Escap­ing the lab­o­ra­to­ry: Apply­ing lab­o­ra­to­ry find­ings in the com­mer­cial world

Gareth Har­vey, Uni­ver­si­ty of Applied Sci­ences West­ern Switzerland

YouTube

By load­ing the video, you agree to YouTube’s pri­va­cy pol­i­cy.
Learn more

Load video

Full Tran­script:

Gareth J. Har­vey:
Cool. So I’m good to car­ry on talk­ing and pret­ty much fol­low the same sort of thing that we’ve just been explor­ing. How can we take the find­ings that we know from the lab-based stud­ies and apply them in the com­mer­cial con­text? But rather than look­ing at, shall we say, some of the nice clean, exper­i­men­tal designs you’ve explored so far I’m going to look at slight­ly more, should we say some of the messy ques­tions that com­pa­nies ask? So to start off with, I just thought I’d say who I am. I say I’m Gareth Har­vey, I’m a Pro­fes­sor of Con­sumer Psy­chol­o­gy at HESGE in Gene­va, and I’m a Con­sumer Psy­chol­o­gist. So I’m par­tic­u­lar­ly inter­est­ed in under­stand­ing how envi­ron­men­tal cues impact behav­ior. So if you go around a super­mar­ket and you can’t find the eggs, the bacon, what­ev­er it is it’s prob­a­bly my fault or some­one like me.

Gareth J. Har­vey:
And so I do a lot of work com­mer­cial­ly design­ing super­mar­ket store lay­outs. So try­ing to design them in terms of where the prod­ucts are, how wide the aisle should be, how tall the shelv­ing should be, the music that’s play­ing, all of those sort of fac­tors hope­ful­ly I’ve con­sid­ered from a psy­cho­log­i­cal per­spec­tive. So rather than doing aca­d­e­m­ic research in terms of pub­lish­ing papers, I work with com­pa­nies and do com­mer­cial research. So as well as teach­ing at a uni­ver­si­ty, I do that com­mer­cial­ly. And I’m the sci­en­tif­ic advi­sor for the UK’s old­est inde­pen­dent design agency. And again, my role is to take prin­ci­ples of psy­chol­o­gy and try and use it to make pack­ag­ing more effec­tive. It’s a bit of basi­cal­ly how com­pa­nies make more money.

Gareth J. Har­vey:
And the way it should work is a com­pa­ny will approach me and nor­mal­ly they’ll give me a brief. Now unlike the ques­tions that you’ve heard before, like which col­or should we use for our pack­ag­ing or for our logo? The ques­tions I typ­i­cal­ly get are a lot more, let’s say blur­ry. So for exam­ple, it’s a case of some­one will say, a com­pa­ny will come to me and say, “We want peo­ple to spend more mon­ey in the aisles.” Or “We want them to spend more time. Devel­op an inter­ven­tion on how we can do that.” Now if you’re inter­est­ed in behav­ioral sci­ence, you’ll be aware that over the last, what? 15 years, there’s been a mas­sive inter­est. Books such as Pre­dictably Irra­tional, Nudge, Influ­ence, The Choice Fac­to­ry they’ve all been pub­lished and lots of peo­ple are real­ly inter­est­ed by them, but they’re pret­ty much all say­ing exact­ly the same sort of fac­tors. So they’re all talk­ing about the same interventions.

Gareth J. Har­vey:
What I’ve been asked to do by com­pa­nies is say, how can I find oth­er things from the con­sumer psy­chol­o­gy lit­er­a­ture and how can we apply them to devel­op nov­el inter­ven­tions and hope­ful­ly give my clients a com­pet­i­tive advan­tage? So a typ­i­cal brief I had, which I was work­ing on ear­li­er on in what? A cou­ple of months ago, client were say­ing, “Right, we want peo­ple to spend more time in the super­mar­ket.” So tra­di­tion­al brick and mor­tar store. And most peo­ple think con­sumers walk up and down every aisle. Well, we’ve got lots of track­ing data that says that’s just not the case. Most peo­ple the way they walk is we use what’s called a race course anal­o­gy. You walk around the out­side and then you basi­cal­ly just dive down the aisle, grab what you need and go back to the safe­ty of the race course, the out­side, the perimeter.

