Marketing funnels have had a bit of time in the sun lately. Mark Ritson believes they are the cornerstone of every marketing strategy detailing their importance in prioritisation as opposed to literal application.
Google has published The Messy Middle generating a lot of noise (it is 90+ pages after all), in which it concludes after several years of research and countless hours of work that ‘it’s complicated’. The model also avoids the non-linear aspect of pathways by forcing you through the middle therefore undoing the metaphor.
Finally, we’ve had Mike Follett at Lumen Research discussing the metaphors in marketing, why we use them and asking for more.
If you take these areas together you would probably conclude that funnels are useful from a conceptual/metaphoric perspective but not from a practical application perspective.
Ritson is right, they are useful but in a conceptual way. Google missed a trick by telling everyone what they already know and spoiling their own conclusions. And I haven’t seen anyone provide Follett with an alternative.
This is a problem. As a communications industry we should be helping brands make sense of the world, not confusing them with impractical concepts or telling them what they already know.
I’ve worked in this area previously, detailing a potted history of the path to purchase and there is a new option. A new model, practical in nature and with an associated change in metaphor that helps make more sense of the complexity inherent in human buying behaviour.
A new model
The first thing we need is something to help visualise the complexity of the path to purchase. One of my regular doodles is a pentagram star. My note pads are full of them. What’s interesting is that you can draw a pentagram without removing your pen from the paper. Each point is connected to every other point. This got me thinking, what if each point was classified as a ‘decision node’ or a point of inflection?
This interlinked model allows all points to be connected to all potential decision nodes. The benefit being there are no pre-formulated pathways or directions. As each node is linked to every other node this theoretically represents a vast number of potential paths to purchase.
This model is based on a fairly standard set of states – if all brands follow AIDA (attention, interest, desire, action) then as an industry we are already comfortable with generalisations. However, like Ritson I believe this could be flexible (to an extent) with variability by category and product. Of course a practitioner would understand the specific consumer states for their business or category but these are broadly similar cross category.
Here is that proposed model based on six different states or nodes, I’ve notionally called this the ‘Hankins Hexagon’, slightly tongue-in-cheek of course.
How does it work? The assumption at its core (and based on truth) is that pathways are non-linear and this has always been the case (despite the AIDA model’s best intentions).
So, what does that mean? Well, simply put, a person can make their own way from A to Z any way they choose. In reality there are very few ‘fixed’ pathways and most are two-way (feedback loops and changes of mind). This model posits that an individual can start wherever and eventually make their own way to purchase, that is if they do buy in the end because not everyone always gets there…
Fairly general categorisations have been applied here as there are commonalities for many categories, even if these ‘nodes’ last for a micro-second. There will always be a challenge thrown in with this generalisation but a good way to overcome it is to use a common everyday purchase like milk as a demonstration.
You need milk quite a lot, maybe once a week. Some people need it every day. For the times you don’t need it you sit in the ‘passive assimilation’ node exposed to all those Arla ads (passive assimilation is where you spend the most time in most categories).
That’s the actual point of advertising after all, to increase the probability that a person chooses your brand over another and in this model you can prioritise where you invest to improve those pathway probabilities.
You then run out of milk, which is the trigger to need. It’s very rare to ‘actively evaluate’ milk but you may wander to the shop and scan the milk section (semi-skimmed, full fat, Jersey, etc), compare (private label versus branded) and then you’ll purchase. The timings may be different to insurance but it’s still broadly the same idea.
It’s worth noting at this point that someone else may just buy whatever is on offer at their corner shop (fewer choices equals no comparison) so their pathway goes passive assimilation, trigger, purchase. Try it out. It should work for most, if not all, categories.
The interesting point is that for different brands and different categories you’ll probably find a dominant path because the lines between these ‘nodes’ are actually probabilities. That’s the actual point of advertising after all, to increase the probability that a person chooses your brand over another and in this model you can prioritise where you invest to improve those pathway probabilities. Much like digital touchpoint analysis, this can then be turned into a decision tree.
Obviously this can go to the nth degree with multiple different pathways, however when aggregated it’s clear that a normal distribution will be in effect (i.e. most common pathways will take up the core 68%).
What those paths are and where they sit with regard impact volume will act as a guide to where marketers should invest to either optimise or grow. This is the key application. Marketers could utilise all those newly-hired data scientists to develop and articulate these probabilities using behavioural data.
Marketers have limited assets but mapping out the above and by using Bayesian probabilistic estimates and methodologies you can prioritise where you invest your cash to maximise penetration. What are the most likely routes to sale in your category or for your brand? Which nodes are important, which less so? Is your brand different from the category? Are you investing in a way that leaves cash on the table?
Essentially what a marketer could do here is use their marketing budget to optimise and keep as many people ‘in play’ as possible, to the point of purchase and then beyond.
Another, perhaps simpler application is that expert researchers could identify new metrics to rank brands based on comparative node scores. This would diminish the probabilistic nature of the model and its newness but it could still be useful (even if a lot closer to old funnel thinking).
Both applications provide clear strategic diagnostic capability when contextually compared to the category as they could show weakness or strength, or movement versus a base. Measuring changes over time to identify trajectories would be the next step down in terms of measurement but no less important.
The relevance of pinball
At this point it’s worth returning to metaphors and one becomes increasingly appropriate. What else deals in probabilities and geometry, utilising certain influencing factors to keep an object ‘in play’? Pinball.
Applying the model, marketers can use their investments as ‘flippers’ or ‘bumpers’ to guide people to final score (that is an extra ball) with the impact of communications acting like geometry to improve the probability you’ll score highly.
Now, I’m not a quant or maths whizz but this technique is eminently actionable. It replicates the pathway analysis often done for digital touchpoint analysis but extends it out. You’d naturally see a normal distribution curve representing the common paths (points scoring) with tails representing those paths that are unlikely in sector.
They say success has many parents so I must admit there is thanks due to James Caig who used a similar metaphor for his post-economic crash IPA paper on teacher recruitment.
Returning to Google and the Messy Middle construct, rather than a squiggly line you can take the model and easily overlay it on the Hankins Hexagon and the labels sit nicely.
The biggest difference being that the hexagon leaves all paths open rather than forcing you through the messy middle. Someone could travel from passive assimilation to purchase immediately, no middle at all with zero behavioural evidence that you’ve been triggered or explored or evaluated.
It’s clear from playing pinball on the Hankins Hexagon that this could help as we have probabilities and prioritisation, which can be applied by marketers (with some mathematics help and research) to make better decisions.
This, after all, is what Ritson wants with his juggling funnels and what Follet wants with new metaphors – useful, applicable tools to enhance shared understanding and planning.
It requires a new way of thinking and further development from a mathematical and measurement perspective but hopefully having brought it to light, better minds can build and enhance and we can finally leave the funnel behind in the 18th century.
James Hankins is a consulting strategist and founder of Vizer Consulting.