Your Sellers Are Not Using the Data You Bought
Between a third and a half of purchased software goes unused, with analytical tools among the worst offenders. Our industry’s fix was to hire a translator instead of opening the audience data black box.
Somewhere in your operating budget is a six- or seven-figure line for audience data. You approved it. It produces real value. And the number of people in your building who can actually operate it is probably in the low single digits.
I should be careful here, because I sold those subscriptions. I ran the commercial side of Scarborough for a decade and was president of Simmons before this. The clients who got the most out of what I sold them all did the same thing: they hired someone whose job was to run it. The data was good and the value was real. I did not read the hire as a symptom at the time.
What it was, was a ceiling. Adoption stopped at however many specialists a company could justify, and the seller on the phone with a furniture retailer got whatever the specialist had time to hand over.
The broader software problem is measurable. Zylo, which tracks more than 40 million software licenses and $75 billion in spend, reported this year that organizations leave 36 percent of their licenses unused. Nexthink, measuring actual usage across more than six million customer environments in an earlier study, put the figure closer to half. Its data also showed a revealing divide: Slack, Teams and Zoom ran above 50 percent utilization, while Tableau and Spotfire sat below 15 percent. The tools people talk through get opened. The tools people analyze with do not.
I could not find a comparable study measuring how broadly broadcasters, publishers and agencies actually use the audience research they buy. But our workflows are unusually good at creating the same problem.
Specialists were the right answer to the wrong question
Salesforce surveyed just over 4,000 sales professionals last autumn and found 40 percent of a rep’s time going to selling and 60 percent to everything else. In our business, a chunk of that 60 is waiting. The seller asks the specialist, the specialist gets to it, and the answer arrives after the advertiser has decided.
Hiring the specialist made the tool usable. It also made the tool a bottleneck because every question in the building now routes through the same two people. Multiply that across a sales floor and you are not looking at a research problem. You are looking at a revenue-per-seller problem that nobody has been assigned.
The consequences of inaccessible audience data are already visible elsewhere in media. When the Association of National Advertisers sought log-level data for its programmatic media supply-chain study, 67 member companies initially expressed interest. Only 21 ultimately participated, with legal and data-access hurdles preventing many others from providing the required information.
The lesson isn’t limited to programmatic advertising. Buyers increasingly want to know where a number came from, what assumptions sit behind it and whether it can be verified. A research specialist may be able to answer those questions eventually. The seller sitting across from the advertiser increasingly needs to answer them in the meeting.
What the conversation could sound like instead
Consider a furniture retailer in Kansas City. Audience data for that market in May shows about 153,000 households planning to spend $1,000 or more on furniture in the next twelve months, out of roughly 1.05 million in the market. That is an intent-level target rather than a demographic shell.
The more useful figure is where those buyers go for information. Furniture store websites lead the market at about 252,000 households. Social media runs under a third of that. Read as a media plan, that argues the retailer’s own site is where the decision gets made, and the job the local properties are hired for is to move a qualified buyer toward it. It also tells the retailer where not to spend, which is usually the part that earns the next conversation.
That is a different sale. Not reach against a demographic, but something an advertiser did not know about their own customer and a route to that person. None of it requires new data. It requires a seller who can reach the data on a Tuesday without filing a request.
The lever is cycle time, and it is an architecture problem
I have reached for the wrong lever here myself. When the data goes unused, the instinct is to buy better data, add more of it or train people on the tool nobody opens. We tried versions of all three. None of them touched the constraint.
Audience intelligence was designed as a periodic deliverable: commissioned, fielded, delivered, shelved and pulled down when somebody needs a number. Every step assumes the question arrives on a schedule and that a trained person will be the one asking it. Neither assumption holds when a seller is sitting with an advertiser trying to answer an unexpected question.
What replaces that model has to behave less like a study and more like a system a seller can query directly, on the day the question arrives, in the market the advertiser named. There is more than one way to build that. What is not optional is reducing the distance between the question and the answer, because the alternative is a subscription whose value is capped by specialist headcount.
What to check before you renew
Take the audience data subscription with the largest dollar figure next to it and count how many of your sellers opened it last month. Not the research team. The sellers. Then determine how many of them could explain where the number came from.
Next, time one unanticipated advertiser question from arrival to pitch-ready answer. That interval, rather than the size of the license or the sophistication of the interface, is what the advertiser experiences.
Finally, read the last ten proposals your team sent and count how many told the advertiser something about their own customer that they did not already know. That is the ratio the renewal conversation should turn on.
If nobody can tell you those numbers, that is the finding. If your sellers cannot tell the buyer where the number came from, that is the more expensive one.