Gareth J. Har­vey:
And what we typ­i­cal­ly find is that we’re try­ing to find ways to encour­age peo­ple to go down. Now we know from envi­ron­men­tal psy­chol­o­gy, peo­ple like feel­ing in con­trol. So if you’ve got an aisle and it’s quite nar­row, peo­ple are less like­ly to go down it and if they do they spend less time there. So it’s all about try­ing to find ways to make it feel more spa­cious. So that’s why super­mar­kets have real­ly high ceil­ings, it’s why the aisles are as wide as we can pos­si­bly make them. It’s why the shelv­ing, in an ide­al world, should nev­er be taller than you, you’d always try and be able to see over it. But one of my clients said, “Well, we can’t make those changes, we’re kind of fixed we need a set num­ber of aisles, is there any­thing else we could do?” So we came up with this idea of well, could we per­mit the shelv­ing slop­ing? So it’s kind of slop­ing away from you, so at least it gives you the illu­sion that there’s more space in the aisle.

Gareth J. Har­vey:
So we came up with this con­cept and we thought, well, that should work. Let’s test it. Now before we rolled it out to test it in a super­mar­ket, well that’s going to get A, very expen­sive and if it does­n’t work we look fool­ish. Now you may say, “Okay, if it does­n’t work you’ve learned some­thing. You’ve learned that you have an unhap­py client and you’ve pos­si­bly lost a lot of mon­ey.” So we try and min­i­mize the chances things don’t work. And the way we start off with that is we test it out, first of all in a lab-based study. So the first process was we start off, we cre­ate a fake shop­ping envi­ron­ment in the lab and we test it in there. It took us four or five iter­a­tions to work out how much the shelv­ing need­ed to slope before it worked in the lab aesthetic.

Gareth J. Har­vey:
Once we were hap­py we got the effect work­ing in the lab-based study, we then rolled it out into a super­mar­ket. And we were actu­al­ly able to test it out in one super­mar­ket in the North­east of Eng­land. So we total­ly redesigned that store with all of these shelv­ings, got it work­ing, and we’re able to see that peo­ple were spend­ing approx­i­mate­ly 15% more in the aisles when this redesign was in place. 15% more in the aisles and again, there was a cor­re­spond­ing increase in the amount of mon­ey that peo­ple were spend­ing. So that was great, it worked. But it’s only been shown to work in one store. So once we did that, it was then test­ed out and it was actu­al­ly rolled out across every super­mar­ket of this one brand across Cape Town. Oh, sor­ry. South Africa start­ing up in Cape town.

Gareth J. Har­vey:
You’ll be glad to know that the com­pa­ny who was pay­ing for this research, they paid for me to trav­el from Gene­va to the North­east of Eng­land to do the research or design it. They could­n’t pay for me to go to South Africa to run the aes­thet­ic, but that’s the way the research typ­i­cal­ly works. That’s the way it should go. But the impor­tant thing to high­light here is yes, we start off in the lab, but we have to actu­al­ly repli­cate in real world. The lab has so many dif­fer­ent prob­lems or chal­lenges that we need to come, the lack of eco­log­i­cal valid­i­ty is one. And far too often psy­chol­o­gists or design­ers if you’re work­ing with them, they’ll over­look some real­ly basic things. So eye-track­ing is at the heart of a lot of prod­uct design. Far too many design­ers they’ll do a great eye-track­ing design. Look at break­fast cere­al, a cat­e­go­ry I work with quite a lot, they’ll do an eye-track­ing study and there’ll be look­ing at some­thing like that. Sor­ry, point­ing on green screen is quite hard.

