September 21, 2025 • 28 mins
Tom Coe, VP of Growth at Retail Insight, shares how data-driven solutions can tackle the perennial challenges of retail operations while empowering store colleagues to deliver better customer service
• Retail Insight provides in-store operations analytics software for over 50,000 stores globally
• The complexity of modern retail operations creates compliance challenges as colleagues juggle multiple responsibilities
• Process execution suffers when labour models are tight and colleagues are overwhelmed with competing priorities
• Date checking for fresh products exemplifies a mundane but critical process where poor execution leads to waste and financial loss
• Smart systems can track inventory expiration dates and direct colleagues to check only specific items requiring attention
• Data collected provides valuable analytics for head office teams to improve forecasting, ranging, and supply chain decisions
• Successful AI implementation in retail requires balancing technological sophistication with practical simplicity
• Data quality is essential for any effective retail technology solution
• Future applications include democratizing analytics through conversational AI and enhancing root cause analysis
#theproductivityexperts
Register for the 2026 Productivity Forum
Find us in the Top 50 Productivity Podcasts
Connect to Simon on LinkedIn
Follow ReThink on LinkedIn
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:05):
Welcome to the Productivity Podcast.
Today, I'm delighted to bejoined by Tom Coe, vp of Growth
at Retail Insights.
Hi, tom.
Speaker 2 (00:14):
Hey Simon, how are you doing?
Speaker 1 (00:15):
Yeah, good, thanks you.
Speaker 2 (00:16):
Yeah, doing good despite the rain outside.
Speaker 1 (00:19):
Yeah, well you can tell we're recording in British
summertime because the kids arebroken up and guess what?
It's started to rain, exactly.
We'll give you something tolisten to for all those
listening on journeys to andfrom the office or while the
kids are in the playground andstuff.
We'll give you some interestinginsight in the conversation
with tom.
So, tom, tell us a bit aboutyourself first, before we talk
(00:40):
about retail insights.
How did you get to be vp ofgrowth at retail insights and
what have you done before?
Speaker 2 (00:45):
yeah, so as you say, vp of Growth at Retail Insight,
I spent most of my career at thesort of intersection of
commercial strategy, operationsand technology.
I'm really focused on helpingretailers and suppliers unlock
performance through data, ai,beta execution, those sorts of
(01:06):
things.
I've been at Retail retailinside about five years now.
Before that, I was in America,used to be a runner, did my MBA
out there when I was a youngerman, I worked at Sainsbury's for
a number of years.
You could say retail wassomething I've been around a
long, long time.
(01:27):
My job now is really aboutdelivering as much value as
possible for our customers, bothwithin what we do for them
already and also how we could domore I suppose newer things
with them.
I really want to be able to askthe question how can we
position Retail Insight with allof our senior and retail
operations stakeholders?
(01:48):
How can we position Retailretail insight as the guys that
you call when you've got aproblem you can't solve
internally, so you need to solveit through data analytics?
Well, I'm going to give calmand retail insight a call and
see if they can support me that.
Speaker 1 (02:01):
That's really what it's focused on right now
perfect, so worked in differentcountries, which I always think
is fascinating and brings somedifferent perspectives on life,
doesn't it?
So just talk to me a bit aboutrunning.
So was that a uni thing?
Was that a Forrest Gump typething?
How did that work?
Speaker 2 (02:18):
It was a university thing, university thing.
I was a keen athlete most of mylife.
I wouldn't say anymore I'm notmuch of an athlete, but running
was what I did at university.
I went to university ofBirmingham sort of a big
heritage as a great athleticsand cross-country school and
managed to go out to America fora couple years did a master's
(02:38):
and I was.
I was a steady distance rightlike a middle distance runner.
But yeah, in a former life nowa former life, it was like a
long time ago.
Speaker 1 (02:47):
Excellent, there you go, there's your, there's your
fact about Tom.
If you meet him anywhere, youcan talk to him about running.
So on onto the onto theimportant stuff then.
So you've touched kind ofbriefly on you want retail
insights to be the kind of go-tofor all those organizations
that that want support aroundwhatever it might be in terms of
data insights process.
But give us a bit more flavourabout what you guys do, maybe
(03:12):
some of the tech you've got,before we get into the detail of
what we're going to talk abouttoday.
Speaker 2 (03:17):
Yeah for sure.
So Retail Insight I'd positionus as the world's leading
in-store operations analyticssoftware provider for retail.
In other words, we're aboutapplying really smart, pragmatic
technology to help tackle theperennial issues of good
shopkeeping.
You know, questions like do Ihave product on the shelf?
(03:39):
Is my data accurate?
How do I stop product goinginto the bin these sorts of
questions that retailers from asingle corner shop on the road
that we all live in to theWalmarts of the world Everybody
asks them and everybody grappleswith them.
To achieve this, we have anarsenal of analytics products
(04:01):
and these are a mixture ofreally, really intelligent
advanced mathematics and some ofthe leading edge technology AI,
ml, real-time streaming andinfrastructure and these cover
things like prompt availability,waste and markdown, phantom
imagery, correction anddetection, online fulfillment,
(04:23):
maximization and a few otherpieces around that, and really
they are very complex productsin terms of the output and what
they actually do, but they'reremarkably simple in terms of
setup.
They need the most basic datapoints out of the EPOS system.
A retailer has four to fivedata points and you can set it
(04:45):
up and you can have an outputwithin 10 to 14 days, which is
pretty unprecedented in the inthe market.
We do this with over 50 000stores globally right now.
This includes people like thecarp in the uk sprouts market in
in north america.
Soon that'll be at sort of 60kplus and, yeah, real heritage.
(05:09):
It's been something we've donefor a very, very long time and I
think we are really seen asexperts in the field and just
continue to double down on thatgrow, continue to focus on what
makes us great and expand outwhat we can do for for our
customers brilliant, so that, Imean, some of those initial
questions are really big onesthat you know, people have been
struggling with for a number ofyears, haven't they?
Speaker 1 (05:31):
but I suppose, more so in the current cross, cross,
cross cost climate and we talk alot on this podcast about, you
know, the dni increases andnational living wage and stuff,
and they're all real and they'reall.
