Using CRISP-DM method to analyze Big Sales Data
Ready to take your sales game to the next level? With CRISP-DM and smartsales.ai, you can analyze and optimize your process for shorter deal cycles...
If you google "How to increase conversions," you'll get output from various subsets of narratives like:
"5/10/13/69 ways to increase conversions...". And it's usually a random mix of the same type recipes:
- Know your customer better.
- Improve UVP.
- Make onboarding easier and sexier, etc.
Well what can you do, that's how SEO works: you have to write simple articles that are accessible to robots.
That said, it's hard to find studies by HBR or reputable academics on what factors actually affect B2B sales conversions.
Knowing conversion on its own doesn't accomplish anything. It is a relative value that is important in comparison. It's best to consider conversion in conjunction with the surrounding context, such as relative to benchmarks, market stage, product, business unit, employee, etc. Or change over time. Conversion is a measure of business process efficiency. Everyone wants to have a high conversion rate, but what does "high" mean? It depends on many variables. Conversion can be useful to identify:
- Bottlenecks in the process.
- Losses in the process
- Best practices
- Effective employees
- Effective campaigns/UVPs/products, etc.
Changing conversions over time shows a change in process efficiency.
Compare yourself to your competitors or how effective you are as a manager.
Every sales director should calculate conversion rates by hand at least once. It's a very creative exercise. In the process of calculating it, you start asking all kinds of questions about the relativity of being and the transience of time. Mostly it is about keeping track of time intervals, so that the numerator and denominator have the same basis for calculation. But let's keep it simple and not get bogged down over it. Also, let's not take into account the formulas sewn into different CRM systems. Simplify to the maximum:
Formulas for conversion calculation
For the calculation we have to choose two stages of the deal that we will compare with each other. Let's divide the deal into 10 stages and choose the stage when we have already formed SQL and the target stage - Closed Won (CW). Usually there is no point in taking the numbers before SQL, because there is a lot of trash and untargeted bids. This area is considered separately for the analysis of the effectiveness of marketing campaigns, let's skip.
In order to use formula (1), we need to take the volume of all deals at a certain period of time in the CW stage and recall how many total SQL was in the same period when these deals were in the SQL stage. Although, for the purposes of our article, it doesn't matter whether we count in money or currency, it will be important for you to draw conclusions.
What's the problem here:
1. It's unlikely that anyone has a snapshot of the pipeline at that point in time to correctly calculate both the denominator and numerator, especially in money.
2. All trades moved differently towards CW and some came in faster, some started much earlier.
3. During this time, the amount of deals tends to change: anything can happen between SQL and CW.
In general, there are a lot of nuances, which are not realistic to take into account without a proper database and complex BI calculations. In the calculations it is difficult, but in work, this formula is still useful.
So, the formula (2) will save us.
We take the pipeline at any point in time and look at the two extreme stages: Closed Won and Closed Lost. Usually by this point, the transaction amounts are correct, allowing us to calculate both conversion by amount and conversion by number. Time span, employees/products/stages and n&l filtering is easy enough.
It is established experimentally that as soon as you display total conversion and conversion of each division/employee on the dashboard and show it to all participants together every Monday, the numbers magically start growing. The effect of psychology on metrics growth.
It doesn't matter which of the two formulas above we use to calculate, it still works both at the same time. By qualifying leads better, you can increase the conversion rate not weakly. We once managed to increase by 18% just by implementing [a better qualification framework](https://blog.salesai.ru/the-complete-guide-of-meddicmeddiccmeddpicc), which had more criteria, i.e. was more accurate.
If we break down the entire transaction into stages and calculate conversions of transitions from one stage to the next, we can identify bottlenecks and work with each one consistently. Obviously, any complex task can be broken down into many smaller ones that are not as complex. Likewise with a deal. You need to identify all the conversions of all the related pairs of stages and start with the lowest one, working through each pair sequentially, going from the worst to the best.