Gareth J. Har­vey:
It works. They’ll get some nice results, but the results are rarely gen­er­al­ized to the com­mer­cial set­ting. Well why? Because in a lab when we’re shop­ping we don’t just see one prod­uct, you’ll see, for a start, we’ll see a wall of prod­ucts. So a wall of Weet­abix. Think about the impact of one Cad­bury’s Dairy Milk bar has. It’s min­i­mal, it’s not good to impact you. What hap­pens when you see a wall of Cad­bury’s pur­ple, lots of Dairy Milk of 12 bars? Actu­al­ly it’s going to attract your atten­tion in a very dif­fer­ent way. You’re not able to see the sides of the pack­ag­ing, it changes how we per­ceive them. And most lab-based stud­ies, fair­ly basic flow but they don’t con­sid­er it. But even if we test them with these walls, it’s not enough.

Gareth J. Har­vey:
Peo­ple behave very dif­fer­ent­ly in a super­mar­ket because, well let’s be hon­est, there’s a huge amount of noise out there. You’ve got a wall of Weet­abix, but you’ve also got the own brand ones, you’ve got oth­er cere­al com­pet­ing for your atten­tion, and you’ve got lots of peo­ple walk­ing past. You might be shop­ping with a, I don’t know, a child who’s scream­ing want­i­ng atten­tion, there might be anoth­er cou­ple hav­ing an argu­ment and you’re try­ing to pay atten­tion and work out what they’re argu­ing about. You’re not pay­ing the same con­scious atten­tion to the cere­al as if you were in a lab-based set­ting when a nice researcher says, “Look at this pack­ag­ing for a cou­ple of min­utes. Oh yeah, what grabs your atten­tion?” Or how­ev­er they phrase it.

Gareth J. Har­vey:
When you’re shop­ping, you might only look at the Weet­abix for a cou­ple of sec­onds. So they’re real­ly dif­fer­ent par­a­digm, so we need to under­stand that a lab-based study is a very dif­fer­ent beast to how we test things in the field. So, as I said, my role is to take these find­ings and apply them into the com­mer­cial set­ting. And I’m try­ing to find dif­fer­ent things in psy­chol­o­gy lit­er­a­ture, so the pub­lished research that actu­al­ly shows things that are a bit dif­fer­ent. But what’s real­ly inter­est­ing is most of the papers that are pub­lished don’t repli­cate in the com­mer­cial set­ting. And there’s some real­ly sim­ple rea­sons why, first of all the set­tings that they use or the designs they use they use real­ly com­pli­cat­ed sta­tis­tics. And yes, they can find an effect, but a lot of the con­sumer site research they’ll be using struc­tur­al equa­tion mod­el­ing, mul­ti­ple regressions.

Gareth J. Har­vey:
So they can show that, say, fac­tor X equals con­sumer pur­chas­ing, but only when they medi­ate out for per­son­al­i­ty, age, and income lev­els. So we do some lots of com­plex tasks to medi­ate or out of those vari­ables, that noise we get rid of it and we show a nice effect. Well when I’m design­ing an inter­ven­tion for com­pa­nies, I can’t con­trol all those sort of vari­ables they’re part of the data I have to work with. So a real­ly sim­ple exper­i­men­tal design that does­n’t use those com­pli­cat­ed stats, have a much bet­ter chance of work­ing in the field. Like­wise, in lab-based stud­ies they have the con­trol for all sorts of noise. It’s a real­ly sim­ple envi­ron­ment. So the only stud­ies that are like­ly to repli­cate into the com­mer­cial sec­tors are ones with a real­ly large effect size. And this is some­thing that we real­ly need to pay atten­tion to. So don’t get too wor­ried about your p‑values, the effect size is prob­a­bly one of the big­ger things to focus on.

Gareth J. Har­vey:
And actu­al­ly com­pa­nies often they may total­ly ignore the p‑value, but they’re more con­cerned about try­ing to under­stand what is the sam­pling error in your research. If we say X had an effect, well how con­fi­dent are we in that? So from the busi­ness domain, we’re going to be look­ing at it. And some of those things we’re just going to be focus­ing on is try­ing to work out well okay, let’s have a look and say they may work, but can we take it to the com­mer­cial envi­ron­ment? First of all, from a fea­si­bil­i­ty point of view, does it trans­fer over? Sec­ond­ly, we could do some­thing if we had enough mon­ey. So we need to think about what’s finan­cial­ly viable. There are some great things and we can do a lot more on the online envi­ron­ment so we can seg­ment based on per­son­al­i­ty, based on how you look. In a bricks and mor­tar store we don’t have those sort of cri­te­ria, so life gets a lot more com­pli­cat­ed when we try to repli­cate some of these fac­tors in that com­mer­cial sec­tor. So it’s just try­ing to think very care­ful­ly which stud­ies we could go for.

