Thomas Bertelsen is the Co-founder and Managing Partner of AI Media planning platform Blackwood Seven and James Procter is the Founder and Creative Director of data led agency One Small Step Collective and they discuss the role of data in planning and predicting media performance and providing a measurement of performance and an accountability to delivering results, speculating that the way to connecting media and creative is data.
You can listen to the podcast here:
Follow Managing Marketing on Soundcloud, TuneIn, Stitcher, Spotify and Apple Podcast.
Transcription:
Darren:
Welcome to Managing Marketing and today I have the opportunity to sit down with Thomas Bertelsen , co-founder and managing partner of Blackwood Seven; a company that brings artificial intelligence to media analytics and planning, welcome Thomas.
Thomas:
Thank you.
Darren:
Thank you for coming to Sydney, you’ve come all the way from Copenhagen, Denmark.
And also here is James Procter, founder and creative director of One Small Step Collective, which is a data led creative agency, welcome James.
James:
Great, I haven’t come that far.
Darren:
No, you’ve come from Melbourne
James:
Yes.
Darren:
In the whole of the world what brings you two together in Australia?
James:
As an agency that believes in strong data that underpins our whole offering we are always looking at technology and where the future’s going. I think that we always recommend to our clients to have change in the middle of whatever they do and we try and do that ourselves.
So we are always looking out through a couple of mediums, not just about ourselves. We learnt about Blackwood Seven and about AI driven media mix modelling and it was really interesting for us in terms of having data and proper attribution at the end of what is essentially still quite a big black hole for marketers in terms of half my advertising money is wasted but I don’t know which half kind of thing.
Darren:
I hate that quote.
James:
It pulls down what we do.
Darren:
I also think there is a fear inside the industry that we are happy for people to keep saying it because there’s a fear it’s actually optimistic. That it’s actually 70% or 80%, so if people go around saying 50 %, I’m comfortable with that, but if it’s 80% what are we doing.
James:
Exactly right so I was meeting with the guys from Blackwood and we’ve been in chats with them for the last two years in terms of their platform, where it’s going and what’s happening with it, and bringing it down here to Australia.
We have just formalised the deal to represent them in Australia and New Zealand, so we are very happy with that.
Darren:
So, Thomas, I first heard about or read about Blackwood early 2016, so that’s a little bit later, when did the company start?
Thomas:
Mid 2013 we started having basically just an idea, having no tech, no platform.
Darren:
An idea.
Thomas:
A good idea and some investors believing in that and believing in us and putting money into the company.
Darren:
I remember when I read about Blackwood Seven, in fact we wrote an article that said, ‘is this the end of media planning as we know it?’ We got quite a lot of response from people especially from traditional agencies.
There is this cynicism about artificial intelligence or machine learning or pattern recognition; how could that possibly replace the intelligence and experience of people working in media agencies? But you are actually not about replacing that are you?
Thomas:
No, I think that is not only within media, I think a lot of industry says artificial intelligence machinery can never replace us. I think that is human nature. We are not about to replace agencies, we might be about to change how media planning is done but hopefully for the better.
We think that the output of what we do, the models do, the platforms do is something that basically should replace a lot of very hard excel work at media agencies trying to get the location right , trying to navigate in the dark, not having the connection between sales and media in the platform.
So, basically replacing that so media agencies can focus on creativity, on strategic consulting of the clients, instead of doing a lot of number crunching in excel sheets.
Darren:
You would have to say in the last decade the number of media channel, media options available has just increased exponentially. To even sit there and believe that one person or a team of people can keep across all of that opportunity and all of that performance data is fairly naive isn’t it?
Thomas:
Yes totally, and the structure and dynamics between media. There’s an explosive number of media out there but also that some media also affects the impact of other media, How to detect that if you don’t even have the sales numbers if you are basically unable to connect media investment with sales.
It’s so complex, so dynamic the media environment, you are not able as a human being, even though you are the best planner in the world, to basically quantify the effect of each media.
James:
And it is built for AI.
Thomas:
Yes exactly.
James:
It’s like a perfect storm for AI.
Thomas:
Yes it is and data’s there because it is a data driven business. Media is very much about investing money and getting return in terms of impressions or clicks or whatever; it’s a data driven business so it’s perfect for AI.
