best gradually get the most distribution so that your best-performing bait is what is on your hook. And you might have different winning content in one ad than you do in another due to the preferences of each ad’s respective audience.

So what had happened in the campaign in question is that the vast majority of respondents had clicked on ads whose headlines and subheads contained something other than the date and location of this auction. We were getting a fantastic cost per click and lots of traffic to the auction catalog, but the auctioneer asked me to shut down the ads, put the auction location in the non-variable text where I had put the sales copy, and restart the ads. He told me he was willing to have less-efficient traffic as long as the comments didn’t scare his seller representative.

I get it. I look at hard data daily and often don’t follow what it shows to be the reality. That’s true in my business and my personal life. Trusting a machine, an algorithm, or artificial intelligence is not natural for human beings. That’s one of the primary tensions that inventors, engineers, and technology executives face every day. 

When Facebook Data Scares You

My client wasn’t losing a single bidder. Anyone truly interested in what he was selling would’ve clicked on the ad to see more photos, more asset information, and auction details like ending time and pickup location. People who complain on a Facebook ad instead of just scrolling by it have an axe to grind and just need an outlet for that rage. If the location and auction end time would’ve been in the ad, they would’ve found something else to complain about in the catalog or auction terms.

Still, the thousands of website visits those ads had efficiently generated couldn’t reach the volume of a couple of angry comments. That’s also part of human nature. We’re humans, and so are our sellers. Humans unevenly trust automation, algorithms, and other perceptive technology. So, I don’t fault the auctioneer for moving away from data-driven content to what would assuage his client; and I immediately made that change to his ads.

What to do with Facebook data

Here’s how I try to convince auctioneers to trust the expressed preferences of the buying public—particularly in terms of variable data in Facebook ads. Look at it like a poll. If 90% of your bidders said they wanted your advertising to use a particular headline, would you switch and use it? If 80% of the people who came to your website said certain information wasn’t important to them, would you still make it a headline? If after seeing these patterns over hundreds of auctions, you refused to adapt to the buying culture, I have one more question for you. What’s more important to you: your comfort or your commission?


Images purchased from iStockPhoto.com

post image

215: 3 Lists Every Auctioneer Wants But Nobody Can Buy

With the rise of big data, entrepreneurs have grown to assume that just about any kind of data is available to purchase. In many cases, that valuable demographic and purchase history is more robust than most of us would ever need. The problem is that it exists in proprietary databases. Those black boxes at best are available for blindfolded lease and at worst compete against us.

Even before Facebook’s ubiquity and Amazon’s dominance, though, auctioneers asked me for the same prospect lists they still assume I can procure. When I tell you that these lists get requested often, I mean every month—sometimes weekly. I still encounter surprise and maybe even disappointment when I can’t deliver them. To save us both from an awkward conversation later, I’ll just explain them here for you.

Real Estate Investors

As of July 19, 2019, we haven’t been able use Facebook’s real estate investor interests to target real estate ads. (I assume that’s to comply with HUD anti-discrimination regulations.) Even before that, we couldn’t target actual investors—only people whose Facebook habits showed an interest in investing. As of right now, we can’t purchase a list of individual people who invest in real estate. We can target companies whose standard industrial classification (SIC) falls under real estate development, management, or brokerage. We can ask for highest-known executives in those firms and phone numbers and legal/opt-in emails where available. In certain databases, we can pull people who own homes but don’t live in them. A list broker can sort that by net worth and/or annual household income. It’s a long shot, but that’s currently our best option.

Land (or Any Asset Category) Buyers

There isn’t a commercial source for those records. Landowners, yes. Land buyers, no. Independent auction companies should have a list of past bidders and buyers from auctions—hopefully sortable by asset category. Those lists should be queryable in order to pull only past buyers and/or bidders. Until you have critical mass, you can use that list only for direct mail. Once you do get several hundred buyers and/or bidders, you can use Facebook’s lookalike audience tool to find similar prospects. Until then, one option would be to partner with a joint venture company who does have a list large enough for lookalike audience potential. If you drive the Facebook traffic to your website, you can then start using your Facebook pixel data to create a prospect base. Using information sign-up forms and bidder registrations on those joint-venture auctions, those who do respond can become seeds to start or accelerate your own list. 