Gareth J. Har­vey:
Test­ing in the com­mer­cial sec­tor has some real­ly inter­est­ing com­pli­ca­tions that we need to be aware of. First of all, we can do a study that is a hun­dred per­cent legal, we have eth­i­cal approval from a uni­ver­si­ty or your com­mer­cial ethics boards, but pub­lic per­cep­tion is very dif­fer­ent. A cou­ple of years ago Dai­ly Mail ran a big cam­paign talk­ing about secret­ly filmed while you shop. Some of cus­tomer’s biggest stores secret­ly track­ing you. It was on the six o’clock news, it was on the 10 o’clock news. Research I was con­duct­ing, I was on page three of the Dai­ly Mail labeled as Dr. Evil. Every­thing we did was ful­ly GDPR com­pli­ant, but the pub­lic per­cep­tion there’s a big uproar about it. Again, there’s a lot more atten­tion to cer­tain things about Cam­bridge Ana­lyt­i­ca, manip­u­la­tion. So we need to think very care­ful­ly about these sort of things.

Gareth J. Har­vey:
You need to be com­fort­able when you do these sort of things, actu­al­ly what are the impli­ca­tions? Now when these sort of head­lines came up, com­mer­cial­ly I made a big dif­fer­ence. I had a very inter­est­ing meet­ing with the vice chan­cel­lor of my uni­ver­si­ty, try­ing to explain why I was called, Dr. Evil by the Dai­ly Mail. Prob­a­bly one of my favorite head­lines about me ever, but we need to be aware of them and think about the neg­a­tive impli­ca­tion of doing these sort of research. Com­mer­cial research, again, the time­frames and what com­pa­nies are expect­ing is total­ly dif­fer­ent. So if I do research, a good exam­ple I did an fMRI project a cou­ple of years ago. The point where the uni­ver­si­ty gave us eth­i­cal approval to run the study, the client who was pay­ing for this expect­ed the full project to be completed.

Gareth J. Har­vey:
They were expect­ing the project to go from com­mis­sion­ing to deliv­ery with­in about four weeks. They did­n’t care about paper, that was irrel­e­vant. They want­ed a one-page sum­ma­ry and they just want­ed an oral pre­sen­ta­tion of the results. That’s all they need­ed to make a deci­sion. So it’s try­ing to think about some of them. So there’s slight dif­fer­ences there. The oth­er thing is when you’re work­ing with these com­pa­nies actu­al­ly try­ing to work out how you get access to the data can be tricky. You may be being paid by a live super­mar­ket or a major brand, they want you to make sure you devel­op an inter­ven­tion that will increase sales. But they won’t give you the sales data to val­i­date whether you’re right or wrong, it’s com­mer­cial­ly sen­si­tive. So there’s some real­ly big issues around those sort of things. So actu­al­ly, how can you come up with a proxy to see what increas­es sales? How do you get access to cer­tain data? It’s challenging.

Gareth J. Har­vey:
I’m run­ning field tri­als, noise is part of that response. These ran­dom events make life fun. So for exam­ple, once I was run­ning a project and col­lect­ing data in Liv­er­pool, whole day’s film­ing in the store cost­ing mul­ti­ple tens of thou­sands of pounds, so well over 30,000. I’m not the biggest foot­ball fan. Liv­er­pool were play­ing in the FA cup, con­se­quent­ly from that day we had hard­ly any peo­ple in store, it was a total­ly dif­fer­ent sort of envi­ron­ment and it made a dif­fer­ence in how peo­ple attend­ed. But whole day’s worth of film­ing was wiped out because I failed to check the foot­ball sched­ule. Things like when it’s extreme weath­er, heat or rain, it changes where peo­ple go to the store and you can’t just gen­er­al­ize if you’re col­lect­ing data from one day, so we need to think more about it. And actu­al­ly repli­ca­tion is so key in this com­mer­cial research.