They're all meant with theright intent, but they're all a
cost to any organization,specifically where you've got a
big number of people which tendsto be in the customer facing
(05:54):
environment.
So it's a true challenge, forsure.
So, one of the things we weregoing to focus on today and yes,
we will talk about ai as we weget to the end of the podcast,
as it feels like we're obligedto do, but we were going to kind
of focus in on compliance and,like you said, you worked in
Sainsbury's when you wereyounger and I've worked in
(06:15):
retail all my life.
It is the holy grail in retail,always has been, probably
always will be, is the reality,and for those that work in
retail, I'll get that instantly.
For those that may be listeningthat don't.
Why is that?
Well, you've got lots of that.
Well, you've got lots of people,you've got lots of physical
locations.
You've then got got what I'dcall and I don't mean this
(06:37):
disrespectfully a random factorof customers, because some will
interact with you, some won't,some will be nice, some won't.
But you've also got peoplemaking decisions.
So a store manager, unless youcan really help them with a
process, will make a differentdecision in store A than store B
, than store C, so on, so on.
So when you're speaking withcustomers and your team, what do
(07:00):
you typically see in that spacearound?
Lack of compliance or whatdrives poor compliance?
Speaker 2 (07:10):
Yeah, you're absolutely right.
It's a huge challenge,something that, if I think back
to when I worked in a store manymoons ago, you could almost see
the compliance challenge as youworked on the shop floor
alongside your colleagues,people that maybe weren't
engaged as much on the job asthe retailer probably wants them
(07:32):
to.
And that's a massive, massivechallenge and tackling it is not
easy, I think, particularly inan era where, as you say, labour
models are incredibly tight.
You know razor, razor thin interms of tightness at the moment
.
So ensuring that the resourceyou have deployed is executing
the process, the tasks, the waysof working that you need them
(07:53):
to is has never been moreimportant, because the money
you're spending on this laboryou need them to do the job you
need them to do.
Otherwise that's just a sunkcost.
I think there's a couplereasons for for poor that we see
.
No doubt there's many more, butthese are sort of the big ones
that come to mind.
(08:13):
I think there is a challengearound process in-store and the
job these associates andcolleagues need to do is really
complex.
It's overcomplicated.
Often when I worked in-store itwas a bit simpler.
You were an associate or acolleague who worked in dry
grocery.
So you knew the 10, 12 hoursyou worked on the products, the
(08:37):
process, the ways of working,the picking and binning out the
back, how you replenished, howyou took the products back out,
how you dealt with the customers, et cetera.
With less labor on the floor,colleagues now need to do way
more.
You might need to replenishproducts, you might need to
accept deliveries, you mightneed to shop online orders, you
might need to jump onto theself-checkouts, you might need
(08:58):
to go and collect trolleys andamongst all that, there's new
process around self-scanning astore and things like this,
which is making it very noisyaround process and colleagues,
ultimately, are trying to figureout well, what am I supposed to
focus on, what are mypriorities?
And it's not a surprise thatthings get lost because there's
(09:21):
just so much to focus on and it.
Speaker 1 (09:24):
I found that interesting because we've been
on a journey again as far backas I can remember of
simplification, simple stores,one best way.
I'm just quoting all the kindof strapline across the years
from various differentorganizations.
Yet yet you're right, becausewhen you reflect actually
(09:45):
multi-skilling, a get.
Because why do you want justpeople set on checkout, somebody
you can just work, sat oncheckout somebody you can just
work in delhi, somebody you canjust work in ambient whatever,
because it limits yourflexibility with a smaller pool
of people.
But then we've kind of enabledsome of these people with tech
that that might be a challenge.
They're not used to tech, don'twork with tech outside of work.
Then we've given them mostinstances delivery, uber, eats
(10:09):
on demand, click and collect,picking goods, returning goods,
layer on top, then shrink theft.
You'll have everyone listening,will have seen it on various
social media platforms andyou'll see lots of colleagues
wearing the cameras now.
So the environment you work inis tricky as well.
It's interesting that we'veprobably ended up in some
(10:31):
respects in a slightly morecomplex world 10-15 years after
everybody's been on asimplification journey yeah,
yeah, you're actually right, Ithink, a more complex world with
less labor.
Speaker 2 (10:44):
To deliver against that complexity, I think you
could have had some amount ofsegmentation or separation of
responsibilities.
When this person's going to besolely responsible for, you know
, merchandising, planning andonline ordering, this person's
be solely responsible for freshand produce, but now it's the
world's a bit um a bit morechallenging.
(11:04):
The store people are doingwearing multiple hats and doing
multiple, multiple things, whichis which is very difficult,
very difficult and are there anyspecific examples you can share
with us without kind of namingorganizations?
I think a great example is sortof linking maybe the mundanity
of some process into thecomplexity.
(11:25):
If you think about date checkingfor fresh products, it's a
classic example.
If you work in a big sort ofsuperstore hypermarket, you're
going to have five, six, sevenaisles full of fresh products
and your responsibility is toindividually check every product
to make sure that a it's notgone out of date yet and b it's
(11:47):
got a markdown, and making sureyou then apply that markdown and
then later in the day youprobably need to check it again
to make sure the products thathave a markdown have either sold
or they get a second markdownor maybe even a third markdown.
Amongst all that, you've alsogot to rotate products.
I know as customers we often goin and buy the the longest,
(12:07):
longest sort of date item anyway, but colleagues bring forward
the earliest date to try andencourage velocity on those
first.
So I think that's a classicexample where it's just a very
manual process and it's easy tounderstand, as I explained it
there, manual date checking.
It's not a surprise thatcolleagues are disengaged and
(12:29):
compliance is poor and you seeinstances where retailers get
fined and trading standards areinvolved, all these sorts of
things.
Speaker 1 (12:37):
And ultimately I know we talked about cost quite a
bit.
That poor, mundane process ifyou're not kind of doing it
properly and we'll talk in asecond about how you can fix it
costs you money becauseultimately you'll end up getting
fined, you'll end updissatisfying customers and
throwing the product and writingthe cost off.
So you know, if we just workthat through, somebody's made
(13:01):
the product, that company'sbought it.
They've then shipped it to thewarehouse, the warehouse have
shipped it to the store, thestore have put it on the shelf,
probably touched it a couple oftimes, and then we throw it away
.