James:
I would also say yes, the platforms use to augment current media agencies but also we find a lot of the times its reinforcing gut feel or assumptions they already have and quantifying them.
It’s not always saying, well we were doing it completely wrong, it’s saying that we were doing it incredibly right to 80- 90 % and here’s how we can optimise based on that assumption.
Thomas:
I totally agree because very often we see that some of the not very wise media decisions that have been made by a company, an advertiser, that comes from the advertiser them self.
They insist on having this type of media, I won’t mention anything, but because one of their board members lives in a certain area and they just want to be present in that area one way or another.
And the media industry has been fighting that for so long but based on the fact that it is a board member then they can do nothing. Now they have proven data, it has no impact on sales. Let’s skip in and do something else.
Darren:
Or at least if you are going to do it, understand that it is a cost to the investment so you will just write that off.
I’m glad you bought that up, Thomas, because one of the things we often hear from marketers is, ‘oh my media agency gives me the same schedule over and over again, it’s almost like it’s just come off the photocopier’.
Yet they are wanting something different almost as if something different is going to be something better. If it is data driven its actually better anyway because it’s proven by the collective data that’s gone into there on past behaviour.
Thomas:
I have been on the media agency side for many many years and I have not experienced one client that doesn’t kind of say, ‘we want your stuff, come on suggest something new’. And whenever the media agencies do that, ‘no, no we cannot risk that; let’s stick to the old plan.’ So it’s also that you want something but you really don’t act according to that.
But coming back to the media agency’s role, I totally agree that having the models and the platform that can actually show which media plan to go forward with, it actually allows you to experiment.
Trying out new media channels, new media combinations, new media publishers and the week after, the month after you actually know, was it worth it. Did it actually drive extra sales or didn’t it? It allows you to experiment without the cost.
Darren:
So, James, from a creative perspective what do you see as a real benefit here for you as a creative agency? I mean this platform is very media, sales, brand focused so where do you see the benefit for you and the agency?
James:
It’s a good question and I think you and I Darren ask the same question when taken through the platform itself and it is something creative’s have been trying to do for ages in terms of measuring the effect of creative. So what effect does creativity have on this and that was something we discussed at length with Thomas.
I guess the big thing is that the model can take in any piece of data that has an effect on the KPI. So getting down to a publisher level on media, great so understand the media but also at that point we can also say this is this campaign, this is this campaign, this is this campaign, and do what people know as AB testing on different media as well.
I see that as really exciting and getting feedback on that from a creative point a view. I think creative people are better with the more data we have. We are naturally curious more about what makes things tick and we are better with a tight brief and better with the proper amount of data behind it.
We use Crimson Hexagon, we use Brandwatch, so these are social analytics tools to really understand the consumer, to understand language that really helps insights and strategy get better.
Darren:
I’m going to stop you there because that makes you quite unusual because a lot of creative people have reacted very badly to the idea of data informed creative process. That they see this as trying to create a formula for creativity and that formulas don’t work.
James:
I don’t view data as doing the job of creative. The job of creative is still the X factor and the magic that goes on it. It will certainly measure and attribute that X factor but at the same time I see the bigger job of data as being able to firm up the questions that we are asking.
Who is the target market, what do they think, what do they do, what interests them, what doesn’t so that we have a greater understanding of them at the start so that when we do creativity into the top of it we are in a much better chance of getting it right more of the time than less.
I think that it’s not going to be something where it’s black and white, it will just mean that we’ll hit the target a lot more because of the fact that our insights are a lot better.
Darren:
Thomas, it’s not just creativity that you capture in the model. You mentioned before how sales and media investment are sort of the core part of the platform but you’re capturing all the other things that affect sales aren’t you; from weather to what other activities are happening?
James:
Economic outlook, unemployment.
Darren:
It’s quite comprehensive.
Thomas:
Yes.
Darren:
It must have taken quite a while to model all of the variables to build the platform in the first place.
Thomas:
All of the variables are depending on the customer because some customers have sales that are affected by data points out there that other customers do not have. So the model is basically customer specific.
What we have built and used quite a lot of money on building is basically you are taking all the data that is specific to customers and then getting an intelligent model out of that.
The modelling framework basically, the engine we have built so when we have the data we can build the model extremely fast and we can update them really fast as soon as we have new data in an existing model we can update it.
