In a recent post by Eric Peter­son (Web Ana­lyt­ics Demys­ti­fied), he brought up the inter­est­ing topic of what it means to be “data-driven” and pro­posed that the data-driven busi­ness is a myth. He actu­ally went so far as to say, “A ‘data-driven busi­ness’ would be doomed to fail.” That’s a bold pre­dic­tion, and a bit too omi­nous for me.

Before I get into the sub­tle seman­tic dif­fer­ences between being data-informed and data-driven, I’d like to start by focus­ing on the inter­pre­ta­tion that a “data-driven” orga­ni­za­tion will blindly fol­low what­ever its data tells it to do. In all my years in web ana­lyt­ics con­sult­ing, I’ve never run into an orga­ni­za­tion that is pre­pared to let the data con­trol the deci­sion mak­ing process. Influ­ence — yes. Inform — yes. Inspire — yes. Con­trol — no. It reminds me of an episode from the TV show, “The Office”, where Michael Scott and Dwight Schrute went out to win back lost clients with gift bas­kets. As they were try­ing to find a par­tic­u­lar client using a GPS device, the fol­low­ing encounter hap­pened between Michael Scott who was dri­ving and his trusted “Assis­tant to the Regional Man­ager” Dwight:

GPS: Make a right turn.
Dwight: Wait, wait, wait! No, no, no! It means bear right, up there.
Michael: No, it said right. It said take a right.
Dwight: No, no, no. Look, it means go up to the right. Bear right over the bridge, and hook up with 307.
Michael: Maybe it’s a short­cut, Dwight. It said go to the right. [turns right]
Dwight: It can’t mean that! There’s a lake there!
Michael: The machine knows where it is going!
Dwight: This is the lake!
Michael: The machine knows— stop yelling at me!
Dwight: No, it’s— there’s no road here! [car dri­ves into lake]


Data-driven orga­ni­za­tions seek out rel­e­vant data to help inform and shape, not dic­tate or con­trol, their key busi­ness deci­sions. They’re not going to drive into a prover­bial lake because their web ana­lyt­ics GPS tells them to — at least not when their busi­ness sense or intu­ition dis­agrees with the deci­sion. Both logic and intu­ition (com­mon sense in the case of Michael Scott) are needed and equally impor­tant to the deci­sion mak­ing process. They can act as valu­able checks and bal­ances to each other. Data can reveal when your gut feel­ing is far askew, and intu­ition can ground your high-flying cal­cu­la­tions in rel­e­vant past expe­ri­ence. They work together and need to be balanced.

When I think of being “dri­ven” in any­thing (fam­ily, work, val­ues, etc.), I think of con­vic­tion and deter­mi­na­tion. Hav­ing a data-driven mind­set is a com­mit­ment to ensur­ing all forms of data are actively con­tribut­ing to mak­ing bet­ter busi­ness deci­sions. For me it’s about fix­ing and cor­rect­ing an exist­ing imbal­ance in order to find more equi­lib­rium. Whether we like it or not, busi­ness deci­sions are still pre­dom­i­nantly dri­ven by intu­ition. Data hasn’t had an equal foot­ing at the deci­sion mak­ing table — since, well, the dawn of data. It’s still fre­quently viewed as a nice-to-have, not as a need-to-have. It’s wel­comed with open arms when it sup­ports a par­tic­u­lar posi­tion but can be dis­missed and belit­tled when it doesn’t. By encour­ag­ing indi­vid­u­als and orga­ni­za­tions to be more data-driven (and account­able), I’m look­ing for that eye-of-the-tiger hunger for data and insights that still isn’t as preva­lent as it should be in our data-rich dig­i­tal age. By push­ing for a data-driven mind­set and approach, I’m advo­cat­ing for data to receive the same level of con­sid­er­a­tion and appre­ci­a­tion as intu­ition already receives. It’s def­i­nitely not about replac­ing intu­ition with data (that’s a false dichotomy) but about get­ting data a chair at the big peo­ple table.

When­ever you have a con­flict between the two sides of logic and intu­ition dur­ing deci­sion mak­ing, you’d ratio­nally expect peo­ple to rec­on­cile the dif­fer­ences in their minds before act­ing. Regard­less of the type of per­son you are — “data-driven” or “data-informed” — you’ll decide to gather more data if the data appears to be wrong or doesn’t agree with your intu­ition. On the flip side, a data-driven per­son will ques­tion their assump­tions if their intu­ition feels way off base and the data looks sound. What will the data-informed per­son do? Prob­a­bly still ques­tion the data if it doesn’t agree with their intu­ition. That’s the prob­lem. Being data-informed is just too pas­sive or weak. While some peo­ple strug­gle with the seman­tics of “data-driven”, I strug­gle with the seman­tics of “data-informed”. Is some­one data-informed if they reg­u­larly receive a sched­uled report in their Out­look inbox? Do we expect them to have at least looked at the report? Do we expect them to have prop­erly inter­preted and under­stood the report? Do we expect them to be open-minded about con­sid­er­ing the data if it con­tra­dicts their estab­lished views? For me, data-informed cre­ates a weaker stan­dard than data-driven does. Rather than lean­ing into the data ready to pounce on an opti­miza­tion oppor­tu­nity (data-driven), I envi­sion some­one lean­ing back wait­ing for some­thing obvi­ous to hit them in the face (data-informed).