So it's not just it cost meI'll make it up four pound 99
trifle, it's the layers ofsalary that go before that to
physically get it on the shelfthat you paid for.
(13:21):
Yeah, you're absolutely right,you're absolutely right.
So so, fixing it then, so thatwe'll all have seen that, we'll
all have seen yellow labels, redlabels, whatever organization
you're working in, this, thiskind of supermarket example, how
can you guys help fix it?
Speaker 2 (13:39):
I think, a level.
There's an amount of diagnosticand process mapping.
Where are colleagues spending alot of their time doing things
that's really high frequency andcould be automated or improved
or augmented with technology?
We talk about the idea oftechnology augmenting process in
store.
How can you create almost abionic or superhuman associate
(14:04):
that's got all the data, theinsight, the technology at their
hands to perform process to themaximum so ultimately they're
freed up to serve and sell?
Because that remains the bigdifferentiator for how retailers
can win on on manual datechecking.
The classic example would be canyou just build a ledger which
is essentially inventory holdingof what products you have when
(14:29):
those items expire, and then youwork that into your markdown
process, which you probablyalready have?
You clearly already have to saythat you know today the first
markdown, I need to scan fiveindividual skews of vanilla
cheesecake because I can seethey're expiring today, on the
31st, and then tomorrow I needto scan these items and then you
(14:50):
check in new items as they comeand it's a self-updating ledger
system that colleagues areresponsible for.
So they're buying into theprocess.
They know that good processexecution is going to make their
job maybe this afternoon ortomorrow afternoon easier.
So there's an amount of sort ofbuy-in because they know that
they're going to make theirlives a bit easier there.
(15:10):
I think it's just about quitesimple technology that removes
some of that mundanity, makes ita bit quicker, a bit easier and
ultimately gets colleagues abit more engaged in that process
as well then that must give abenefit.
Speaker 1 (15:27):
So they kind of using that through your software.
That must give a benefit.
A central level, ie the headoffice, there must be some stats
or management information thatthey can glean from that yeah, a
huge amount of insights andanalytics off the back.
Speaker 2 (15:41):
You can see what individuals, what stores, what
times of day these processes arehappening.
So you know you typically willhave a markdown period happen
between hours X and hours Y.
You can realize when thoseprocesses are happening out of
hours.
If they happen too late you'vegot less time to sell products.
If they happen too early, youmight be cannibalizing full
(16:04):
price sales that you might haveotherwise got and then, beyond
that, bringing in more ways tosee what colleagues are facing
in store.
Are they dealing with too manymarkdowns?
Do I need to edit myforecasting replenishment plans?
Do I need to figure out betterways to improve that
replenishment process to supportless waste on the shelf so
(16:27):
there's less going into themarkdown process at the end of
it.
So a huge, huge amount.
That's just a lot of scratchingthe surface.
Speaker 1 (16:32):
A huge amount of opportunity to explore from a,
you know, insights and analyticsperspective on that side yeah,
I mean in simple terms, you canhelp reduce the volume that gets
thrown in the bin, so that's afull on cost and make sure that
the markdowns are intelligent,priced correctly, based on the
you know vanilla cheesecakeexample so they actually sell
(16:55):
through.
Then, stepping a stage backfrom that, stop, stop people
checking every item every day.
So we're just targeting theones that we know that need
reducing.
So there's a huge labor free upthere.
But then, yeah, back to yourfinal point.
The analytics then can start tobe interrogated to say, well,
(17:15):
why are we always reducing thevanilla cheesecake?
Do we need to ship it in threesrather than sixes?
Do we need to sell it at allbecause we reduce more than we
ever sell at full price?
There's different bits ofinformation for kind of the ops
team, the buying team, theranging team, et cetera, that
they can start to work with.
Is that right?
Speaker 2 (17:34):
Yeah, absolutely.
It's that root causing, youknow, diagnosing what issues
consistently happen on the shopfloor.
And then how can I make broader, enterprise-wide decisions to
remove those challenges in thefuture?
Do I need to change the range?
Do I need to alter the forecastand demand plan?
Do I need to change thepromotional plan?
Do I need to work withsuppliers to alter the case pack
(17:55):
that comes to store?
Those are just a few of theexamples that could be taken at
head office level to helpimprove things for the colleague
on the shop floor and,ultimately, the customer.
Speaker 1 (18:07):
Good, so exactly let's, let's talk about ai.
It feels like we have to talkabout it on every, every podcast
.
So what kind of are you seeingin the world of, specifically in
retail ai?
So any good examples you canshare, any poor examples.
Speaker 2 (18:24):
Yeah, yeah, lots, lots in both camps, I think AI
and retail operations.
I wrote a piece about this acouple of months ago.
It's really challenging becausethere's so many issues on the
shop floor.
You have to compete with thechaos of the shop floor, the
lack of the shop floor, the lackof data quality.
Getting around those aremassive, massive problems.
(18:49):
I'm yet to see fully-scaled AIthat is autonomous and just
deals with those perfectly andhas no issues.
It doesn't really exist.
I think the best examples, themost successful implementations
of technology that has, aibalances the sophistication of
(19:09):
those technologies and thesimplicity.
So you can have the mostcomplex, advanced machine
learning model in the world, butcan you deliver it to a
colleague in the store in asimple, digestible way that, as
we're just talking about, a theyengage with, b makes them buy
into the process and c directsthem to do a task that they can
(19:35):
achieve, and then maybe d?
Can you see the impact of thetask that they've just done on
the shop floor?
So an example might be ourintelligent inventory alerting
system, inventory insight, whichis basically about flagging
potential instances of phantomor shadow inventory, essentially
where there's a mismatchbetween pi and what you actually
(19:56):
have in the store.
This is a really advancedsystem.
It's got, I think, 20 plusmachine learning models that
balance and make predictionsabout your imagery position
across the entire shop.
But it brings in really quiteelegant and pragmatic analytics
layer and decisions system whichproduces insight in a way that
(20:22):
is quite human and I thinkthat's really important, instead
of just issuing a generic stockcommand around.
Go and check this stock, simon.
It's going to give the context.
This item has not sold for fivedays.