The benchmark for the market on average is you update the model in 6 – 8 weeks; we can do it in less than 24 hours.
Darren:
Wow.
Thomas:
But the data put into the model has to be adapted to the type of client and their marketing environment.
Darren:
I remember reading the two main influences on the sale of Coca Cola is the weather and distribution. It’s interesting because often we think about working in the marketing media advertising area, that that activity actually mostly contributes most of the sales, when in actual fact what we are talking about is marginal influence. Incremental increases on top of a base.
Thomas:
That is why you need all the other data points because otherwise you could look at all the sales curves and okay, we did a campaign and sales went up, hey we performed well, it was a good campaign but what you don’t see is actually you increased your distribution at that point in time because the sales guys knew that a campaign was coming up and you also reduced the price to stimulate sales.
So what is that; is it distribution or the price or was it the media campaign that actually worked.
Darren:
So all that goes into the platform.
Thomas:
Exactly.
Darren:
So future predictions will actually take that into consideration, or at least prompt you to answer the question as part of the planning process.
Thomas:
Exactly.
Darren:
Interesting. How long has the relationship been going, its relativity new isn’t it?
James:
It’s relatively formally new and yet at the same time we’ve been working for a year and a half with Thomas and the guys at Blackwood Seven on a big bank insurer in Australia that we’ve done the models for. So yes we’ve been working for a long time with the guys.
Darren:
And has it changed the way that you go about the creative process or the briefing and creative or even the way you present creative ideas to the client, without naming the client. Has it changed your process?
James:
Yes, it has. I think the bigger outcomes for Blackwood Seven are certainly in the media space and in the business space. I think the big business insights that are gained through the platform are certainly ones that translate back up to the questions that are asked at the marketing department and the questions that are solved by the creative process.
So in a sense even though the platform is at the end of that process there, the insights that are given work back to the front. We see a lot of marketers that are pulled back to the board room table because they’ve got a business case.
Darren:
Supported by data.
James:
And that’s the big thing.
Darren:
Not gut feel.
James:
Exactly right. You’ll see the CFO’s will set up in the thing because it’s not just this black hole of I think, and sales will be incrementally better than last year; it’s we do this thing and we do this thing and this is what’s going to happen.
Darren:
It must be interesting from the point a view of being able to project a model that’s investment, media channel choices and sales for instance and then see depending on the creative that goes in there, how close the delivery of the model is or whether it exceeds it.
I mean as a creative getting that sort of feedback, is it better than winning a gold something or you like that as well?
James:
I don’t think anyone is going to give up a gold something. This next generation of creatives really know that effectiveness is what it’s all about. If you are showing that you impact a client’s business, you show the effectiveness of marketing, there’s a much better chance of keeping the client and doing better work.
The trust that goes with that is the first thing you need to do and then to be able to say once they trust you, I think we can take a leap with this because I understand the brand, I’ve proven I understand the brand and you know that I do and I think we can take a leap here to kind of do this. You certainly get this from modern creatives.
Darren:
So you have to say you are willing to be more accountable to delivering results.
James:
You have to be, where’s it going to go? With the data getting more and more accessible in the future and we’re having more and more algorithms churning out more and more stuff we know more about why things happen and the way in which they happen.
So I think accountability is not an option, it’s something that needs to be a fundamental in what we do.
Darren:
Everyone says they are accountable to results but then they hope like hell that no one has got the results.
James:
Exactly right, or that the results are so woolly, or that someone goes my data versus your data. You know I think that’s the big thing about Blackwood Seven as well is the unification of messages inside the organisation.
What are we all pulling for? So is the sales teams going this is my data and the marketing team saying this is my data, so there is only one data set there and everyone inputs the data to understand that KPI.
I think that when a whole of a business understands that, that’s when you see everyone pulling and efficiencies are amazing in that space.
Darren:
So, Thomas, you and your partners would have come from media advertising backgrounds.
Thomas:
Yes.
Darren:
You would have been I’d imagine working both in the period when media and creative worked together and when media and creative were separated.
Thomas:
Yes, that’s correct.
Darren:
So there’s a famous saying that a guy called Harold Mitchell who’s very famous in the Australian market said, ‘once they are separated you can’t put the toothpaste back into the tube’, that it will be impossible for them to come back together.
