How decision intelligence could democratise analytics

Deci­sions, deci­sions, deci­sions. We make them every day: bad ones, good ones, seem­ing­ly incon­se­quen­tial ones that have pro­found impli­ca­tions. It’s no won­der “analy­sis paral­y­sis” plagues so many of us, even if the stakes are as low as pick­ing a brand of ketchup in a super­mar­ket aisle.

When the stakes are high­er – say, run­ning a multi­bil­lion-dol­lar com­pa­ny – busi­ness­es need to strate­gise. Many years ago, data col­lec­tion meant hir­ing con­sul­tants, con­duct­ing thor­ough A/B test­ing and in-depth mar­ket analy­sis. More recent­ly, the explo­sion of big data and smarter ana­lyt­ics soft­ware has allowed for “busi­ness intel­li­gence”, or BI: real­ly deep, gran­u­lar ana­lyt­ics, help­ing organ­i­sa­tions to carve out the cor­rect course of trav­el based on the infor­ma­tion they hold about the past and present. 

While busi­ness intel­li­gence has proved enor­mous­ly valu­able to organ­i­sa­tions of all stripes, a new prac­tice is emerg­ing: deci­sion intel­li­gence. Put sim­ply, this is the com­mer­cial appli­ca­tion of arti­fi­cial intel­li­gence (AI) to make bet­ter busi­ness deci­sions. It aims to take data-led insights a step fur­ther. It’s been called the “new BI”, but real­ly it means broad­en­ing the capa­bil­i­ties estab­lished by busi­ness intel­li­gence in a much more intu­itive, smarter way.

Handling complexity

All the data in the world is use­less if it doesn’t teach you any­thing. Just look at the NSA, which bus­ied itself by dili­gent­ly col­lect­ing all that it could on Amer­i­can cit­i­zens, ren­der­ing its sur­veil­lance sys­tems inef­fec­tive, accord­ing to whistle­blow­er William Bin­ney. Use­ful data needs to be sep­a­rat­ed from the junk, and busi­ness­es need to draw insights from it. 

Deci­sion intel­li­gence could offer a solu­tion.  Merg­ing data sci­ence with social sci­ence, it promis­es to democ­ra­tise ana­lyt­ics, apply­ing machine learn­ing to data and help­ing peo­ple make deci­sions. The social sci­ence aspect is cru­cial. While machines are get­ting much bet­ter at doing the heavy lift­ing for labo­ri­ous man­u­al tasks, they still lack the nuance and under­stand­ing that humans pos­sess. 

A deci­sion intel­li­gence ini­tia­tive might mean exam­in­ing sets of data, run­ning poten­tial out­comes through machine learn­ing mod­els, and pre­sent­ing deci­sion mak­ers with poten­tial cours­es of action to take, all in plain Eng­lish. 

This isn’t mere buzz from ven­dors. Ana­lyst firm Gart­ner believes a third of all large organ­i­sa­tions will employ ana­lysts that prac­tice deci­sion intel­li­gence, includ­ing deci­sion mod­el­ling, as soon as 2023. That’s part­ly because it’s becom­ing far more com­plex to make busi­ness deci­sions: a recent Gart­ner sur­vey found that over half of respon­dents (65%) believed the deci­sions they made were more com­plex than two years ago, while 53% said there was more pres­sure to jus­ti­fy or explain their deci­sions. 

Big tech leadership

As is often the case, big tech is shap­ing the trends. Pio­neer­ing com­pa­nies with the resources to spear­head exper­i­men­ta­tion, such as Google, are rethink­ing their approach to deci­sions and data. In 2019, Google hired its first “chief deci­sion sci­en­tist”, tech­nol­o­gy evan­ge­list and influ­en­tial data sci­en­tist Cassie Kozyrkov, to help meld data-led AI tools with behav­iour­al sci­ence.