It's got 20 units on hand.
It sold 15 units yesterday.
(20:44):
We think there's an issue reallysimple, digestible colleagues
in store probably understand andappreciate why they've been
alerted to that item to check it.
Because I think that's a bigissue that colleagues will get
alerts from systems in store ortechnologies and eventually so
many bad calls.
You're just going to think thatevery call you get from this
(21:06):
system in the future is going tobe a bad call.
You completely lose the storeteams.
So I think that's a goodexample.
A bad example.
I think a lot of peopleretailers and some vendors will
deploy dynamic markdown systems.
We've got a dynamic markdowncapability at retail insight, so
(21:26):
this is just about capturing asmuch margin as possible whilst
minimizing waste throughexploration-based discounting.
Yeah, a lot of the technologiesand systems, and you can go and
see some of this and experienceit in some stores, but they'll
deploy machine learning modelsthat consider loads of factors,
(21:49):
for example, the volume of fullprice items in store.
How is that going tocannibalize discounted items?
But they really fail toappreciate, I think, the full
value chain in retail ops.
So these systems might assumethat the on-hand system, the
on-hand count for these items,is correct and it will make
(22:09):
discounts based on the fact thatI've got 20 items of that
cheesecake that we talked aboutearlier, but really 12 of those
items are phantom and yoursystem's not correct and it's
making decisions based oninaccurate, flawed data.
And we can all know what theoutcome is going to be
discount's going to be all overthe place and the chances are
(22:31):
you're either not going to sellit and throw away a great
product, or you're going to sellit at a massive discount and
potentially lose a lot of money.
So I think it's how systemsthat are really intelligent rely
on data that is flawed, isinaccurate.
I think that's the big thingthat we started to talk a lot
about in the last sort of 12months is there is a connected
(22:54):
tissue between our suite ofproducts.
How can availabilities lost,sales drive figuring out what
items that have phantom imagery,and then how can phantom
imagery level set on hands instore to drive the right
discounting levels within ourmarkdown system.
(23:16):
And there's sort of this ebband flow between all of them
that feed into each other.
Speaker 1 (23:21):
The key and it's, I suppose, something that rethink.
We've we've learned and there'sthis big unknown, I suppose,
about ai and it's moving soquickly.
The bit that people don't talkabout is and I think you've just
hit the nail on the head is thequality of data that's needed
to drive, drive any algorithm,any system, ai or not?
(23:44):
Yes, and the consequence ofhaving that data not to the
quality it needs.
And again, I think most peoplelistening to this that work in a
retailer will have questionsabout data sources, quality of
data, consistency of data.
And until people get a grip onthat and you know it's
(24:07):
gold-starred and we all knowwhere it comes from and the
sources are reliable andanomalies are detected and moved
out or dealt with we're notgoing to harness it to its full
potential.
Because the software is there.
It's like anything.
It's only as good as what youfeed in, right?
Speaker 2 (24:24):
Yeah.
Speaker 1 (24:26):
Yeah, you're absolutely right.
Okay, so we'll just close onwhere you you're.
You are retail insights interms of using ai in products.
How you're thinking about itmight shape your products and
therefore your offerings tocustomers in the future.
Speaker 2 (24:44):
It's a great question , I think, across a number of
guises, there's probably twothat really stand out.
I think the first is how AI candemocratize access to our
insights and analytics acrossorganizations.
I think, historically, bianalytics has been served and is
customized in a way that worksfor the retailers, but AI
(25:05):
unlocks that and creates almostself-serve for us, and I think
creating giving our customersthe chance to speak to our
system essentially and have asemantic layer in place which
makes that conversation possible, and an AI system that can
produce a really great outputthroughout that conversation is
(25:27):
powerful.
You know, if I'm sat down, I'ma retail operations leader, a
customer, and I've got ouravailability system, and I'm sat
in a store wondering why whyavailability is so poor in this
store, because the sales arefantastic, had a brilliant week,
and you can have a conversationwith our system, understand
what's driving it, and I thinkthat's really exciting, really
(25:49):
exciting.
We hear that as well, andthat's something we're
definitely capitalizing on.
I think the other area is how AIcan to that point I spoke about
earlier around diagnostics.
We're doing some really greatwork around root cause analysis,
so figuring out why out ofstocks keep happening and then
(26:11):
making corrective action at ahead office level, an upstream
level, that essentially cancelsthat out in the future.
Ai is an amazing augmentationof that process to help us move
through that root cause treemuch quicker, much more
elegantly and, I think, get toan answer that retailers are
happy and comfortable with muchfaster, and I think will mean
(26:32):
that what is a very complexsystem in terms of root cause
analysis there's so many reasonswhy this might be happening can
be deployed, managed and usedin a far more efficient way,
easier way in the future, andthat's really exciting.
I think those are probably twoexamples that really stick out,
(26:54):
but there's many, there's many.
Speaker 1 (26:55):
Yeah, no, I like those.
I'm just thinking back to mystore manager days.
How would I feel if my areamanager or somebody from the
center came in and startedinterrogating the stock system
by asking it questions?
There's nowhere to hide, right?
But ultimately, if you're asmart store manager or team
leader colleague, you'll haveinterrogated it yourself first.
(27:17):
So that surfacing data quicklyis great.
I think again, there'll be loadsof organizations centrally
where they've got theirmonitoring, supply performance
and all those other things thatthat go on.
So surfacing that quickly, youknow, looking at suppliers with
poor, poor supply rates, allthose kind of things.
It'll all be happening now.
I don't think anybody deniesthat.
(27:38):
But getting there quicker,flagging it earlier, predicting
it even, will make such a a bigdifference to that whole
end-to-end state.
Yeah, be interesting to see,but I can only see positives
from it.
Yes, definitely We'll pausethere, tom.
Great conversation, really goodto hear about the great things
(27:59):
you're doing at Retail Insights.
If somebody wants to get intouch with you, find out more,
have a chat about maybe, how youcan help, what's the what and
where are the best places forthat to happen I would say reach
out on on linkedin as a classicone, or or drop us a note on
our website and we can have achat.
Speaker 2 (28:17):
We can come to store, grab a coffee, walk around and
have a chat about what you'refacing, what you're challenged
with, and see how I might beable to help.