We’ve even seen a big trend with a lot of creative agencies putting up their hand and saying we offer media as well, ‘we are full servicing’. The interesting thing for me is that it’s actually the application of data that is making it possible, not to put the toothpaste back into the tube, but to invent a whole new tube, isn’t it?
Thomas:
Yes.
Darren:
Because we are not just talking about bolting media on the back of a creative idea; it’s actually the start, middle and end of the campaign development isn’t it?
Thomas:
I totally agree and that’s one of the reasons why I started in the media agency industry many years ago. Basically the media agencies were separated out from the creative agencies and over the years the media agencies have gained more and more power, some people may think too much power but that was based on being data driven.
Maybe not data driven in the sense that could explain how much media actually affected sales but actually being accountable. I think that was one of the key things that a creative business, advertising agencies missed out on; that was their accountability.
Hopefully tools like ours can actually start combining accountability both from the media side and the creative side so you unify what is extremely important for unifying media and creativity. You cannot separate it.
Darren:
Did you and your partners think that would be an outcome or were you just trying to come up with a better way of modelling and planning and making media more accountable?
Did you think about the fact that this could be potentially the thing that brings media and creative back into the same playing field?
Thomas:
Maybe I should say yes now but no. What we wanted to do initially was to bring the clarity to the advertiser. Actually, the first sentence we put on our web page 6 years ago was that we bring the control back to the advertiser.
Because the marketing department has been the advertiser and has basically for so long been unaware and poured the money into media, the main part of the marketing budget but what is the outcome?
So giving a tool to advertisers to basically underline the importance of marketing; marketing being done well or not well. So that has basically been the aim to bring insights and knowledge to the advertiser’s side in order to basically have the right dialog with the creative side and the media agency side.
Darren:
I worked in advertising as a creative when media was presented in the last 10 minutes. You spent 50 minutes talking about the idea because that was the sexy thing. It was very visual and then the poor media person was left 10 minutes at the end, put up some Excel sheets, that no one could read, on the projector and that was it, which is really interesting.
I never understood why, when 10 times more money was usually spent on media investment than creative, why that imbalance was there; 50 minutes on creative, 10 minutes on media.
And also why media agencies weren’t infinitely more trusted because trust has become a big issue in media with the rebates and all that type of thing when you would think that if you’re giving someone so much of your marketing budget you would have to have a level of trust anyway.
Now, what I see with Blackwood Seven is the opportunity to re-instill trust because it becomes both parties, creative and media, are working to hold themselves accountable with the client to deliver results.
Thomas:
I very much agree. Your opening comments about some agencies saying ‘we’ve seen this before and it won’t replace us’—this is a very good point in time where there is distrust towards the entire media agency business from the client side.
Having this kind of tool can prove them wrong but there is a reason why we allocate as we do. It’s not because we earn more money over here as is regarded as a natural thing from the advertisers side. It’s basically because it stimulates your business, your sales the most doing it this way.
It’s a very good starting point for discussion and basically building trust between parties because it gives the clarity on what works and what doesn’t.
Darren:
I remember a media procurement person for a global client coming to Australia and wanting to run a tender. We asked what the spend profile was and about 40% of their budget was in outdoor. And the media procurement person, having a media background, said, ‘that seems very high’.
And we’re sitting there smiling because we knew that outdoor was giving some of the best rebates to the agency that you could have so why wouldn’t the agency move more of the money into outdoor.
With a system like this where you’re accountable to results, you can’t make decisions based on where you’re going to get the biggest rebate can you? Unless you start manipulating the data.
James:
I think the pushback is natural and change is hard in a lot of these things but I don’t think it’s just the media industry that’s going to see data wash over it.
Darren:
Of course not.
James:
I think it will accelerate in this industry because the data is available and the products like Blackwood Seven are available. That transparency that Blackwood Seven gives.
Darren:
I would disagree. I don’t think it gives transparency; I think it gives accountability. The whole argument of trying to be transparent is actually a fool’s errand because just simply being transparent doesn’t make it better but having a way of measuring performance.
You can have great performance and not be transparent. Who cares if you’re getting great performance? Transparency is a default that we’ve gone to because a lot of marketers, media agencies, and creative agencies have all said it’s too hard to measure performance so let’s be transparent as a way of rebuilding trust.