Gareth J. Har­vey:
If you think about how we start­ed off doing the research where we’re say­ing, lab-based study one store in the North­east of Eng­land and then we repli­cat­ed in South Africa. Well if you’re work­ing for a brand, how many brand or how many stores are you going to test in? Are you going to be test­ing across mul­ti­ple for­mats? How many dif­fer­ent geo­graph­ic regions are you going to test in before you’re con­fi­dent in your results? Because we do get big, large amount of vari­a­tions. And it’s try­ing to think, okay, what are we going to accept? How much vari­a­tion can you accept? What are the com­pe­tent appli­ca­tions and the time­frames to deliv­er? So there are some unique chal­lenges. But for me, it’s one of the most excit­ing things to do. But there are cer­tain things that we real­ly need to con­sid­er before you take or if you are going to work commercially.

Gareth J. Har­vey:
You can get the great ideas, but the things are you’ll prob­a­bly nev­er get any papers with doing this sort of research. If a com­pa­ny is pay­ing you to do this research, unless it’s going to be part of their pub­lic­i­ty stunt, you will have to sign a non-dis­clo­sure agree­ment. Why would they want you to pub­lish a paper that their com­peti­tors could get that research and they could get it for the price of £30 to buy that one paper, when that client has spent £20,000 on this one project? It’s a bit of a dif­fer­ence, so it does change how you do research. There’s also some real­ly inter­est­ing impli­ca­tions over your legal respon­si­bil­i­ties when you do com­mer­cial work. And it’s some­thing as psy­chol­o­gists we tend to not to think about, what did you actu­al­ly promise the client? Now you may say I’ve got this great idea, it will be able to increase sales by X percent.

Gareth J. Har­vey:
Now your con­tract that you actu­al­ly sign may not say that it’s just a research pro­pos­al, but any­thing that you say about in your mar­ket­ing mate­r­i­al, any phone calls, emails about it they actu­al­ly can be part of the con­tract. And legal­ly you can get into some dodgy grounds if you’re not care­ful. So if you’re a behav­ioral sci­ence and you’re work­ing com­mer­cial­ly, it’s a real­ly impor­tant idea to actu­al­ly start to under­stand what are your legal respon­si­bil­i­ties. Things are going to go wrong and it’s some­thing to be aware of. So it’s a case of try­ing to work out well, what hap­pens if some­thing goes wrong? Are you liable? Is it a case of the uni­ver­si­ty is going to be liable? Or is it your com­pa­ny? And if it’s your com­pa­ny, is there a dif­fer­ence between you and your com­pa­ny? Are you the same legal enti­ty? And it’s real­ly impor­tant to start work­ing out what are those dif­fer­ences, because yeah, it has some fair­ly seri­ous impli­ca­tions for you.

Gareth J. Har­vey:
And the final one to start bear­ing in mind is, I do com­mer­cial research because it’s great fun, I also quite like get­ting paid for it. Actu­al­ly try­ing to work at how you pay or get paid for com­mer­cial work, it’s a real­ly dif­fer­ent sort of ball game. What do you want to get paid for? Is it just based on hours worked? Prob­a­bly not. Is it more, we’re more like­ly to be focus­ing on how much val­ue added do you get to the project? My pre­ferred way of actu­al­ly try­ing to get paid is by results. So for exam­ple, if I can increase sales by 1%, or 2%, what­ev­er I will get a cut out of that. Now gen­er­al­ly speak­ing, lawyers and accoun­tants real­ly don’t like that because if we rolled it out glob­al­ly, that can be an awful lot of mon­ey very quick­ly. So nor­mal­ly some accoun­tant steps in and just pays me a flat fee. But it’s a great mod­el to start look­ing at.