Speaker 1 (28:26):
Perfect.
Appreciate you taking the timeout, tom, always great to chat
and we'll catch up soon.
Thanks a lot, simon.
All the best.
Welcome to the Productivity Podcast.
Today, I'm delighted to bejoined by Tom Coe, vp of Growth
at Retail Insights.
Hi, tom.
Speaker 2 (00:14):
Hey Simon, how are you doing?
Speaker 1 (00:15):
Yeah, good, thanks you.
Speaker 2 (00:16):
Yeah, doing good despite the rain outside.
Speaker 1 (00:19):
Yeah, well you can tell we're recording in British
summertime because the kids arebroken up and guess what?
It's started to rain, exactly.
We'll give you something tolisten to for all those
listening on journeys to andfrom the office or while the
kids are in the playground andstuff.
We'll give you some interestinginsight in the conversation
with tom.
So, tom, tell us a bit aboutyourself first, before we talk
(00:40):
about retail insights.
How did you get to be vp ofgrowth at retail insights and
what have you done before?
Speaker 2 (00:45):
yeah, so as you say, vp of Growth at Retail Insight,
I spent most of my career at thesort of intersection of
commercial strategy, operationsand technology.
I'm really focused on helpingretailers and suppliers unlock
performance through data, ai,beta execution, those sorts of
(01:06):
things.
I've been at Retail retailinside about five years now.
Before that, I was in America,used to be a runner, did my MBA
out there when I was a youngerman, I worked at Sainsbury's for
a number of years.
You could say retail wassomething I've been around a
long, long time.
(01:27):
My job now is really aboutdelivering as much value as
possible for our customers, bothwithin what we do for them
already and also how we could domore I suppose newer things
with them.
I really want to be able to askthe question how can we
position Retail Insight with allof our senior and retail
operations stakeholders?
(01:48):
How can we position Retailretail insight as the guys that
you call when you've got aproblem you can't solve
internally, so you need to solveit through data analytics?
Well, I'm going to give calmand retail insight a call and
see if they can support me that.
Speaker 1 (02:01):
That's really what it's focused on right now
perfect, so worked in differentcountries, which I always think
is fascinating and brings somedifferent perspectives on life,
doesn't it?
So just talk to me a bit aboutrunning.
So was that a uni thing?
Was that a Forrest Gump typething?
How did that work?
Speaker 2 (02:18):
It was a university thing, university thing.
I was a keen athlete most of mylife.
I wouldn't say anymore I'm notmuch of an athlete, but running
was what I did at university.
I went to university ofBirmingham sort of a big
heritage as a great athleticsand cross-country school and
managed to go out to America fora couple years did a master's
(02:38):
and I was.
I was a steady distance rightlike a middle distance runner.
But yeah, in a former life nowa former life, it was like a
long time ago.
Speaker 1 (02:47):
Excellent, there you go, there's your, there's your
fact about Tom.
If you meet him anywhere, youcan talk to him about running.
So on onto the onto theimportant stuff then.
So you've touched kind ofbriefly on you want retail
insights to be the kind of go-tofor all those organizations
that that want support aroundwhatever it might be in terms of
data insights process.
But give us a bit more flavourabout what you guys do, maybe
(03:12):
some of the tech you've got,before we get into the detail of
what we're going to talk abouttoday.
Speaker 2 (03:17):
Yeah for sure.
So Retail Insight I'd positionus as the world's leading
in-store operations analyticssoftware provider for retail.
In other words, we're aboutapplying really smart, pragmatic
technology to help tackle theperennial issues of good
shopkeeping.
You know, questions like do Ihave product on the shelf?
(03:39):
Is my data accurate?
How do I stop product goinginto the bin these sorts of
questions that retailers from asingle corner shop on the road
that we all live in to theWalmarts of the world Everybody
asks them and everybody grappleswith them.
To achieve this, we have anarsenal of analytics products
(04:01):
and these are a mixture ofreally, really intelligent
advanced mathematics and some ofthe leading edge technology AI,
ml, real-time streaming andinfrastructure and these cover
things like prompt availability,waste and markdown, phantom
imagery, correction anddetection, online fulfillment,
(04:23):
maximization and a few otherpieces around that, and really
they are very complex productsin terms of the output and what
they actually do, but they'reremarkably simple in terms of
setup.
They need the most basic datapoints out of the EPOS system.
A retailer has four to fivedata points and you can set it
(04:45):
up and you can have an outputwithin 10 to 14 days, which is
pretty unprecedented in the inthe market.
We do this with over 50 000stores globally right now.
This includes people like thecarp in the uk sprouts market in
in north america.
Soon that'll be at sort of 60kplus and, yeah, real heritage.
(05:09):
It's been something we've donefor a very, very long time and I
think we are really seen asexperts in the field and just
continue to double down on thatgrow, continue to focus on what
makes us great and expand outwhat we can do for for our
customers brilliant, so that, Imean, some of those initial
questions are really big onesthat you know, people have been
struggling with for a number ofyears, haven't they?
Speaker 1 (05:31):
but I suppose, more so in the current cross, cross,
cross cost climate and we talk alot on this podcast about, you
know, the dni increases andnational living wage and stuff,
and they're all real and they'reall.
They're all meant with theright intent, but they're all a
cost to any organization,specifically where you've got a
big number of people which tendsto be in the customer facing
(05:54):
environment.
So it's a true challenge, forsure.
So, one of the things we weregoing to focus on today and yes,
we will talk about ai as we weget to the end of the podcast,
as it feels like we're obligedto do, but we were going to kind
of focus in on compliance and,like you said, you worked in
Sainsbury's when you wereyounger and I've worked in
(06:15):
retail all my life.
It is the holy grail in retail,always has been, probably
always will be, is the reality,and for those that work in
retail, I'll get that instantly.
For those that may be listeningthat don't.
Why is that?
Well, you've got lots of that.
Well, you've got lots of people,you've got lots of physical
locations.
You've then got got what I'dcall and I don't mean this
(06:37):
disrespectfully a random factorof customers, because some will
interact with you, some won't,some will be nice, some won't.
But you've also got peoplemaking decisions.
