Browse Tag: data

What are retailers thinking about in 2018?

I attended the eTail Nordic 2018 event and wanted to summarise some findings from selected keynotes on 2.10. – 3.10.2018.

Peter Mold, CDO, ICA

Peter Mold, the Chief Digital Officer of Swedish grocery and pharmacy giant ICA was first up. He stated ecommerce for groceries still hasn’t picked up in Sweden, concluding only 2% of the purchasing volume was coming in through digital channels in 2016.

One of the issues with ecommerce for groceries are the expensive and cumbersome logistics. As a grocer, you need to store, pick and transport fresh produce taking care of the cold chain all the way to the consumer, making it expensive.

Mold continued by describing the importance of data telling us that data is what feeds AI, which will become even more important.

Mold described the future of physical retail being all about the customer experience. He said retailers need to provide good experiences and even become a sort of hub connecting like-minded people or people with similar interests.

One of his views did not resonate that well with the audience. It was the future of voice vs keyboard controlled devices. Mold believed that we’ll see the keyboard vanish within the next five years. Asking the audience, there was quite some scepticism around this happening.

I’d like to support Mold’s view here. Looking at where efforts by Apple, Google, Amazon and other tech giants are, it’s AI enabled voice control. The technology is slowly but surely growing out of being a curiosity into being a usable interaction method.

Being a Finn, I expect to be among the last to use natural voice control, but in English the technology is maturing fast.

Mikael Lenneryd, Director of Digital and Loyalty, Apoteket

Mikael Lenneryd from Apoteket described a case from his former employer, Samsung. He told the story about how they connected their own first party data with the second party data from a telco to run better digital campaigns.

They used the combined data to target real people based on their interests. This was a challenge both technically and legally.

The technical challenge was combining the data and feeding it into the normal programmatic buying architecture. The legal issues were around sharing data between two large companies and then feeding it onto the other actors in the value chain. It took seven months to sort out the legalities to be able to execute.

But the results were good. Average click to conversion had a 35x uplift and the best segments had a 50x uplift. Which is, in one word, impressive.

The campaign was run on a regular promotion and not an extra sweet deal. So the results should be comparable. The product was a Samsung S8 with a 24-month contract.

Mei Chen, International Business, Alibaba

Next up is Mei Chen from Alibaba. Personally, I found this keynote very interesting given I have almost no knowledge about the Chinese market, but have been interested in it for a while.

The first shock is grasping the sheer volume of the market. The Chinese version of Black Friday is Singles Day. During the sale, Alibaba hit more than a billion dollars per hour for a 24-hour period. This means they totalled over 25 billion dollars during the day and night.

One key difference comparing China with the US or Europe is the age of the consumer. China has a huge number of young consumers and they want to stand out. This can be done by choosing what to wear, what to put on their faces (particularly women) or what to eat.

Alibaba is carrying a lot of European and Nordic brands as well. Given the wish to stand out, products coming from small and (from a Chinese point of view) obscure markets might have a lucrative opportunity ahead. And that’s how Alibaba wants to be seen: as an opportunity to move product, not a threat to local retailers focusing on Europe or a particular country within the region.

In China, marketplaces have a much stronger position than in the US or in Europe. Chen’s take was that the Chinese consumers don’t see a point in using a lot of different sites when you can find almost everything bundled together in just one place as well.

This is a trend we can see in the US as well with Amazon seeing that they have surpassed Google as the most used search engine for product search. So maybe this is the future of retail with platform players acting as supersized department stores and landlords deluxe for whomever wants to get their product out there en mass?

Caroline Carlqvist, Program Manager Personalization, Zalando

Caroline Carlqvist, working on personalisation for Zalando, talked about the challenges of fashion in particular. She described a small boutique in Italy where the owner got to know her so well her boyfriend could shop there for her and the shopkeep would make sure the clothes were a fit for her, in both style and size. That’s what Zalando wants to be as well, but at scale.

The particular challenge for fashion is size. It’s difficult to know if a medium is the same medium as you’ve gotten used to with another brand.

Sizing connected back to data. Zalando has a lot of data about returns and the reason for said returns. This data can help find the right fit. If you have a lot of people who bought brand A in medium sending back brand B in medium since it’s too small, Zalando can warn the next buyers of brand B to size up if they see medium from brand A as a good fit.

Zalando is also pushing this data back to the manufacturers so they know if their brand is consistently getting a bad fit from multiple people.

The threat of Amazon and the Intersport approach

There was a debate hosted on Amazon. I’m not going to dive into more detail about the debate, but I’d like to highlight something interesting presented by InterSport. They have started a cooperation with Tmall (a part of Alibaba) in China.

The partnership is an experiment into combining the upsides of physical retail and ecommerce into one package. Instead of bringing your shopping with you, you can stroll bag-free to lunch and get your new sneakers delivered to your house two hours later.

Louisa Nicholls, John Lewis

Louisa Nicholls from John Lewis shared some insights from the UK market. This is interesting for Nordic and Swedish retailers given that Amazon is already active in the UK market.

Nicholls pointed out that consumers are moving from buying more stuff for fulfilment into searching for experiences.

Given the trend, John Lewis have started adding services to their offering to lure people into the stores. E.g. their style studio service gets an uplift of 30% measured in revenues compared to fashion shoppers who don’t use the service.

Nicholls added to the stories by others before her on the value of data. She told us retail used to be about three things: location, location and location. Now that has changed into data, data and data.

Summary and reflections

Amazon is, of course, on everyone’s mind given the rumours about them entering the Nordic markets. Sweden would be first up, given it’s the biggest population in the region. The severity of the threat seems to divide people.