That awful saying, ‘opening up the kimono’, why do I need you to do that? I just need you to deliver the results I need so let’s forget about transparency and talk about performance and holding each other accountable.
And one of the things I like, Thomas, about the platform is that it does both sales and brand building as well. So you can do short, medium, and long-term objectives.
Thomas:
That is one of the really interesting things about the model; that you can introduce, not just short-term but long-term effects like, for instance, a brand parameter. Within recent years we have seen a lot of focus, not only from the media side but also the creative side of stimulating short-term sales.
You have a lot of digital tools showing how on a minute by minute basis how creative and media works. But building the brand so that next year we have higher sales or are less dependent on having a lower price but can increase our prices. That can be captured by the models.
Darren:
I’d imagine that as the data goes in it starts to allow for that earlier brand building investment, to what the contribution would be to sales now.
Thomas:
Yes.
James:
Yes. Whether it’s captured by itself as an item but then there is also a part of the model which is baseline sales where you look at that baseline of sales and go ‘year by year, are you wanting that to grow?’ That’s our free hit without the marketing.
Darren:
Or at least the non-advertising.
James:
That brand saliency. So when I’m in the supermarket at that zero moment, what’s going to make me go over here or over there. It’s a massive thing. And these are big discussions that have happened for so long in our industry about what’s the worth of brand? What’s the worth of all these things?
Making sure you can start to quantify them in a way in which you just weren’t able to. And that’s really due to processing power and algorithms.
Darren:
So, James, you said that you’d been working together for about 18 months with a client. I imagine any model starts off with you dump as much information in but over time it would get better and better, as in more refined in the way that it can predict outcomes.
Is that the experience in just 18 months or is it a longer period of time that it takes?
Thomas:
No, that is the experience, both from a model experience but also from the perspective that the client gets this wonder that has been built based on their data and perhaps they have the power and energy to dig out more data, other data variables and introduce them into the model.
So, it’s not that we build the model and it’s running for 2, 3 , 4 or 5 years, you can actually get new data in, refine the data to get an even better model. But the more data we have over a longer period the better the model gets from a technical perspective as well.
Darren:
One of my favourite quotes is, ‘all models are flawed; some are useful’. You just accept that any model is going to have some flaws in it but it’s better to have a useful model than no model at all.
James:
Totally, and in Blackwood’s case, the model doesn’t get released until it’s 90 plus percent. So, you’re not going to say, let’s build the perfect machine.
Darren:
Statistically, the confidence interval is 90 plus percent?
Thomas:
Exactly right. But in optimising a model, a lot of the time it’s reduction, taking out things we thought might have an effect but it turns out they didn’t really have an effect on the model.
I think that’s interesting from a client’s point of view in terms of championing it up through the hallways saying ‘we’ve got this data set here’ but it didn’t really have much of an effect at all on the model.
So, the ability to answer questions and also questions they didn’t have, and Thomas will talk a bit about it, once the model is done, it starts with media but it ends in amazing points where there are deep dives into what affects pricing. Where’s the sweet spot for pricing for me? At what time of the season should I be here? If my competitors are doing this from a pricing point of view what should be my reaction to optimise our sales?
Those kinds of questions are big questions for companies rather than should I buy a 30 second or a 15 second spot? And which one of my suppliers is moving more product for me? In what way is that directly related to my media or creative? They have other questions that come up once the model’s been built and they’re really interesting.
Darren:
And that would be, Thomas, because of the focus on sales as a way of measuring performance right?
Thomas:
Yes. But not just sales. Any KPI that the client thinks this is the reason why we are doing marketing. It can be sales, it can be transactions, traffic to a physical store or online store; basically any data point you can quantify.
Darren:
I said sales because I’m sure you’re both aware there is this big trend that’s come out of the WFA in Europe and the ANA in America about marketing as a growth driver and the change from being about just the reputation or brand management to actually driving business, so sales suddenly becomes a really important metric.
The danger has been in the past that the only measures a lot of marketers have had has been last-click attribution and so we’ve seen this big shift in investment away from media into online where they can measure that last click. And they’re giving 100% attribution to that last click.