So a store manager, unless youcan really help them with a
process, will make a differentdecision in store A than store B
, than store C, so on, so on.
So when you're speaking withcustomers and your team, what do
(07:00):
you typically see in that spacearound?
Lack of compliance or whatdrives poor compliance?
Speaker 2 (07:10):
Yeah, you're absolutely right.
It's a huge challenge,something that, if I think back
to when I worked in a store manymoons ago, you could almost see
the compliance challenge as youworked on the shop floor
alongside your colleagues,people that maybe weren't
engaged as much on the job asthe retailer probably wants them
(07:32):
to.
And that's a massive, massivechallenge and tackling it is not
easy, I think, particularly inan era where, as you say, labour
models are incredibly tight.
You know razor, razor thin interms of tightness at the moment
.
So ensuring that the resourceyou have deployed is executing
the process, the tasks, the waysof working that you need them
(07:53):
to is has never been moreimportant, because the money
you're spending on this laboryou need them to do the job you
need them to do.
Otherwise that's just a sunkcost.
I think there's a couplereasons for for poor that we see
.
No doubt there's many more, butthese are sort of the big ones
that come to mind.
(08:13):
I think there is a challengearound process in-store and the
job these associates andcolleagues need to do is really
complex.
It's overcomplicated.
Often when I worked in-store itwas a bit simpler.
You were an associate or acolleague who worked in dry
grocery.
So you knew the 10, 12 hoursyou worked on the products, the
(08:37):
process, the ways of working,the picking and binning out the
back, how you replenished, howyou took the products back out,
how you dealt with the customers, et cetera.
With less labor on the floor,colleagues now need to do way
more.
You might need to replenishproducts, you might need to
accept deliveries, you mightneed to shop online orders, you
might need to jump onto theself-checkouts, you might need
(08:58):
to go and collect trolleys andamongst all that, there's new
process around self-scanning astore and things like this,
which is making it very noisyaround process and colleagues,
ultimately, are trying to figureout well, what am I supposed to
focus on, what are mypriorities?
And it's not a surprise thatthings get lost because there's
(09:21):
just so much to focus on and it.
Speaker 1 (09:24):
I found that interesting because we've been
on a journey again as far backas I can remember of
simplification, simple stores,one best way.
I'm just quoting all the kindof strapline across the years
from various differentorganizations.
Yet yet you're right, becausewhen you reflect actually
(09:45):
multi-skilling, a get.
Because why do you want justpeople set on checkout, somebody
you can just work, sat oncheckout somebody you can just
work in delhi, somebody you canjust work in ambient whatever,
because it limits yourflexibility with a smaller pool
of people.
But then we've kind of enabledsome of these people with tech
that that might be a challenge.
They're not used to tech, don'twork with tech outside of work.
Then we've given them mostinstances delivery, uber, eats
(10:09):
on demand, click and collect,picking goods, returning goods,
layer on top, then shrink theft.
You'll have everyone listening,will have seen it on various
social media platforms andyou'll see lots of colleagues
wearing the cameras now.
So the environment you work inis tricky as well.
It's interesting that we'veprobably ended up in some
(10:31):
respects in a slightly morecomplex world 10-15 years after
everybody's been on asimplification journey yeah,
yeah, you're actually right, Ithink, a more complex world with
less labor.
Speaker 2 (10:44):
To deliver against that complexity, I think you
could have had some amount ofsegmentation or separation of
responsibilities.
When this person's going to besolely responsible for, you know
, merchandising, planning andonline ordering, this person's
be solely responsible for freshand produce, but now it's the
world's a bit um a bit morechallenging.
(11:04):
The store people are doingwearing multiple hats and doing
multiple, multiple things, whichis which is very difficult,
very difficult and are there anyspecific examples you can share
with us without kind of namingorganizations?
I think a great example is sortof linking maybe the mundanity
of some process into thecomplexity.
(11:25):
If you think about date checkingfor fresh products, it's a
classic example.
If you work in a big sort ofsuperstore hypermarket, you're
going to have five, six, sevenaisles full of fresh products
and your responsibility is toindividually check every product
to make sure that a it's notgone out of date yet and b it's
(11:47):
got a markdown, and making sureyou then apply that markdown and
then later in the day youprobably need to check it again
to make sure the products thathave a markdown have either sold
or they get a second markdownor maybe even a third markdown.
Amongst all that, you've alsogot to rotate products.
I know as customers we often goin and buy the the longest,
(12:07):
longest sort of date item anyway, but colleagues bring forward
the earliest date to try andencourage velocity on those
first.
So I think that's a classicexample where it's just a very
manual process and it's easy tounderstand, as I explained it
there, manual date checking.
It's not a surprise thatcolleagues are disengaged and
(12:29):
compliance is poor and you seeinstances where retailers get
fined and trading standards areinvolved, all these sorts of
things.
Speaker 1 (12:37):
And ultimately I know we talked about cost quite a
bit.
That poor, mundane process ifyou're not kind of doing it
properly and we'll talk in asecond about how you can fix it
costs you money becauseultimately you'll end up getting
fined, you'll end updissatisfying customers and
throwing the product and writingthe cost off.
So you know, if we just workthat through, somebody's made
(13:01):
the product, that company'sbought it.
They've then shipped it to thewarehouse, the warehouse have
shipped it to the store, thestore have put it on the shelf,
probably touched it a couple oftimes, and then we throw it away
.
So it's not just it cost meI'll make it up four pound 99
trifle, it's the layers ofsalary that go before that to
physically get it on the shelfthat you paid for.
(13:21):
Yeah, you're absolutely right,you're absolutely right.
So so, fixing it then, so thatwe'll all have seen that, we'll
all have seen yellow labels, redlabels, whatever organization
you're working in, this, thiskind of supermarket example, how
can you guys help fix it?
Speaker 2 (13:39):
I think, a level.
There's an amount of diagnosticand process mapping.
Where are colleagues spending alot of their time doing things
that's really high frequency andcould be automated or improved
or augmented with technology?
We talk about the idea oftechnology augmenting process in
store.
How can you create almost abionic or superhuman associate
(14:04):
that's got all the data, theinsight, the technology at their
hands to perform process to themaximum so ultimately they're
freed up to serve and sell?
