Automated Product recommendations – Does Software Better Than Human Sales Staff?

The article deals with the possible uses of automated product recommendations in sales and customer service. There are buyers who prefer to buy on the internet for there they do not have to talk to human sales representatives. They probably had some experience with vendors staff hiding their cluelessness behind a lot of talk and are trying to sell their products on stockage at anyone who does not get away fast enough.

The problem looks hard to solve. But wait: for a few years now it is possible to extract valuable data points from primarely web activities and connect them with realised sales. The connector are either self-learning models or static traditional models. Models simulate behaviour and results.

The interested customer is looking für information before buying. He or she may ask a specialised salesperson and consultant who has all the relevant answers ready and can think in the customer’s interest. This relieves the customer. Or they, the potential clients, may look for publicly available information and customer review to avoid the advice of an inexperienced and overburdened sales representative.

Many sales representatives for complex products are quickly overwhelmed by the questions and expectations of well-informed prospects. The untrained salesperson feels attacked, the sales talk goes off the rails, the interested party prefers to buy from a supplier without advice.

The web part of customer journey

Depending on the product, between 80% and 90% of the buyers begin their customer journey with an search engine. So it is important for the seller to know what it is reported in the web, take care of the correctness of web information and avoid contradicting information. If there is human sales staff, they are supposed to assist the web customer journey and take care of a seamless process.

The consultant with all the facts in mind and the fully automatic customer advisor are extremes, the practical truth in the company lies in between.

Can the web be the better customer advisor?

Yes and no. The informations in the web and the informations given by a human sales person are supposed to go without contradiction. Nobody likes lies and deception.

Artificial intelligence makes use records of any communication with customers, for example chats, mail, data entries. This is compared with predetermined success goals. The model recommends something and can adjust itself. Something salespeople do intuitively, too, if the selling company lets them. Artificial intelligence therefore means less control for the management. On the other hand, opportunities are used that might otherwise have been overlooked. Where to start now?

Sellers are supposed to make as much information as possible available in a neatly structured way. Many people would like to find out more before making personal contact, but cannot find the information they are looking for. If publicly available documentations saves half an hour of calls per working day for the seller this justifies an expense of €2,000 for an improved sales website or other type of customer information – and that for telephone staff at a standard wage. Informed customers are better customers.

The meaning of human interaction

The ideal salesperson knows the buyers needs and wishes. He ors she build the best possible solution with his expertise. In between, looking up details and availability on the computer is essential for complex inquiries.

Advice form humans or from machines and websites?

Whether or not a human intermediary remains necessary for information collection and transmission depends on the product itself and on the personal preferences of the interested party. Some can evaluate and implement the freely available information, such as product descriptions, customer reviews and more for their own needs, while others cannot.

Good sales people relieve the customers. To do this, they must put the interests of the customer first and not fight for sales at any price. Consulting software in the sales pitch gives standardized information, if the software is capable of extracting the recommendation which work and then the client might be really happy.

Correct information with automated product recommendations
Correct product informations saves money for buyer and seller

Success has many reasons

Incidentally, the screen of the gentleman can turn directly to a prospective customer without the computer interfering. And if he needs information from the cloud or machine, he has it quickly at hand.

How prices are perceived from seller and buyer

Pricing affects everybody. Selling your apartment, selling used furniture, doing salary negotiations or finding prices for your products. How to find a price liked by the seller and buyer? In my seller history I found situations with low prices where the customer voluntarely added a gift, and many situations where nobody wanted to buy. The reason for this can be “no market for the product” and “price is prohibitively high”. The latter one means there is a market for the product, but not at the charged price.

The next chapters give structure to this complicated matter.

The classical theory of pricing for optimal is very simplified. It assumes a linear demand function, with only one determinant, the price: demand=f(price) . This means only demand determines the price. What happens if demand has many more determinants?