Gareth J. Har­vey:
The chal­lenges if you’re using it though is, how can you make sure that you actu­al­ly trust them? So, for exam­ple, how do you know that if you’ve actu­al­ly been able to increase sales by a cer­tain per­cent­age? So it’s a case of, can you get access to that data because com­pa­nies aren’t will­ing to share that data with you most of the time. So it’s one of those sorts of things to think about. But it’s if you want to take your research, because we’ve got some great insights in psy­chol­o­gy, think about how we can take them out. So that’s every­thing I was just going to quick­ly men­tion. If you’ve got any ques­tions, feel free to email me them per­son­al­ly. My details are some­where there. And thank you. Hope you found it vague­ly interesting.

Speak­er 2:
Oh, absolute­ly. Incred­i­bly inter­est­ing. Thank you ever so much Gareth. For every­one in the chat. Can you let us know if you’ve found that inter­est­ing. Write inter­est­ing in the chat, tell us how you feel about it. If you have ques­tions, put them into the ques­tion and answer sec­tion. Alex is going, sor­ry. Gareth is going to be able answer these while we’re here basi­cal­ly. I think one ques­tion that would be real­ly use­ful for a lot of peo­ple here, would you be able to say some­thing about how, as I think a lot of our audi­ence are psy­chol­o­gists, how would they get into the field of con­sumer psy­chol­o­gy? How do they start get­ting into these kinds of jobs?

Gareth J. Har­vey:
So I start­ed off actu­al­ly as part of my PhD. So it was a case of I approached an adver­tis­ing agency I was inter­est­ed in work­ing with them and they actu­al­ly end­ed up invit­ing me to a meet­ing with Cial­di­ni, Robert Cial­di­ni. Which was a heck of a great start as a PhD stu­dent. And then they intro­duced me to com­pa­nies. The prob­lem is there’s so many ques­tions com­pa­nies have but they don’t think of them as psy­chol­o­gy based ques­tions. So it’s a case of actu­al­ly you’ve got knowl­edge if it’s about what grab some­one’s atten­tion, that’s great for adver­tis­ing. How do peo­ple learn a mes­sage? Well that’s mem­o­ry, again it’s on domain of psychology.

Gareth J. Har­vey:
But in a busi­ness school they nev­er get taught it, a mar­ket­ing man­ag­er often does­n’t think about it. So there’s noth­ing wrong with actu­al­ly going out and sell­ing your­self or approach­ing com­pa­nies to start doing that. The key is actu­al­ly think­ing about how can you take what you know and speak in a lan­guage that’s rel­e­vant. Because they’re not going to want to paper it you’re going to have to be able to speak in the lan­guage that’s rel­e­vant, and pos­si­bly con­dense it and make it into a for­mat that they’re will­ing to understand.

Speak­er 2:
I think that’s great. I think Pasquale, if you can get your slides up and just one final ques­tion, is there a book or a text or a web­site you could rec­om­mend for peo­ple who are inter­est­ed in learn­ing how to do that?

Gareth J. Har­vey:
I think prob­a­bly one of the best places to start off is Richard Shot­ton’s book, The Choice Fac­to­ry. So it just goes through some of the 24 basic behav­ioral bias­es that we use as a start­ing point, but they’re the ones that most peo­ple know. Once you’ve known those, what real­ly dif­fer­en­ti­ates you is actu­al­ly under­stand­ing the oth­er things. So that’s why we start look­ing at things such as the Jour­nal of Con­sumer Psy­chol­o­gy and actu­al­ly just being aware of what’s being pub­lished all the time. Because there are some real­ly cool things that are hap­pen­ing each year but they just don’t make it into the Pop­u­lar Sci­ence press.

 

Get on the Waitlist

BeOnline is the conference to learn all about online behavioral research. It's the ideal place to discover the challenges and benefits of online research and to learn from pioneers. If that sounds interesting to you, then click the button below and sign up for our newsletter. You will be the first to know when we release new content and open applications for BeOnline 2022.

With thanks to our sponsors!

Escap­ing the lab­o­ra­to­ry: Apply­ing lab­o­ra­to­ry find­ings in the com­mer­cial world