Get­ting started with de­mand for­cast­ing

So­phis­ti­cated plan­ning is mod­est, but grow­ing, in gro­cery.

Progressive Grocer (India) - - Contents - By John Karolef­ski

So­phis­ti­cated plan­ning is mod­est, but grow­ing, in gro­cery.

Wheatsville Food Co-op, in Austin, Texas, re­lies on a va­ri­ety of tech­niques to fore­cast de­mand for ev­ery­day shop­ping. There’s a decade’s worth of sales and move­ment data in its point-of-sale sys­tem that can be ac­cessed to de­ter­mine the ac­tiv­ity of a par­tic­u­lar item or cat­e­gory.

For sea­sonal vari­a­tions cre­ated by hol­i­days such as Thanks­giv­ing, the co-op can see what pop­u­lar tra­di­tional items shop­pers bought. By com­par­ing past sales dol­lars, more re­cent cus­tomer count and av­er­age bas­ket, a rea­son­able es­ti­mate can be made for the cur­rent year.

“While we know that stuff­ing and tur­keys are Thanks­giv­ing sta­ples, we have also no­ticed the rise in de­mand for gluten-free bread prod­ucts and ve­gan meat al­ter­na­tives,” notes Niki Nash, pack­aged man­ager for the only re­tail food co­op­er­a­tive op­er­at­ing in Texas. “As spe­cial­ized di­ets be­come more main­stream, our va­ri­ety and amount of these prod­ucts have in­creased. Each year, we mon­i­tor these non­tra­di­tional ver­sions of tra­di­tional fa­vorites for pop­u­lar­ity and ad­just our or­ders for the next year. By us­ing a per­pet­ual in­ven­tory man­age­ment sys­tem, we make sure that we’re never out of key items.”

Whether they’re co-ops, in­de­pen­dent gro­cers or large re­gional food chains, com­pa­nies must main­tain the right bal­ance of sup­ply and de­mand to pre­pare for surge pe­ri­ods such as Thanks­giv­ing, as well as for ev­ery­day shop­ping. Out-of-stocks dur­ing the hol­i­days — or any time of the year, for that mat­ter — may have long-term neg­a­tive im­pli­ca­tions for shop­per loy­alty. In other words, cus­tomers change stores.

“Back-of-an-en­ve­lope cal­cu­la­tions and sim­ple spread­sheets don’t cut it when you’re mak­ing fore­cast­ing de­ci­sions about as­sort­ments, pric­ing, pro­mo­tions and sup­ply chain. These ar­eas de­pend on ac­cu­rately pre­dict­ing con­sumer de­mand, and that ac­cu­racy de­pends on an­a­lyt­ics,” says Dan Mitchell, a re­tail in­dus­try con­sul­tant for SAS, a Cary, N.c.based provider of an­a­lyt­ics, busi­ness in­tel­li­gence, and data man­age­ment soft­ware and ser­vices.

Gro­cers can use de­mand fore­cast­ing in a num­ber of ar­eas to­day, in­clud­ing la­bor and staffing, per­ish­ables, re­plen­ish­ment of shelf-sta­ble goods, re­sponse to pro­mo­tions, and un­der­stand­ing the ef­fect of price changes by prod­uct and ge­og­ra­phy. Some com­pa­nies are ob­vi­ously fur­ther along than oth­ers on the path to ef­fec­tive de­mand fore­cast­ing. In fact, many are just beginning the jour­ney to us­ing these tech­nolo­gies ef­fec­tively.

“It’s im­por­tant to note that the nec­es­sary tech­nol­ogy is out there, though gro­cery is tra­di­tion­ally low in adopt­ing these ex­ist­ing tech­nolo­gies,” says Mike Neff, a part­ner at New York-based con­sul­tancy Kurt Sal­mon. “How­ever, with the ex­is­tence of these com­pa­nies mov­ing to cloud-based ap­pli­ca­tions, costs for the tech­nol­ogy are com­ing down. Now, smaller play­ers in the mar­ket­place can af­ford to pay and play in a way that’s most ef­fi­cient and cost-ef­fec­tive. The trend in get­ting more tech­nol­ogy to ex­ist will pro­vide bet­ter ca­pa­bil­i­ties in small and large gro­cery spa­ces.”

Strate­gic Po­si­tion

What spe­cific things can gro­cers do to start off on the right foot?