People Who Want [fill in the blank]

This is the one that makes me audibly laugh, when I open the email. Men have joked my entire life that they never know what the women in their life want. If that’s true, that takes out 50% of the people whose wants we can capture and query. Even if it’s not true, we would need a Minority Report-style system to mine this desire data. Do we really want companies to know our private thoughts? We can hit this goal obliquely through a list of past bidders or buyers on similar items, assuming their need or want wasn’t satiated since that last auction. We can purchase lists of some interests and purchase history, and we can target non-real estate ads to even more interests on Facebook directly. But the best we can do is get adjacent to wants and let artificial intelligence engines do their magic. There’s no list for people who want a mower or a Coca Cola sign or 20 acres.

That said, not having these lists and even not being able to acquire these lists doesn’t mean our advertising has to be ineffective or inefficient. It just means we need to do more homework, more experimenting, more tracking, more data analysis. Yes, that’s more work. Yes, that’s a completely different skill set than a silky auction chant or a well-executed sales pitch. Yes, it’s not the way your dad did it.

Over the next two decades, conglomerates and aggregate sites are going to put hundreds of bid callers and even auction marketers out of business. They’re going to pay people to do this data curation work. Instead of trying to buy data, they’re going to mine their own. Those of us who follow their example will most likely be the ones in 2035 who are still advertising auctions at all. 

Because I don’t manage websites, host online bidding, or run auction software for my clients, I will be dependent on auction professionals like you to procure, store, and query this data. Thankfully, most current auction software and platforms make this doable, if not easy. That gives me hope, and I’m glad to be in your good hands.

Stock images purchased from iStockPhoto.com

post image

Reach the Bidders You Didn’t Know You Were Missing

There’s a sneaking suspicion in many auction marketers—and definitely in their sellers. We wonder if there was a stone unturned, a motivated bidder that wasn’t reached by our advertising.

Did we cast a big enough or tight enough net?

Missing Bidders PosterWhat people weren’t in our mailing list broker’s database?
Who didn’t read the newspaper during the weeks prior to the auction?
Who didn’t drive past our sign out on the highway?
Did any emails go unopened or straight to junk folders?
Did we choose the right demographic selectors on Facebook?

The auction community prides itself in bringing the whole market to bear on an asset at once. We tell potential sellers that we’ll deliver true market value. We rightly trumpet our concentrated advertising campaigns.

Still, there’s that whisper, that gnawing question—especially when the auction price is low and even more so when it was an absolute auction. Did we find everybody?

One of the biggest developments in advertising over the past couple of years has been a partial solution to that mystery. This development has made mailing lists more powerful, web traffic more valuable, and Facebook just short of necessary for finding buyers.

Big Data for Small Businesses

In addition to the vast amount of data users give Facebook about themselves, Facebook also buys data from outside sources and matches that information to its user base. Bank and mortgage lender records. Vehicle ownership. Purchase histories. Web site visits. As a result, this data gets woven into an astounding web of connected dots. Using advanced algorithms, Facebook can then match people with common denominators.

So, after you find the people you think are likely buyers, Facebook can find people who look just like your intended audience. With Facebook’s Lookalike Audience tool, both purchased lists and in-house lists can be matched with people just like them for use in Facebook ads.

With the free Facebook Pixel code installed on your website, you can also now direct Facebook ads to people who recently visited your auction’s page or the page of a similar auction on your site. Then, with the Lookalike Audience tool, you can advertise to people who look just like the people who came to your website.

Over the course of your advertising campaign, as more and more people view your auction’s page on your site, Facebook can learn more and more about the people coming to your site and hone the audience of your Facebook ads.

Facebook Loop

So, whether you start with just a Facebook list of demographics [B] or if you upload lists to Facebook [A], you can create a set of ads that learn and improve their effectiveness over time. You can access an automated database that keeps getting more robust. Your advertising can reach people in the cracks between the groups of people you can find yourself.

An Impressive But Imperfect Solution

Is this Facebook solution circle a silver bullet? No. This is just one medium that reaches less than 80% of the population. Does this mean you’ll definitely find more and better bidders? No, but it’s a superlative start. It’s a more robust solution than what you’ve got now.

Could this concept confront our ignorance? Absolutely.

Recently, I’ve noticed that several of my clients’ Lookalike Audience ads have significantly outperformed not only their uploaded lists but also the Facebook audiences built with the demographic selectors we chose for prospective buyers. In other words, Facebook knew who would visit these websites better than I or my clients did. For the decades of auction marketing experience between all of us, that’s humbling.

It’s also exciting. Now, our lists of past bidders and email subscribers are more valuable. Now, our web traffic can be more meaningful. Now, purchased lists don’t have to be exhaustive. We just need to find a critical mass to get the ball rolling.

Now, we can find the people we weren’t finding—even with our best laid plans.

Illustration built by request from Fiverr.com
Stock images purchased from iStockPhoto.com