James:
Which is amazing. And that is one of the big points in terms of being able to give credence and attribution to offline media and the effect they have. I would say to the offline media and the TV stations in particular, that these models quantify exactly the effect that these have on sales but also on other medias such as digital media, in terms of making them more effective. It’s a big day for offline media as well.
Thomas:
The amount of money wasted on last-click attributions insight is enormous. It’s like claiming it was the last beer Friday night that made me drunk.
James:
It was actually.
Darren:
No, it was between the first and the last that it happened.
Thomas:
We have seen a revival for some media that is basically not modern or smart for the marketing departments to discuss, like newspaper or radio advertising, good old-fashioned media that has basically been forgotten by the markets. And the media want to do the digital stuff and suddenly we see, not because they are good media but because price has decreased so much that it is relevant, it is very fruitful and they are back on the plan.
James:
That’s interesting in that the initial clients in Blackwood Seven, the ones with data in hand, digitally-based clients; what was their media of choice? Digital. But the biggest answers that came out of using Blackwood was their spends into offline media so they should be pushing hard into offline media.
Darren:
And haven’t we seen that? Everyone talks about the disruption caused by Amazon, Airbnb, Uber and the rest; all online platforms that when they started used online media. Now we’re seeing them use all the non-digital or traditional media.
James:
How much is Facebook spending next year?
Darren:
On outdoor, even newspapers are getting some of the spend because they are starting to realise that a lot of what you’re doing is at the very bottom of the funnel where you’re converting whereas you’ve got to keep topping up the top.
If you’re just converting at the bottom and not filling the top and I know the funnel gets people going ‘oh, it’s such an old model’. But it’s a very good way of thinking about the dynamic that happens there.
James:
Completely.
Darren:
So, what’s the future? That’s a big question. Do you see more clients potentially in Australia embracing this way of working?
James:
Yes, I think that data and martech will only increase. As clients have better access to data and their KPIs are based around different data sets you’ll find the usage of data to grow more and more. And once it’s demanded by clients it’ll be done by agencies.
Data is certainly not going anywhere. Blackwood is certainly the start of where it’s going to be going from a media point of view. I love the fact there’s no one else that does predictive at the moment. There’s heaps of media mix modelling that can get up to that point or look backwards but the predictive nature of Blackwood is what separates it out.
And that allows marketers to plan properly and to freebase; instead of having a trial in a new medium, they could have a look at what a different media mix would do to their plan without having to actualise it because the model will tell them what happened.
I think you’ll see potentially more risks taken—well non-risks because they’ve already modelled it. Advertisers going in media that you wouldn’t expect them to be in because their knowledge of the effect on their KPI is a lot more certain.
Darren:
But Thomas, you’d admit that this is not a platform for every advertiser. If you’re incredibly small the upside may be quite hard. Are you really looking at the larger advertisers?
Thomas:
Yes.
Darren:
The bigger the advertiser the bigger the upside.
Thomas:
Yes, you could say that. But from our point of view, if you look in the coming years we would also build some self-service solutions where smaller advertisers could actually use the platform.
So, it’s currently for the advertisers with a bigger budget but since everybody is getting squeezed on the market, marketing budgets are squeezed internally, agencies are squeezed on their revenues. And until people become more efficient, it could be modelling but all types of automation where you think faster and utilise data faster to get there faster is something that is coming to a high degree in the market.
James:
It starts at the big end of town but it’s coming down really quickly, to be democratised down into the small end. Thomas was talking about adding a SAS platform so that smaller marketers can use the model and the algorithm and can just plug straight into it.
It makes a lot of sense, a lighter version of the bigger thing. And that theory of QuickStart; at this end an incredibly robust predictive model and yet here, in very quick succession because these data points are incredibly available, can we get to a point where it’s not as long and it doesn’t cost as much to get to 70 or 80% or to have a look and have some clarity around what are the big drivers on my KPI rather than into that future space. They’ll be there in the not too distant future.
Darren:
Look, we’ve run out of time but it’s been a great conversation. Thank you, Thomas and James. Just one last question for both of you. Who do you think is the biggest competitor to what you guys are doing?
Ideal for marketers, advertisers, media and commercial communications professionals, Managing Marketing is a podcast hosted by Darren Woolley and special guests. Find all the episodes here