Because that remains the bigdifferentiator for how retailers
can win on on manual datechecking.
The classic example would be canyou just build a ledger which
is essentially inventory holdingof what products you have when
(14:29):
those items expire, and then youwork that into your markdown
process, which you probablyalready have?
You clearly already have to saythat you know today the first
markdown, I need to scan fiveindividual skews of vanilla
cheesecake because I can seethey're expiring today, on the
31st, and then tomorrow I needto scan these items and then you
(14:50):
check in new items as they comeand it's a self-updating ledger
system that colleagues areresponsible for.
So they're buying into theprocess.
They know that good processexecution is going to make their
job maybe this afternoon ortomorrow afternoon easier.
So there's an amount of sort ofbuy-in because they know that
they're going to make theirlives a bit easier there.
(15:10):
I think it's just about quitesimple technology that removes
some of that mundanity, makes ita bit quicker, a bit easier and
ultimately gets colleagues abit more engaged in that process
as well then that must give abenefit.
Speaker 1 (15:27):
So they kind of using that through your software.
That must give a benefit.
A central level, ie the headoffice, there must be some stats
or management information thatthey can glean from that yeah, a
huge amount of insights andanalytics off the back.
Speaker 2 (15:41):
You can see what individuals, what stores, what
times of day these processes arehappening.
So you know you typically willhave a markdown period happen
between hours X and hours Y.
You can realize when thoseprocesses are happening out of
hours.
If they happen too late you'vegot less time to sell products.
If they happen too early, youmight be cannibalizing full
(16:04):
price sales that you might haveotherwise got and then, beyond
that, bringing in more ways tosee what colleagues are facing
in store.
Are they dealing with too manymarkdowns?
Do I need to edit myforecasting replenishment plans?
Do I need to figure out betterways to improve that
replenishment process to supportless waste on the shelf so
(16:27):
there's less going into themarkdown process at the end of
it.
So a huge, huge amount.
That's just a lot of scratchingthe surface.
Speaker 1 (16:32):
A huge amount of opportunity to explore from a,
you know, insights and analyticsperspective on that side yeah,
I mean in simple terms, you canhelp reduce the volume that gets
thrown in the bin, so that's afull on cost and make sure that
the markdowns are intelligent,priced correctly, based on the
you know vanilla cheesecakeexample so they actually sell
(16:55):
through.
Then, stepping a stage backfrom that, stop, stop people
checking every item every day.
So we're just targeting theones that we know that need
reducing.
So there's a huge labor free upthere.
But then, yeah, back to yourfinal point.
The analytics then can start tobe interrogated to say, well,
(17:15):
why are we always reducing thevanilla cheesecake?
Do we need to ship it in threesrather than sixes?
Do we need to sell it at allbecause we reduce more than we
ever sell at full price?
There's different bits ofinformation for kind of the ops
team, the buying team, theranging team, et cetera, that
they can start to work with.
Is that right?
Speaker 2 (17:34):
Yeah, absolutely.
It's that root causing, youknow, diagnosing what issues
consistently happen on the shopfloor.
And then how can I make broader, enterprise-wide decisions to
remove those challenges in thefuture?
Do I need to change the range?
Do I need to alter the forecastand demand plan?
Do I need to change thepromotional plan?
Do I need to work withsuppliers to alter the case pack
(17:55):
that comes to store?
Those are just a few of theexamples that could be taken at
head office level to helpimprove things for the colleague
on the shop floor and,ultimately, the customer.
Speaker 1 (18:07):
Good, so exactly let's, let's talk about ai.
It feels like we have to talkabout it on every, every podcast
.
So what kind of are you seeingin the world of, specifically in
retail ai?
So any good examples you canshare, any poor examples.
Speaker 2 (18:24):
Yeah, yeah, lots, lots in both camps, I think AI
and retail operations.
I wrote a piece about this acouple of months ago.
It's really challenging becausethere's so many issues on the
shop floor.
You have to compete with thechaos of the shop floor, the
lack of the shop floor, the lackof data quality.
Getting around those aremassive, massive problems.
(18:49):
I'm yet to see fully-scaled AIthat is autonomous and just
deals with those perfectly andhas no issues.
It doesn't really exist.
I think the best examples, themost successful implementations
of technology that has, aibalances the sophistication of
(19:09):
those technologies and thesimplicity.
So you can have the mostcomplex, advanced machine
learning model in the world, butcan you deliver it to a
colleague in the store in asimple, digestible way that, as
we're just talking about, a theyengage with, b makes them buy
into the process and c directsthem to do a task that they can
(19:35):
achieve, and then maybe d?
Can you see the impact of thetask that they've just done on
the shop floor?
So an example might be ourintelligent inventory alerting
system, inventory insight, whichis basically about flagging
potential instances of phantomor shadow inventory, essentially
where there's a mismatchbetween pi and what you actually
(19:56):
have in the store.
This is a really advancedsystem.
It's got, I think, 20 plusmachine learning models that
balance and make predictionsabout your imagery position
across the entire shop.
But it brings in really quiteelegant and pragmatic analytics
layer and decisions system whichproduces insight in a way that
(20:22):
is quite human and I thinkthat's really important, instead
of just issuing a generic stockcommand around.
Go and check this stock, simon.
It's going to give the context.
This item has not sold for fivedays.
It's got 20 units on hand.
It sold 15 units yesterday.
(20:44):
We think there's an issue reallysimple, digestible colleagues
in store probably understand andappreciate why they've been
alerted to that item to check it.
Because I think that's a bigissue that colleagues will get
alerts from systems in store ortechnologies and eventually so
many bad calls.
You're just going to think thatevery call you get from this
(21:06):
system in the future is going tobe a bad call.
You completely lose the storeteams.
So I think that's a goodexample.
A bad example.
I think a lot of peopleretailers and some vendors will
deploy dynamic markdown systems.
We've got a dynamic markdowncapability at retail insight, so
(21:26):
this is just about capturing asmuch margin as possible whilst
minimizing waste throughexploration-based discounting.
Yeah, a lot of the technologiesand systems, and you can go and
see some of this and experienceit in some stores, but they'll
deploy machine learning modelsthat consider loads of factors,
(21:49):
for example, the volume of fullprice items in store.