This can be:

  1. The look of the selling (web-)site
  2. Do the salesmen make a good job?
  3. Do I get good advice from the staff?
  4. Can I expect some value if I pay more?
  5. Does the product make me feel better?
  6. Do I have the budget for luxury?
  7. What are the prices of the competitors?
  8. The price the seller wants

Point 1-4 are the value to the product added by a good sales team. Example: if a retail store buys a pallet with 1000 packets pasta for 600 € and sells the package for 1,2 € the sold package is not the same product as on the pallet. The services of the retail store are added.

The following graph is an example. It may look different for a specific produkt. The graph with the numbers given there works good for used cars and expensive utilities sold on Ebay or webshops.

What determines the price, pricing
What determines the price, pricing

Example: make advice a valuable good

Many brick-and-mortar retailers complain that their customers do not pay for advice. What they do not tell: how much do their prices differ from common internet sellers without any consultancy? Is there really value added in the store?

In old times, when sales was based on printed catalogs, there was the famous trade costing, with list prices, heavy rebates, and purchase prices. Most retailers ordered their stuff at wholesales, which themselves made their cuts. The result was that end user prices were about 3-10 times as high as the factory price. I understand that many want this time back. The customers had no choice, they had to buy locally. Many wishes and dreams of customers weren’t ever fulfilled for the high costs of maintaining catalogs and storage.

Now many internet retailers order directly at the manufacturers, and they have to look for competing offers worldwide.

Selling Services

There are two approaches for long-term pricing: with interchangeable goods, many manufacturers, in the long run the price will be the production cost plus some earnings. The earnings are the incentive for producing. Examble: a spare part for an historic car. Production cost is 150 €, with this part added the car gains 2000 € value. If the seller knows that and wants a high price, the buyer might look for another source which produces it for 150 € and pays additional 50 € for the favor.

In Services it is a bit different. Production costs is one side. The other is: how much will the buyer profit from my service? If the seller, you, knows that 25-50% of the profit for the seller, the trainer or consultant, are normal.

Is there a formula for prices?

In economic theory the price is equal to marginal use. That means the price equals the benefit the buyer thinks he has from the last unit he or she buys.

If we can measure the benefit in money and realtime, we have the formula.

The benefits are listed at the beginning of the article. The problem is that the customer does so many estimations, that change over time. Also new competitors with low price entrance strategy might show up.

When selling over a website it is quite easy. Look at the ones who buy related to the ones who look at the article. If there are many who look and do not buy then lower the price, if most of the visitors buy then raise. This needs a lot of finetuning. Sometimes visitors look a few times at the articles until they finally buy. When they see that the prices go down, they might wait even longer. So you might want to include the total number of visitors.

How to calculate?

There is a mathematical formula and there are ways to survey it.

We measure the first eight determinants, with due consideration of costs:

determinanthow to find and count
The look of the sales (website) position.Survey of visitors
Are the salespeople doing a good job?not available everywhere, get feedback
Do I get good advice from the staff?Get feedback
Can I expect value if I pay more?Tests and surveys on buying motivation, own samples
Do I feel better because of the product?Tests and surveys on buying motivation
Do I have the budget for luxury?Experiments, interviews
What are the prices of the competitors?Research
The price the seller wantsLook deep inside you

Market research helps with the first six points. Usually it is necessary to repeat the market research every 3-6 months with the well-known market research problem: the target person doesn’t like to answer. They are bored and do not know whatfor they took their time to answer the questions. We must therefore look for other methods. An easy-to-install method is to encourage visitors to comment. You may get 1-5% comments from the customers. This responses add up to valuable data source. It is also possible to conduct studies with paid testers. There are some problems with representation of the correct user group there, additional research helps to circumvent this.

Photo du titre by Egor Myznik on Unsplash

Price Negotiations and Decision Tree

Market research includes personal coaching. This one is about helping the client into negotiations with better information about the market.

The party with the most information has advantages in negotiations. They know the negotiation scope of the other side and th bargaining power do both sides. Other questios are: What do I want? How to avoid getting lost in bargaining? After looking into auctions, where for psychological reasons it is difficult to make rational decisions, I look at the decision tree with estimated earnings and probabilities.