Deb­bie Stan­ton John­son, gro­cery in­dus­try prin­ci­pal at Capgem­ini Con­sult­ing, an in­ter­na­tional con­sul­tancy with U.S. of­fices in six states, ad­vises gro­cers to po­si­tion de­mand fore­cast­ing as a strate­gic ini­tia­tive across the value chain, sup­port­ing cus­tomer loy­alty and ven­dor and em­ployee en­gage­ment while en­sur­ing re­quire­ments are gath­ered across the var­i­ous busi­ness func­tions and or­ga­ni­za­tions.

“Data con­sis­tency, qual­ity and avail­abil­ity are key to a suc­cess­ful de­mand fore­cast,” John­son ex­plains. “Most fore­cast­ing al­go­rithms re­quire more than one year of stores’ TLOG [trans­ac­tion log] data, and for im­proved ac­cu­racy, a min­i­mum of two years of TLOG data. This data should be avail­able for all lo­ca­tions: stores, e-com­merce, cat­a­logs, etc.

“Gro­cery en­com­passes most of the com­plex­ity seen in re­tail for de­mand fore­cast­ing,” she con­tin­ues. “Short life­cy­cle prod­ucts, date-driven prod­ucts, fresh prod­ucts such as pro­duce and meat each have unique and spe­cific com­plex­i­ties. It is es­sen­tial for gro­cers to not only be able to fully un­der­stand the im­pli­ca­tions of the com­plex­ity, but also to en­sure that the ben­e­fits out­weigh the costs.”

To launch a pro­gram, she sug­gests pick­ing a less com­plex as­sort­ment like core gro­cery be­cause it will de­liver vis­i­ble suc­cess with quan­ti­ta­tive ben­e­fits and build mo­men­tum for the prod­uct cat­e­gories.

“Don’t un­der­es­ti­mate the value of data qual­ity,” notes Mitchell, of SAS. “To truly un­der­stand de­mand, you need in­for­ma­tion beyond stan­dard sales and in­ven­tory data. Also con­sider data like prod­uct at­tributes, trad­ing-area de­mo­graphic data, and data about sea­sonal changes like weather and hol­i­days. It’s also im­por­tant to in­vest in ed­u­ca­tion for your team, make the or­ga­ni­za­tional changes that will help you be­come a data-driven en­ter­prise, and roll out fore­cast­ing projects with built-in mile­stones to show ROI from the start.”

Toby Br­zoznowski, EVP of Lla­ma­soft, an Ann Ar­bor, Mich.-based provider of sup­ply chain de­sign, an­a­lyt­ics and op­ti­miza­tion so­lu­tions, adds that de­mand fore­cast­ing needs the flex­i­bil­ity to han­dle sea­son­al­ity and var­ied de­mand pat­terns, and the abil­ity to look in de­tailed gran­u­lar­ity. Hav­ing fore­cast­ing in the cloud is also a ben­e­fit for a fore­cast­ing so­lu­tion, given the high num­ber of SKUS and the abil­ity to use high-per­for­mance com­put­ing, he points out.

Mean­while, Chris­tian Ha­gen, part­ner in the dig­i­tal trans­for­ma­tion prac­tice of Chicago-based con­sul­tancy A.T. Kear­ney, cau­tions gro­cers that fore­cast­ing, re­plen­ish­ment and al­lo­ca­tion are dif­fer­ent for per­ish­ables. She of­fers sev­eral rec­om­men­da­tions to help drive ben­e­fits and per­for­mance:

Costs for the tech­nol­ogy are com­ing down. Now, smaller play­ers in the mar­ket­place can af­ford to pay and play in a way that’s most ef­fi­cient and cost-ef­fec­tive. — Mike Neff Kurt Sal­mon

De­sign per­ish­ables-spe­cific fore­cast­ing and re­plen­ish­ment so­lu­tions

• Lever­age mar­ket-lead­ing ap­pli­ca­tions and tools for fore­cast­ing an­a­lyt­ics

• Stan­dard­ize a toolset for per­ish­ables

• Drive re­plen­ish­ment plan­ning to pri­or­i­tize “fresh­ness”

• De­fine the right met­rics and track with data

Part­ner with sup­pli­ers

• Work with sup­pli­ers to un­der­stand their fresh and per­ish­able sup­ply chain and con­straints

• Co-cre­ate so­lu­tions to en­sure per­for­mance is be­ing op­ti­mized end to end

• Es­tab­lish pi­lots to test im­prove­ment op­por­tu­ni­ties — track the met­rics in these pi­lots to gauge per­for­mance ac­cu­rately.