IBM calls the prac­tice “Deci­sion Opti­miza­tion”, with its cus­tomers includ­ing logis­tics firms, ware­house providers, finan­cial ser­vices organ­i­sa­tions and ener­gy com­pa­nies. Mean­while, Man­ches­ter-based com­pa­ny Peak recent­ly drew $75m in Series C fund­ing from Soft­bank’s Vision Fund, and Quan­tel­lia, the busi­ness found­ed by machine learn­ing pio­neer Lorien Pratt, who has writ­ten a book on deci­sion intel­li­gence, counts major play­ers like Cis­co, SAP, and RBS among its cus­tomers. Typ­i­cal use cas­es span indus­tries like finance, trans­porta­tion, phar­ma­ceu­ti­cals, util­i­ties and oth­ers where sup­ply chain opti­mi­sa­tion is crit­i­cal to main­tain com­pet­i­tive­ness. 

Tra­di­tion­al BI relies on sta­t­ic data, mean­ing organ­i­sa­tions are look­ing in the rear-view mir­ror at a point in time, says Claire Rutkows­ki, CIO at soft­ware com­pa­ny Bent­ley Sys­tems. While it pro­vides a wealth of infor­ma­tion that can be used as the basis of bet­ter deci­sions, the user must know what data to look for and how to inter­pret it, she explains. 

“They need to work with a data office or sim­i­lar func­tion to fig­ure out how to extract the data, manip­u­late it, cre­ate fil­ters, then val­i­date the results before the report or dash­board is pub­lished.”

Deci­sion intel­li­gence is “much more pow­er­ful,” Rutkows­ki adds, because it allows for nat­ur­al lan­guage query­ing and can answer deep­er ques­tions, offer­ing insights on the “why” and not just the “what”. 

At present, how­ev­er, the frame­work for deci­sion intel­li­gence is loose­ly defined. Krish­na Roy, senior research ana­lyst for data sci­ence and ana­lyt­ics at 451 Research, says deci­sion intel­li­gence uses automa­tion and machine learn­ing, which isn’t usu­al­ly the case with plain old busi­ness intel­li­gence.

This seem­ing­ly small diver­gence has sig­nif­i­cant busi­ness impli­ca­tions, which are already prov­ing a pow­er­ful tool for some busi­ness­es.

The role of AI

For bev­er­age multi­na­tion­al Mol­son Coors, there was a need to gain bet­ter insight into its vast, com­plex oper­a­tions, con­tin­u­al­ly improv­ing how these were man­aged at a speed and scale that wouldn’t have been pos­si­ble with­out AI.  

In mid-2020 this led the com­pa­ny to Peak. Togeth­er, they set about assess­ing poten­tial areas of Mol­son Coors’s oper­a­tions where the plat­form could be deployed, speak­ing with key stake­hold­ers with­in the busi­ness in an effort to ascer­tain how their roles could be improved. By Christ­mas, they’d agreed on a tri­al, with work start­ing in Feb­ru­ary 2021 and rolling out nation­al­ly in July.

Mark Elston, dig­i­tal solu­tions con­troller at Mol­son Coors Bev­er­age Com­pa­ny, explains that the com­pa­ny has a huge tech­ni­cal ser­vices team oper­at­ing in the UK. Deci­sion intel­li­gence helps cap­ture insights into those touch­points and quick­ly trans­late them into ser­vice improve­ments.

One of the brewer’s major com­mer­cial projects this year was rebrand­ing one of its flag­ship lagers, Coors Light, to Coors, says Elston. This mul­ti­mil­lion-pound invest­ment meant replac­ing 20,000 dis­pense points with new­ly brand­ed ver­sions, a “mas­sive oper­a­tion where a deci­sion intel­li­gence solu­tion real­ly came into its own”.

Deci­sion intel­li­gence is one of the lat­est ways ven­dors are look­ing to address the skills gap – by automat­ing the deliv­er­ing of busi­ness insights, as well as explain­ing them in a way a non­tech­ni­cal user will under­stand

The sys­tem takes infor­ma­tion from engi­neers’ sched­ules, like planned jobs, most like­ly trav­el routes, skill spe­cialisms and the equip­ment they have in their vehi­cles, then cal­cu­lates where they are able to con­duct addi­tion­al rebrand work with­out sig­nif­i­cant dis­rup­tion, Elston says. 

“Because this is all being done by a com­put­er, it hap­pens live, mak­ing the whole process incred­i­bly dynam­ic and respon­sive. The plat­form was instru­men­tal in get­ting the work done effi­cient­ly and at pace, enabling our cus­tomers to be up and run­ning with new­ly brand­ed dis­pense points faster, with min­i­mal dis­rup­tion.”