Logic faces a big­ger uphill bat­tle than intu­ition does in the decision-making process. Con­sider the metaphor of the Ele­phant and its Rider put forth by Chip and Dan Heath in their excel­lent book, “Switch”. Our emo­tional side is our Ele­phant and our ratio­nal side is our Rider. “Perched atop the Ele­phant, the Rider holds the reins and seems to be the leader. But the Rider’s con­trol is pre­car­i­ous because the Rider is so small rel­a­tive to the Ele­phant. Any­time the six-ton Ele­phant and the Rider dis­agree about which direc­tion to go, the Rider is going to lose.”

If the Rider is reduced to being just a back-seat dri­ver or an informed pas­sen­ger to the Ele­phant, you know who is ulti­mately going to decide the path. At some point the Ele­phant may get impa­tient and will act with­out data. In my view, a data-driven approach needs to respect the strengths of the Ele­phant (speed, energy, cre­ativ­ity, and rel­e­vant expe­ri­ence) and be cau­tious of its weak­nesses (biases, fears, sim­plis­tic heuris­tics, and mis­matched expe­ri­ence). In most cases, the Ele­phant and the Rider want to get to the same des­ti­na­tion; they just dis­agree on the route some­times. Work­ing together with a data-driven empha­sis will ensure the Elephant’s intu­ition receives the proper rigor, scrutiny, and dis­ci­pline from the Rider in order to safely and effi­ciently reach the final des­ti­na­tion. Essen­tially, a ratio­nal “data-driven” approach can help to cal­i­brate a deci­sion maker’s intu­ition over time and increase aware­ness for when gut feel­ings should and shouldn’t be fol­lowed, increas­ing the like­li­hood of suc­cess­ful out­comes. The Ele­phant will walk all over a more pas­sive, data-informed Rider.

For most com­pa­nies, becom­ing more data-driven is a long-term goal, which requires focus on impor­tant key areas such as peo­ple, processes, and tech­nol­ogy. Do we avoid data if it causes us to be par­a­lyzed in our deci­sion mak­ing? No, we fig­ure out ways to max­i­mize the ben­e­fits of data while min­i­miz­ing its draw­backs. For many com­pa­nies, data is a real com­pet­i­tive advan­tage that some firms are lever­ag­ing and oth­ers are not. I hope that by low­er­ing the tar­get from becom­ing data-driven to being sat­is­fied with merely data-informed, orga­ni­za­tions don’t lose momen­tum or fall short of their orig­i­nal vision for orga­ni­za­tional suc­cess. Know­ing Eric, I think he’s try­ing to make us think a lit­tle more rig­or­ously about the phrases we use and the posi­tions we take. In my mind, the data-controlled busi­ness is the real myth, while the data-driven orga­ni­za­tion is still a desir­able standard.

John Hunter
John Hunter

Data driven does not mean you remove judgement (though many act as it does). Data driven mean you use data and when you are making a judgement you use data (and notice when you are not doing so). It isn't as simple as just looking at a spreadsheet and acting based on what numbers you see. You have to understand what the numbers mean. You have to know how they were collected. You have to understand when the numbers seem to be wrong. As an example, when I was, first getting an organization to try and use data a supervisor gave me data on phone calls. The data was obviously wrong. It showed far more calls than I knew were handled in that office. A supervisor needs to know their business. But their reaction was that we were suppose to use data now, this is the data. Anyway I went and talked to the person collecting the data, they counted a phone call, then if another came in they put #1 on hold and counted the next one, put #2 on hold went back to #1 and counted it again (3 now), then another came put #1 on hold... The numbers were at least inflated by 200-300%. You must understand the data and make sure the data is telling you what you need to know to make decisions. Software often leads people to believe the numbers presented are somehow super valid as a computer gave the number. This is dangerous. People need to understand the data and far too often people don't. That is a serious problem.

Brent Dykes
Brent Dykes

Philip, I'm glad the post resonated with you. I agree with you that acting on the data can be a huge challenge for many companies. Being data-driven needs executive and company-wide commitment because data won't generate any value on its own if it's not acted upon. Cheers, Brent.


I agree wholeheartedly. Arguments about semantics just detract from more important conversations. There isn't a company out there that implements an analytics tool and then puts the business on autopilot. In fact, it seems to be quite the opposite. The biggest concern of many companies, when it comes to analytics, seems to be both getting and applying actionable data.