Real-time Re­al­ity

In April 2016, re­search by the Kurt Sal­mon con­sul­tancy found that 55 per­cent of food spend­ing isn’t done in con­ven­tional gro­cery stores, but in al­ter­na­tive classes of trade. Why? It de­ter­mined that out-of-stocks at store level aren’t be­ing tol­er­ated by con­sumers. In fact, gro­cers are faced with a big­ger prob­lem than in years past.

“Also, you can’t just look at point of sale or one data op­tion,” says Kurt Sal­mon’s Neff. “More and bet­ter data is crit­i­cal. In ad­di­tion to syn­di­cated data, it’s time we add a third op­tion in e-com­merce. Click-and-col­lect e-com­merce for gro­cers is much more com­mon and im­por­tant. It pro­vides bet­ter info that is closer to the trends of what’s hap­pen­ing at a par­tic­u­lar store level while meet­ing the needs and de­mands of cus­tomers. Folks can lever­age the tech­nol­ogy and dif­fer­ent lev­els of im­mer­sion into that tech­nol­ogy that will take tra­di­tional points of sale, syn­di­cated data and e-com­merce data to pro­vide in­for­ma­tion to gro­cers that wasn’t avail­able be­fore.”

Ron Wil­son, an­other part­ner at Kurt Sal­mon, stresses the need for real-time data to in­ter­act and en­gage to cre­ate agility and get bet­ter at track­ing per­pet­ual in­ven­tory.

“DSRS [de­mand sig­nal repos­i­to­ries] aren’t good enough to ad­dress prob­lems to­day be­cause they’re look­ing on a daily ba­sis. The only way it’ll truly be solved [is] if we can get to a place where we can in­ter­act and en­gage con­sumers in a mo­bile en­vi­ron­ment that’s used col­lab­o­ra­tively in real time in both mo­bile and re­tail to get them these core prod­ucts that we need,” Wil­son says. “Un­til then, we rely on rel­e­vancy of daily data and mar­ket re­tail data, which can take three to four weeks, so it’s too late.

“There needs to be a Big Data ap­proach.” he con­tin­ues, “that al­lows gro­cers to look at things in real time and cre­ate and use the types of science and an­a­lyt­ics to start pre­dict­ing po­ten­tial im­pacts, and be­ing able to pro­vide a way to solve them. Ap­ply­ing sci­en­tific math­e­mat­ics to how one can ad­dress and un­der­stand de­mand, and then us­ing in­no­va­tion to ap­proach it, will drive more value — though it may in­cur more cost.”

Re­al­iz­ing Ben­e­fits

Cost not­with­stand­ing, the good news is that more gro­cers have adopted so­lu­tions and pro­cesses for de­mand fore­cast­ing fo­cused on re­plen­ish­ment and in­ven­tory man­age­ment. They are re­al­iz­ing ben­e­fits in the ar­eas of im­proved prof­itabil­ity, in­creased prod­uct avail­abil­ity, op­ti­mized in­ven­tory and re­duced in­ven­tory hold­ing costs.

“In­te­grat­ing pro­mo­tional plan­ning with de­mand fore­cast­ing is es­sen­tial to tighten fore­casts,” says John­son, of Capgem­ini. “Lead­ing gro­cers are mod­el­ing their data to im­prove new store open­ing and holiday/pro­mo­tional pe­riod fore­casts. For new store open­ings, es­tab­lished and cur­rent store clus­ter­ing and seg­men­ta­tion mod­els en­able fore­casts to project sales for sim­i­lar stores. These mod­els pre­dict im­proved fore­casts and dis­play pat­terns his­tor­i­cally rel­e­vant in other clus­ters.

“What is in­creas­ingly the ‘Holy Grail’ for gro­cers,” she con­tin­ues, “is to lever­age un­struc­tured data like so­cial trend­ing to pick up on cat­e­gories and brands that will ex­pe­ri­ence de­mand . ... While there are def­i­nite chal­lenges to im­ple­men­ta­tion, the key to suc­cess for or­ga­ni­za­tions is to adapt and uti­lize re­al­time in­for­ma­tion.”

Con­sider data like prod­uct at­tributes, trad­ing area de­mo­graphic data, and data about sea­sonal changes like weather and hol­i­days. It’s also im­por­tant to in­vest in ed­u­ca­tion for your team. — Dan Mitchell SAS

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.