How Marketing Technology is Evolving – TfM Congress 2017
This year’s TfM 2017 digital marketing congress and trade show focused on Artificial Intelligence (AI) and Automation, MarTech, Content Marketing, Impact of GDPR and Social Media. This was in contrast to last year’s congress which also saw Neuromarketing and Big Data being prominently featured. From the world of finance both FinTech and Blockchain were notable by their absence.
Martin Tavener and Jeremy Waite from the IBM Watson team kicked off the keynote sessions with an overview on the use of cognitive computing to create a differentiated customer experience.
Although AI is omnipresent it is often incorrectly positioned as a future replacement for humans, it should be viewed as a complementary tool to help humans perform tasks with greater efficiency - both speakers refer to AI as augmented intelligence. Another key point to note is that once programmed, any AI system must be able to learn from mistakes and incorrect decisions.
The IBM Watson system is based on three levels foundational cognition (signal skills like language, speech, vision, empathy), knowledge organisation (reasoning based on data ingestion, enrichment, entity resolution) and higher reasoning (organisational skills centred on conversations and trade-off analytics).
Martin Tavener demonstrated several interesting examples of Watson APIs in use.
Red Bull for example uses insights based on analysing an individuals’ behaviour on social media to map out the personality of athletes they sponsor and check that there is alignment with the brand culture – allowing potential “adjustment” of athletes’ behaviour. (https://www.ibm.com/watson/services/personality-insights/)
Shopdirect uses chatbots to help customers find what they are looking for. Customers are invited to login via social media (increases company reach) and when they for search for a skirt for a summer party for example, along with the diverse options displayed, a bot asks for the date of the party. The system is programmed to check the weather forecast for that day and then suggests a matching jacket since chilly conditions have been forecasted. Although a logical and obvious way for upselling, a considerable amount of planning is needed along with real-time analysis which IBM Watson handles.
Other applications include using sales data and commerce insights for merchandise stocking coupled with intelligent sequencing e.g. general inventory levels and best sellers are all analysed, so that the top products can be automatically displayed on a landing page.
Jeremy Waite used a live demonstration with the Watson system being steered by voice control. At current levels, customer data doubles every 12 months and IBM predicts is that by 2020, 30% of all browsing will be conducted by voice search using Alexa and Siri for example, hence their investment in this field.
In the demonstration, an automated voice control experiment was conducted to show how an e-mail campaign can be fine-tuned – taking less than 12 mins to do what would take an agency 30 days to complete “manually”! For this demo, the objective was to fix e-mail campaigns which are under-performing e.g. open rate, click thru rate, on-site tracking by analysing and correcting subject lines – all with voice control. Following the command: “look at best and worst sentiment scores over last 12 months”, the system analysed 60mil records and served up the required lists which were analysed. The e-mail campaign subject headlines were then changed in real time and the new campaign activated by hitting “run” – all in under 12 minutes.
The take home message from this keynote talk was to put faith in people and not blind trust in technology.
Later during the first day of the congress, a panel discussion on AI was chaired by Dave Chaffey.
The panellists debated whether AI should be acknowledged as being the natural progression following on from the Industrial Age and the Internet Age. There was some discussion centred around whether the “A” should stand for “augmented” rather than “artificial”. The reasoning was that you require a strong human element to establish or define what you want to achieve in a business setting and then set up the appropriate AI system to help with the data crunching. One should view AI as an intelligent system which can help you work faster – a human accelerator not a replacement.
Furthermore, AI alone should not be viewed as the answer to resolving business issues. There is too much temptation to buy a great bells and whistles AI system only find that the issues remain - a recent HBR article addresses this topic (https://hbr.org/2017/07/the-trouble-with-cmos).
Another example given was the development of an edited highlights segment from a full five-set tennis match. This was achieved with the prior analyses of over 50 million historical tennis points to define key points of interest and then appropriate material was selected based on the type of points won and the highlights were collated.
At last year’s congress AI systems from Salesforce and Microsoft were also referred to, but this year it appeared that IBM has made huge progress in this area leaving the competition somewhat behind.
Tasmin Todd gave a very interesting talk on how customers are moulding today’ e-commerce landscape. There is a massive increase in online ordering and parcel shipment. Apart from the logistics of warehouse delivery and stock replenishment, there are several knock-on effects. For example, all new apartment blocks are designed with bigger storage rooms, traffic in London is up 10% due to the increased number of couriers underway, and companies are monitoring the levels of personal orders being delivered to office premises.
Today mobile has firmly established its position as the decision-making device driving desktop conversion with people surfing on their mobiles during their commute, and then placing the order via the desktop once home.