One way is to break up the stage into several more "sub-stages" to influence each one more effectively. If the client can't make a big move to meet right away, offer them some small steps. For example, not a paid pilot, but a free test for a week. Even easier is a demo, even less is a reference visit, more is a case study, etc.
If managers can't solve some problem head-on, then obviously you can try different hypotheses on how to solve this problem in a roundabout way.
For example, if clients don't respond well to a meeting, suggest they go to a webinar or download an e-book before the meeting. Hypothetically, both should be easier to take than the meeting. The main thing is not to proliferate them to infinity and to remember to keep the end goals in mind. Options in one plane can be an infinite number, but in the chain it is necessary to remove the most inconclusive.
And it will help to remove unnecessary steps that only delay the transaction, it is obvious that it is necessary to analyze won and lost transactions in order to identify successful practices, in order to constantly recycle the process, throwing out unnecessary movements. To do this, it is necessary to reconstruct step by step the movement scenarios of 5-10 successful trades, find among them repetitive events and fix them.
The same must be done with the lost trades. And change those steps of the process that were in the losing trades, but were not in the winning trades. It's something like Process Mining on your knees.
Win/Loss analysis allows you to find out why and how SQL turned into a new client (or not).
Win/Loss analysis is the process of examining past transactions to evaluate why SQL became CW or CL. The knowledge gained from such an analysis can play an important role in growing your business and increasing revenue.
It is generally believed that price is the most important factor in making a buying decision, but this is not always the case. Finding out what other aspects played a role in winning or losing can ultimately give you an advantage over your competitors' sales tactics.
Win/Loss is a type of CustDev, but after the deal. to conduct Win/Loss, you need to prepare two questionnaires and go to the customers, asking all the same questions. The analysis of the answers allows you to get a lot of conclusions that you have not even guessed.
This is basically a combination of items 2, 5 and 6, where you collect the whole picture and understand who your client really is, how you found him, why he bought you and what value he really got from you. Most likely not the value you think they got. A job well done on all three counts won't linger with the result because it will benefit all stages: both marketing and sales and product.
But it's impossible not to remind about universal ways of working with any problems, or rather ways to turn a problem into a task:
Getting both methods right allows you to generate enough hypotheses on how to improve conversion rates and run them into testing while the research described above is being done in parallel.
Reverse brainstorming allows just a list of hypotheses of the reasons why conversion is low to work with them with different marketing tools, content, etc.
If you consistently ask the question Why...? at least 5 times, you'll always get to the root cause. And once you understand the root cause, you can use other tools to treat it.
The "5 Why's" when looking for the causes of low conversions
The root cause is the root cause of the original event (low conversion) that, when addressed, prevents the original event from happening again.
If we can identify the root cause of the problem, then we are able to identify the necessary corrective actions to solve the problem.
Knowing how to count conversion across all transaction cuts/variables will allow you to influence it more effectively. Make more informed and effective decisions, don't waste resources on dead leads, manage your marketing campaigns, don't keep ballast on staff and your business will be healthier and more resilient to all sorts of VUCA/BANI.
You'll grow faster and be able to compare yourself to competitors in your industry, region, etc.
The next step will be to build a system that will predict process failures and signal in advance the risks of lower conversions. But that's another article)
smartsales.ai is an effective tool for making decisions based on data obtained directly from customer meetings, where customers' words are turned into clustered data without distortions, cuts and wasted time, to know exactly how to increase conversions.
Ready to take your sales game to the next level? With CRISP-DM and smartsales.ai, you can analyze and optimize your process for shorter deal cycles...
Remote management and team motivation is still a blind spot for many sales directors. Let's look at what mistakes prevent proper communication with...
Case study about SMB that counts money and requires a 100% return on investment. Read how we got an ROI of 328% already during the POC.
Be the first to know about new B2B SaaS Sales insights to increase the conversion rate with the tools and knowledge of today’s industry.