The long tail

Effectiveness of on-line marketing depends on the available tool set. Customers become immune to omnipresent marketing and to succeed one has to shuffle through different strategies or show them they can benefit from it…

I remember times when a landline was something to dream of, and the moment the dream finally came true. Unfortunately I can also recall the moment when the telephone became a nuisance. It was ringing again, and I was about to hear a sweet voice announcing “Your number has been selected…” Our number was unlucky enough to be caught into the databases. 

Telemarketing made distance selling easy. As the number of phone users skyrocketed, so did the applicability of telemarketing. It kept spreading to different fields until its use finally got out of hand with the monetisation of the databases. This insolent exploitation stemmed from the belief that no one would give up on such an indispensable medium of communication. People developed various defence mechanisms. Some kept blocking the intrusive numbers and for some the ringing sound became indistinguishable from the background noise. Some learned to just put the receiver down. Gradually such business model became obsolete. Or rather it became popular elsewhere.

the cookie

Internet may be an ocean of knowledge but it is also a hunting ground. Marketing follows potential clients: on sites where they spend considerable time they would surely see ads. That is the price to pay for free surfing. What do we pay with? Personal data: clicked items, time spent on different web pages, our areas of interest, our shopping lists. Digital trackers will promptly pick up traces of our on-line behaviour.

To collect this information browsers need the infamous cookies, small files that store client’s ID which allows for the aggregation of data describing client’s web activity without using his or her name. There are countless companies that offer data collection mechanisms, from the internet giants like Google and Facebook, local e-commerce hegemons – Allegro, Ceneo, down to totally niche solutions. There are also businesses that combine and utilise the aggregated data.

Déjà vu

This model does not allow for a direct contact like email or a phone call, but will be perfect for intermediary action – serving ads that might be of interest to clients. Rings a bell? Are you seeing ads of a laptop you checked yesterday? Same or similar? It’s a bait to lure you back to the shop.

Maybe the client will come back to finish shopping. But what if he’d bought the product somewhere else, or she’d changed her mind? Not much reason to come back. But the ads will keep popping up for a while, echoing past clicks.

Echo backed by AI

To alleviate this, algorithmic models moved on, analysing not only the recent events but the whole shopping history as well. Thus by comparing matching data of other people it can infer needs and preferences and present customers with relevant ads.

How accurate is this guesswork? It is error prone, depending on the volume of data at hand. AI algorithms (and specifically: Machine Learning algorithms) follow the simple rule: more data – better accuracy, but the accuracy will never reach 100%. See the diagram below.

Google changes the game

Where to get the data from? The biggest players have them. The smaller the shop the longer it takes to collect enough data – i.e. the longer it takes to reach the designed efficiency. Patience will be rewarded. Or will it? Not necessarily, because Google, the master-of-the-cookies, announced important changes, starting in 2022. Then Chrome browser will not allow any external pages to track users using cookies. This cookie-wall will keep recommendation systems away from data. So, we’re back to the drawing board.

Successful salespeople

How to serve customers who do not sport this long tail of their shopping history? Or enter the shop incognito? The best strategy is to co-operate. After all this is what seasoned salespeople do. They talk with the client, gradually gaining all necessary bits of information to finally suggest the best solution. 

Instead of trying to guess the needs of our customers we can ask their current (here and now) needs and preferences. Equipped with this knowledge we can offer a handful of much better matching recommendations for the customer to choose from. Such method will directly translate into higher e-commerce conversion rates. A satisfied customer, freed from misplaced recommendations, will be happy to com back for another purchase. And this is exactly how feeCOMPASS® works.

If you are interested in feeCOMPASS solution, please learn more about functionalities of our solution and examples of applications in selected online stores.