Auctions – the risc of following the group

Auctions are popular for antiques, used items and real estate. In the bidding war the offers may run away while the auctionator counts highet and higher. This war requires at least two bidders. Participants see and hear competitors bid higher and higher and infect one another. The bidding continues, so the race goes on for not having a defeat. Giving up means the object is gone. Due to the short decision-making period and the group, spontaneous decisions are made, which be confirmed experimentally. As soon as the bid is accepted, the bidder goes back to his rational normal and recalculates: the high purchase price makes the investment unprofitable. The buyer tries to get out of the contract. This phenomenon occurs less often if a security deposit to be presented limits the bids.

How a decision tree helps

We try to look at every possible scenario and calculate the probability of that scenario. The example is about a freelance teacher, who applies for teaching a class. How much money will she get?

decision tree usable for price negotiation – probabiliy and earnings

Sandra is applying to be a lecturer at the Montgomery Training Center. She has no experience and is pretty much broke right now, so she urgently needs paid work, her lower limit is 20 euros an hour. But she wants to achieve the maximum possible hourly rate in order to look good in front of her friends and colleagues and to earn money.

Sandra negotiantes with the dean, Ms. Dr. Teufel. She wants more earnings for the school and save money spent on teacher salaries. If teachers are scarce or one has special qualifications, she can pay up to 40 euros per hour. She tells the applicants that there are lecturers who work almost on a voluntary basis. So it tests their price scope downwards. That the training center also hires expensive lecturers is not for the appliants ears.

Doing background research helps

Sandra asks herself how much she could ask. She estimates that the school can spend up to 50% of the participiants fees for lecturers salaries[2] . The difficult question about the income of the training center can be answered with the help of price lists of the company, sales figures from the Federal Gazette (corporations have to publish there), inquired or estimated numbers of participants. Sandra does the math and comes to a maximum of 40 € per hour.

Sandra analyzes her competition. What is the likelihood that another qualified person will do it cheaper? This is very high when teaching in university cities. Public statistics on wage levels can help, or a survey among friends. Industry associations often have fee statistics. She heard from friends that they often only pay 25 euros per lesson. So what to do in the price negotiations?

Price Negotiations and Background Research: Add Probabilities to the decision tree and play with scenarios

The probability that a sufficiently qualified applicant appears charging only charges 25 euros per hour is 50%. Let’s figure out whether Sandra should play it safe and demand 25 euros per hour or whether it is worth asking 40 euros per hour. The contract lasts for 200 hours, i.e. 5000 euros at 25 euros per hour and 8000 euros at 40 euros per hour.

Sandra realizes that she still doesn’t know enough about the lecturer market. What is the probability that in the case of a rejection by Dr. Teufel, a new job of the same type appears that brings in at least € 40 an hour? It is not possible to ask the competition, they will hardly tell the truth. Miss Dr. The devil is also talking about volunteering, so no wages at all.

Looking at the decision tree: if demanding 25 € per hour Sandra has the job safe and gets 5000 Euros. Demanding 40 € there is a 50% chance that she will earn 8000 Euro and 50% that she needs to go cleaning houses for 12,50 Euro each hour, that makes 2500 €. In addition and weighted by probability this goes to 5250 €, slightly more than talking about the low price.

External Factors – Prestige and Feeling Safe

Sandra possibly thinks she definitely needs the teaching job. She no longer wants to clean and needs references. Then she plays safe and only charges 25 € per hour. Demanding high prices is only worthwhile to a limited extent, as the previous analysis shows. The opportunity costs in the form of stressed nerves can be an argument for low claims.[3]
————————————————————–

  1. [2] To know this average value, information about the business model of the schools is necessary. The basis is the commercial calculation with cost price, handling costs – and profit surcharge and sales price. The lecturer’s fee is the cost price, the remuneration paid by the participants per lesson is the sales price. The trading surcharge includes rooms, advertising, administration and risk. [3]
  2. [3] Between 2008 to 2010 I did some tests with groups of 30 participants about that subject. The results were that very young people tend to ask too little money, older people tend to ask too much in price negotiations for work.