Celebrities are being used less frequently in part due to the excessive costs involved for their engagement in supporting a brand. A new set of micro-influencers on Pinterest and Instagram with growing consumer appeal has now taken the place of the celebrity.
On a more down to earth level Sly Yuan presented a case study showing how competitions and social platforms allows customers to do the “heavy lifting” for a campaign. An Australian beer producer had defined an objective to generate more beer sales by targeting females and creating a lifestyle setting for responsible drinking. Using a tie-up with a BBQ retailer and a combination of discount vouchers given with every 24-pack of beer, a competition was set up using Facebook and an e-mail marketing campaign focussing on owning the summer with all the various lifestyle components such as BBQ, relaxing with friends etc. At the end of the campaign beer sales were increased by 5% yr-on-yr with an annual category lift of 3.5%. Social media channels were used for word of mouth promotion and $9mil revenue was generated over a 2-month period; no ATL communication was used. A real-world example with figures demonstrating a clear ROI. (https://tractiondigital.com/australias-largest-promo-lifts-sales/)
Ed Relf gave a presentation showing that any industry can be ripe for disruption. He shared some key learnings from the disruption of the laundry industry. Laundrapp is a washing / dry cleaning home-collection service designed to make life easy for the consumer. Any industry is ripe for disruption once you have clear consumer insights showing an unmet need coupled with service / product consumption patterns. Factor in preferred communication channels and content, with consumption time during the day, and one can persuade a change in behaviour eventually making it habitual.
With Laundrapp, so-called “vanity metrics” such as likes, shares and App downloads are of secondary importance to purchase of the laundry services – the “only metric which counts”. Very detailed analyses of customer life-time value (LTV) broken down into segment analysing communication consumption habits generated some key insights. The original assumption was that busiest time for orders would be on a Sunday – in effect Monday morning between 7-8am proved to be the busiest time for orders via cell phone with people on their commute to work.
Ease of use is key with App download and off you go – no irritating login via Facebook or Google! Apart from the App (which is a one-button swipe action to order collection by a Laundrapp driver), direct mailing is used extensively along with social media, as well as SMS push messaging with autofill for redemption of promo codes. The whole concept has been designed to make the customer experience as smooth and as “hands free” as possible.
Furthermore, all Laundrapp drivers undergo extensive customer training since they are the main human interface, and are full-time employees not zero-contract hires.
Andy Mulcahy chaired a very lively panel discussion on the impact of the Black Friday and Christmas peak on retailers. Whether we like it or not Black Friday is here to stay and since 2014, Black Friday went from a weekend event to a week-long event.
The actual day on when Christmas falls also affects purchase patterns – so it is crucial to look at historical data in first line and try to identify trends. Many people can remember in 2014, how warehouse replenishment and delivery all collapsed in the run up to Christmas. Nowadays sales patterns more predictable due in part to better use and understanding of analytics.
Currently the spikes in the 7-week period going from Black Friday to Christmas are getting shallower and the consumer spend period is getting longer as people spend in the days leading up to Black Friday and not only on the Black Friday weekend itself. This has reduced the rush in the period leading immediately up to Christmas.
In 2016, December 19 was the peak online sales day for the UK with peak offline trading occurring on December 22 / 23 as people bought from a physical store, or selected online order with store pick up, to avoid the non-delivery risk with a pure online order.
During the discussion it was recognised that while Amazon do a very good job of priming the market, many smaller retailers would rather that Black Friday – a US import - did not occur since in the UK, Christmas is a “bigger” event than in the US. As such retailers have high CPA costs for Black Friday due to the need to go for volume with PPC being the prime cost driver. There was agreement that Black Friday is now an established part of the Christmas run up with the associated and expected discounts. A lot of debate centred around whether Black Friday is a discounting event or a marketing event.
Interestingly, mobiles now feature more heavily as we have better payment ecosystems available to complete an order. Peak orders (often from mobile devices) are received during the week between 20:00 and 22:30, with instore pick up expected the following day either over lunchtime or on the way home. Other times for online orders are seen on the morning commute or during the lunch break. To gauge just how big Black Friday has become, one should look at YouTube which has a colossal amount of material explaining how one should prepare for Black Friday – not just retailers but also consumers.
Regrading trends and predictions there was agreement that new occasions will continue to be imported (e.g. Chinese New Year), people will respond more to discounts, more personalisation e.g. coupons will be seen, and finally volume / top line sales will continue to grow. However, panellists also remarked that viewing this timeframe for generating top annual revenues may no longer hold for the future, since consumers are now showing a desire to shop at any time of the year and expect discounts and coupons throughout the year.
These are just a selection of some of my personal highlights from a whole collection of interesting and stimulating talks. This congress / trade show is an excellent way of keeping up with all the developments in the field of digital and will take place again next September.
#brandcrusader