As a reflection on Amazon’s possible level of threat, I’d like to reminisce on video rentals in Finland. In an article published in 2013, the two major video rental companies thought companies such as Netflix and HBO are a curiosity, not posing any real threat for them. In hindsight, it’s painfully obvious how wrong they were. Disruptors have a way of looking harmless early on.

Data, both the importance of data and how to use it properly, was a part of most keynotes and case studies. Well, it was the main topic of my case study as well, so I might be a bit biased. Yet I’d conclude the significance of data is well understood in retail, but many companies still lack in tools and tactics to actually make the data work for them.

The future of the physical store was also a big debate. This is natural, given most incumbents come from the physical space and have expanded into ecommerce. The issues is, no one really knows what role brick-and-mortar will play in the years to come.

 

AI, data ja automaatio

”Robotit tulevat ja vievät työpaikkamme.”

Suurin piirtein näin lehdistö on otsikoinut meneillään olevan murroksen mm. täällä, täällä, täällä ja täällä. Niinpä innostuin tiivistämään omia ajatuksiani automaatiosta, joka on koko trendin pohja. Varoitus: aion seuraavaksi käyttää monia inhottavia insinööritermejä.

Jotta jotakin voisi automatisoida, täytyy ymmärtää prosesseja. Prosessit koostuvat kahdesta osasta: päätöksistä ja toimenpiteistä. Ensin tehdään päätös tarvittavasta toimenpiteestä ja sitten suoritetaan itse toimenpide. Toimenpidettä seuraa päätös parhaan jatkotoimenpiteen suorittamisesta ja sama toistuu.

Insinöörit ovat aina olleet hyviä automatisoimaan toimenpiteitä. Robotti hitsaa kaksi palkkia yhteen tehtaassa. Ohjelmisto koostaa raportin. Tiskikone puhdistaa astiat.

Päätösten automatisointi on paljon vaikeampaa, sillä päätökset ovat monimutkaisia. Tarkastellaan tätä käyttäen autoa esimerkkinä.

Auton suorittamien toimenpiteiden automatisointi on lasten leikkiä. Hieman yksinkertaistaen autossa voi painaa kaasua, painaa jarrua ja kääntää rattia. Kolme helppoa toimenpidettä. Ensimmäisen vuoden automaatio-opiskelija osaa säätää kolmea ulostuloa samanaikaista, varsinkin, kun kaksi näistä on toisensa poissulkevia.

Silti itse ajavan auton tekeminen on tavattoman vaikeaa, koska päätösten automatisointi on tuskaisaa. Koska painan kaasua ja miten paljon? Mihin suuntaan käännän rattia ja kuinka paljon?

Auton ohjaamiseen liittyviin päätöksiin tulee dataa monesta lähteestä. Mitä ympärillä tapahtuu? Millaisia rajoituksia on voimassa? Minne auton halutaan päätyvän?

Syy itsestään ajavan auton isoihin kehitysaskeliin on, että käytössä olevat työkalut ovat parempia keräämään ja analysoimaan saamaansa dataa. Ohjaavien algoritmien ei tarvitse olla valmiiksi ohjelmoituja kaikkiin mahdollisiin tilanteisiin, vaan kone itsessään oppii. Edistysaskeleet ovat nimenomaan päätösten, eivät toimenpiteiden automatisoinnissa.

Päätökset syntyvät datan avulla. Siksi suosittelen kaikkia automaatiosta kiinnostuneita aloittamaan datasta. Olennaista on ymmärtää mitä dataa yrityksestä löytyy jo nyt, mitä muuta dataa voidaan kerätä ja miten laadukasta kerätty data on. Data määrittää millaisia päätöksiä sen avulla voi ja kannattaa automatisoida.

Yhteenveto kaavan muodossa: automaatio = data + algoritmi + toimenpide. 

 

Ihminen ei pärjää datalle, tässä kolme todistetta

20140531-110327-39807069.jpgIhminen on usein kelvoton päättäjä. Runsaslukuiset vinoumat ohjaavat logiikkamme sivuraiteille tai päistikkaa metsään. Siksi objektiivista dataa ahmivat koneet pesevät meidät tarkkoja päätöksiä kaipaavissa töissä.

1. Robotti ajaa ammattikuljettajaa paremmin

Itseajavista autoistaan tunnettu hakukoneyhtiö Google paljasti konferenssissa, että sen robottiautot ajavat paremmin kuin ammattikuljettajat. Robotit viettävät vähemmän aikaa läheltä piti -tilanteissa ja jarruttavat pehmeämmin kuin koulutetut ihmiskuskit.

2. Algoritmi syrjäytti ihmisen riskisijoittajan johtoryhmässä

Riskisijoittaja Deep Knowledge Ventures nosti Vital-nimisen algoritmin johtoryhmäänsä tekemään sijoituspäätöksiä. Vaikka tempauksen motivaattori lieneekin julkisuus, ajatuksessa on perää. Miksi muuten valtaosa kaikesta pörssikaupasta käytäisiin koneiden toimesta ihmisten sijaan?

3. Lääkärit eivät pärjää koneille

Riskisijoittaja Vinod Khosla linjasi, että paras tapa kehittää terveydenhuoltoa on hankkiutua eroon valtaosasta lääkäreitä. En epäile hetkeäkään, etteivätkö koneet parantaisi terveydenhuollon laatua, kun jo yksinkertainen tarkistuslista vähensi leikkausten aiheuttamia komplikaatioita runsaasti. Lisäksi lääkäreiden diagnooseja on parannettu jo yli vuosikymmen ohjelmistojen avulla.

Gigatolkulla tarkkaa ja objektiivista dataa ahmivat koneet tulevat väistämättä pärjäämään ihmisiä paremmin monissa päätöksentekointensiivisissä ammateissa.

Kuva: www.freeimages.com, Sasan.