How is that going tocannibalize discounted items?
But they really fail toappreciate, I think, the full
value chain in retail ops.
So these systems might assumethat the on-hand system, the
on-hand count for these items,is correct and it will make
(22:09):
discounts based on the fact thatI've got 20 items of that
cheesecake that we talked aboutearlier, but really 12 of those
items are phantom and yoursystem's not correct and it's
making decisions based oninaccurate, flawed data.
And we can all know what theoutcome is going to be
discount's going to be all overthe place and the chances are
(22:31):
you're either not going to sellit and throw away a great
product, or you're going to sellit at a massive discount and
potentially lose a lot of money.
So I think it's how systemsthat are really intelligent rely
on data that is flawed, isinaccurate.
I think that's the big thingthat we started to talk a lot
about in the last sort of 12months is there is a connected
(22:54):
tissue between our suite ofproducts.
How can availabilities lost,sales drive figuring out what
items that have phantom imagery,and then how can phantom
imagery level set on hands instore to drive the right
discounting levels within ourmarkdown system.
(23:16):
And there's sort of this ebband flow between all of them
that feed into each other.
Speaker 1 (23:21):
The key and it's, I suppose, something that rethink.
We've we've learned and there'sthis big unknown, I suppose,
about ai and it's moving soquickly.
The bit that people don't talkabout is and I think you've just
hit the nail on the head is thequality of data that's needed
to drive, drive any algorithm,any system, ai or not?
(23:44):
Yes, and the consequence ofhaving that data not to the
quality it needs.
And again, I think most peoplelistening to this that work in a
retailer will have questionsabout data sources, quality of
data, consistency of data.
And until people get a grip onthat and you know it's
(24:07):
gold-starred and we all knowwhere it comes from and the
sources are reliable andanomalies are detected and moved
out or dealt with we're notgoing to harness it to its full
potential.
Because the software is there.
It's like anything.
It's only as good as what youfeed in, right?
Speaker 2 (24:24):
Yeah.
Speaker 1 (24:26):
Yeah, you're absolutely right.
Okay, so we'll just close onwhere you you're.
You are retail insights interms of using ai in products.
How you're thinking about itmight shape your products and
therefore your offerings tocustomers in the future.
Speaker 2 (24:44):
It's a great question , I think, across a number of
guises, there's probably twothat really stand out.
I think the first is how AI candemocratize access to our
insights and analytics acrossorganizations.
I think, historically, bianalytics has been served and is
customized in a way that worksfor the retailers, but AI
(25:05):
unlocks that and creates almostself-serve for us, and I think
creating giving our customersthe chance to speak to our
system essentially and have asemantic layer in place which
makes that conversation possible, and an AI system that can
produce a really great outputthroughout that conversation is
(25:27):
powerful.
You know, if I'm sat down, I'ma retail operations leader, a
customer, and I've got ouravailability system, and I'm sat
in a store wondering why whyavailability is so poor in this
store, because the sales arefantastic, had a brilliant week,
and you can have a conversationwith our system, understand
what's driving it, and I thinkthat's really exciting, really
(25:49):
exciting.
We hear that as well, andthat's something we're
definitely capitalizing on.
I think the other area is how AIcan to that point I spoke about
earlier around diagnostics.
We're doing some really greatwork around root cause analysis,
so figuring out why out ofstocks keep happening and then
(26:11):
making corrective action at ahead office level, an upstream
level, that essentially cancelsthat out in the future.
Ai is an amazing augmentationof that process to help us move
through that root cause treemuch quicker, much more
elegantly and, I think, get toan answer that retailers are
happy and comfortable with muchfaster, and I think will mean
(26:32):
that what is a very complexsystem in terms of root cause
analysis there's so many reasonswhy this might be happening can
be deployed, managed and usedin a far more efficient way,
easier way in the future, andthat's really exciting.
I think those are probably twoexamples that really stick out,
(26:54):
but there's many, there's many.
Speaker 1 (26:55):
Yeah, no, I like those.
I'm just thinking back to mystore manager days.
How would I feel if my areamanager or somebody from the
center came in and startedinterrogating the stock system
by asking it questions?
There's nowhere to hide, right?
But ultimately, if you're asmart store manager or team
leader colleague, you'll haveinterrogated it yourself first.
(27:17):
So that surfacing data quicklyis great.
I think again, there'll be loadsof organizations centrally
where they've got theirmonitoring, supply performance
and all those other things thatthat go on.
So surfacing that quickly, youknow, looking at suppliers with
poor, poor supply rates, allthose kind of things.
It'll all be happening now.
I don't think anybody deniesthat.
(27:38):
But getting there quicker,flagging it earlier, predicting
it even, will make such a a bigdifference to that whole
end-to-end state.
Yeah, be interesting to see,but I can only see positives
from it.
Yes, definitely We'll pausethere, tom.
Great conversation, really goodto hear about the great things
(27:59):
you're doing at Retail Insights.
If somebody wants to get intouch with you, find out more,
have a chat about maybe, how youcan help, what's the what and
where are the best places forthat to happen I would say reach
out on on linkedin as a classicone, or or drop us a note on
our website and we can have achat.
Speaker 2 (28:17):
We can come to store, grab a coffee, walk around and
have a chat about what you'refacing, what you're challenged
with, and see how I might beable to help.
Speaker 1 (28:26):
Perfect.
Appreciate you taking the timeout, tom, always great to chat
and we'll catch up soon.
Thanks a lot, simon.
All the best.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Are You A Charlotte?
In 1997, actress Kristin Davis’ life was forever changed when she took on the role of Charlotte York in Sex and the City. As we watched Carrie, Samantha, Miranda and Charlotte navigate relationships in NYC, the show helped push once unacceptable conversation topics out of the shadows and altered the narrative around women and sex. We all saw ourselves in them as they searched for fulfillment in life, sex and friendships. Now, Kristin Davis wants to connect with you, the fans, and share untold stories and all the behind the scenes. Together, with Kristin and special guests, what will begin with Sex and the City will evolve into talks about themes that are still so relevant today. "Are you a Charlotte?" is much more than just rewatching this beloved show, it brings the past and the present together as we talk with heart, humor and of course some optimism.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.