TLDR: Marketers are usually learning from the past results of their campaigns and adjust their strategy to hopefully meet their goals at the end of the month. What if you know the results of your month even after the first few days? You could take mitigating action soon enough to achieve your goals, you could please your boss that this month will be OK, or maybe you could even optimize your campaigns towards the future! Sounds impossible? Learn more about how marketingintelligence.io utilizes rocket science math to do this.
A story of status quo – you cannot forecast the future results of your campaigns
Imagine Peter – a CMO in a famous online retailer selling winter jackets. Peter has a 30.000 USD budget for this month and should deliver 200.000 USD in sales.
Peter has a lot of experience from previous campaigns so he will allocate certain budgets to certain channels but still running marketing campaigns can be like betting on the weather. You never know how your creatives resonate with the target audience, what competitors plan, how expensive your ads will be etc. so the only thing that Peter can do is to wait for some results, optimize his campaigns, wait for results again, see how his budget is getting exhausted and repeat this process. So this is exactly what Peter is doing…
Peter’s team brings first results after one week and it seems that all channels are behind the target but not that much and as it is the beginning of the month there is still a good chance to catch up. Peter just says to his team to bid a bit higher so that the traffic increases and so do the sales.
At the next meeting, Peter is very concerned – increased traffic did not boost sales as much as anticipated – we are at 70.000 USD of the revenue target and what is worse, nearly ¾ of the budget is exhausted and only half of the month is remaining! All people in the team agree on a battle plan including a lot of changes. We have to double the speed of sales!
Every team member worked at least 10 hours a day – recreated most of the campaigns, added new creatives, played with bidding and followed all possible best practices. Peter is checking Google Analytics on a daily basis waiting for good results.
The first days look ugly but the team considers it normal since campaigns are new and algorithms of Facebook, Google and other networks are picking up. But after a few other days, the team knows that it is not going to be good.
The last week of the month begins with a traditional meeting where the teams are presenting results. Peter knows already from Google Analytics that revenue is only at 110.000 USD but what was even more concerning was that we have only last 5.000 USD left. The team was throwing the guilt at other departments, competitors and of course not forgot to mention that the company “is selling too expensive crap, so no wonder that people do not want to buy…”
Peter knew that even when they do a miracle the last week they will not deliver 200.000 USD sales. He should go to his boss and tell him the news – but how much should I promise him?
“We have found too late that marketing will not deliver desired goals and in attempts to turn it around we have made it even worse… “
Is this familiar to you? Continue reading…
How Marketingintelligence.io employs math to change this story?
Let’s rewind time. We have the same company, same Peter, same team and same initial set-up – the only difference is that Peter is not just reading data from the past but he is actively using all his Marketing Intelligence data.
Peter allocated certain budgets to certain channels, his team drove the campaigns the same way and brings first results after one week. Peter and the team see in their Marketing Intelligence dashboard that all channels are currently behind the target but they have a very productive discussion when they see estimated future development of every channel in their Marketing Intelligence dashboard. You know why?
Imagine that Marketing Intelligence has downloaded data from all possible platforms including a large history of different campaigns, the performance of different target segments, products, channels… Simply everything that you can find in Google Analytics, Google Ads, Facebook, CRM, RTB system, etc. Marketing Intelligence creates models including the current performance of running campaigns, similarities with previous campaigns, calculating seasonality, deviations and complete conversion path for every possible cookie. These models can utilise all the data from the past and get trained to calculate the future development of your campaigns in just a couple of days with 92% accuracy!!
Thanks to Marketing Intelligence Peter and his team sees exact predictions of what will be revenue at the end of the month delivered by every single channel even calculated for every single future day.
This enables them to identify that Facebook campaigns are trending well but they need to immediately do something with Google and decrease RTB campaigns. The team is happy that they know what to do and every next meeting is getting Peter and his team closer and closer to their target.
In the third week, Peter sees that they will actually achieve 220.000 USD with their budget and therefore goes to the office of his boss to tell him this great news.
The end result after one week is 217.000 USD but his boss does not mind this 3.000 difference:
“Knowing what will likely be our results in the future enabled us to increase the final order from our supplier and make sure that all our customers receive their products before Christmas.”
Sounds like black magic? Where is the weather and what if competitors go crazy?
We see that our models are very precise because they are already calculating with competitors’ future development based on current data and usually there are not drastic unexpected moves. But you are right that even the best mathematical model cannot predict that for instance, new coronavirus appears. Even in such a case, Marketing Intelligence would learn after a few days that there is something significant going on and change predictions accordingly. So the question is why not to utilise fully your historical data fully to foresee the future?