Sto­ries like this are becom­ing more com­mon, with deci­sion intel­li­gence mak­ing a dif­fer­ence to busi­ness­es right now. 

Increased adop­tion will like­ly change the pace and qual­i­ty of deci­sion-mak­ing, enabling organ­i­sa­tions to speed up the entire process, quick­ly glean insights and devel­op plans to either act as coun­ter­mea­sures to their dis­cov­er­ies or dri­ve oppor­tu­ni­ties hard­er, says Rutkows­ki. Bent­ley Sys­tems is cur­rent­ly lay­er­ing deci­sion intel­li­gence tools into its data tech­nol­o­gy stack to draw addi­tion­al insights from its data.

“Com­pa­nies will be nim­bler with their strate­gies, which should dri­ve bet­ter per­for­mance and bet­ter results. Cor­rec­tive action will be more focused when we under­stand the ‘why’ and will like­ly be more suc­cess­ful,” she adds. “When com­pa­nies dis­cov­er pos­i­tive insights on a giv­en data set, they can fur­ther lever­age strate­gies and ini­tia­tives that are dri­ving growth, expand­ing it even fur­ther.” 

And for busi­ness­es strug­gling to recruit pro­fes­sion­als to inter­pret the data they gen­er­ate and draw insight from it, deci­sion intel­li­gence could pro­vide some reprieve for alle­vi­at­ing the data skills gap, which the British gov­ern­ment has acknowl­edged is an issue in the UK.

Deci­sion intel­li­gence “promis­es to bring ana­lyt­ics to the mass­es”, explains Roy. “It’s all about pro­vid­ing insights to answer busi­ness ques­tions, with­out the user requir­ing smarts in ana­lyt­ics or data.”

Democratising analytics

Busi­ness­es have cre­at­ed self-serve data plat­forms for employ­ees, mean­ing they don’t need to be data spe­cial­ists to draw insights and make bet­ter deci­sions. Deci­sion intel­li­gence prod­ucts could sim­pli­fy this process, mak­ing ana­lyt­ics avail­able to every­one that would ben­e­fit from it. 

“Ana­lyt­ics has yet to be democ­ra­tised, so peo­ple with­out skills in using analy­sis tools or under­stand­ing strug­gle to get the insights they need to do their job,” says Roy. “Deci­sion intel­li­gence is one of the lat­est ways ven­dors are look­ing to address the skills gap – by automat­ing the deliv­er­ing of busi­ness insights, as well as explain­ing them in a way a non­tech­ni­cal user will under­stand.”

Organ­i­sa­tions that want to get start­ed should first research how AI and deci­sion intel­li­gence could fit into their busi­ness, advis­es Elston. Key stake­hold­ers should be brought into the con­ver­sa­tion ear­ly to hear how the tech­nol­o­gy will ben­e­fit them. Mol­son Coors spent months on this ini­tial research and con­ver­sa­tion phase, which “paid div­i­dends” in achiev­ing buy-in lat­er on, says Elston, and allowed the com­pa­ny to hit the ground run­ning.

Com­pa­nies will be nim­bler with their strate­gies, which should dri­ve bet­ter per­for­mance and bet­ter results. Cor­rec­tive action will be more focused when we under­stand the “why” and will like­ly be more suc­cess­ful

Start sim­ply, Elston adds. Ear­ly deploy­ments should be easy to imple­ment. They should deliv­er val­ue quick­ly to build con­fi­dence with lead­er­ship teams as well as those that are using the solu­tion.

“Intro­duc­ing this tech­nol­o­gy effec­tive­ly requires the rel­e­vant stake­hold­ers to embrace change, and there is no bet­ter way of get­ting buy-in to that than a suc­cess­ful proof of con­cept,” he says. 

Deci­sion intel­li­gence will prob­a­bly be a stan­dard com­po­nent for busi­ness­es one day, but it will be “crit­i­cal for com­pet­i­tive­ness” in the mean­time, says Roy. Busi­ness­es would do well to explore the field today and take at least some of the pain out of deci­sion-mak­ing. 

---