The way we do business has fundamentally changed, but it does not count for everything. The way we are responding to tenders in 2020 has barely changed from the way we responded to tenders in 2000. Tendering is time consuming and involves a lot of resources while the processes are error-prone, the chances of winning are low and there is no continuous improvement in place at all. Technology is lacking behind. Organizations are mainly using email, Excel and Word while responding to tenders. Sometimes a knowledge-based system is used to capture the most commonly used answers to generic questions. The knowledge and experiences are in the minds of people and are gone when they change roles or leave the organization.
Using Artificial Intelligence (AI), and specifically great technologies like NLP, Machine Learning and Computer Vision combined with smart data approaches and efficient UX could transform the way we respond to tenders completely. In this post we focus mainly on the first couple of subprocesses of the Tender Management process. The sub processes within tendering involving management, price calculations and settings, workflows, approvals, evaluation and submission are saved for another post and not described in this post. Below you will find 5 amazing examples to change Tender Management in a way that it feels more like 2020 rather than 2000.
Tenders are published on various websites. TED (Tenders Electronic Daily) is the online version of the ‘Supplement to the Official Journal’ of the EU, dedicated to European public procurement, and the most famous one within Europe. The first step would be that relevant new tenders get automatically discovered and prioritized by connecting to online and public sources. Smart search mechanisms can be applied to extend the coverage by manual searching. To be fair, this is all straight forward and perhaps not solving the biggest problem out there.
A second step would be that by the use of AI, a list of anticipated tenders get automatically generated. This is a list of tenders that could be published in the future based on for example past award notices and contract expiration dates. Afterall, it is better to start engaging with the customer and influencing the requirements before a tender gets announced and published.
Once the pack of documents related to a new tender is received, assessing the documents is a first step. What if you could analyze and enrich the documents, and extract the most important information out of it? AI can help with a basis analyses and data enrichment. Award criteria and weight or scoring mechanisms like price, quality, terms of delivery and knock-out criteria can be distracted from the documents. Same counts for other key elements like key dates, region or country, terms and conditions, budget, recurrence, certificates required, reference required and even content that refers to signs that a specific competitor was involved in defining the requirements. Finally the probability to win the tender can be estimated by analyzing historical tender data, historical sales data and, if possible, empirical pricing data of competitors. After assessing the tender documents, your tender team is able to more easily make a solid bid or no-bid decision.
It can be very useful to analyse historical tender documents and find look-a-like tenders. Although every tender response should be unique, some elements and solutions can be re-used and/or rewritten. This will save the team plenty of time and the ability to focus on the creative elements and tactics to win the tender.
The received tender documents are not the only source of information. AI can help you to mine public data and create a profile containing information like certain themes, context and content related to the company who published the tender. This will enable you to learn more about your customer, its vocabulary and important topics and to write a high quality response which should result in a higher win rate.
Within this domain we see two main use cases where AI can be applied. While writing a response it often comes to mind that a certain answer was given before in one of the tenders in the past. First, make it easier and better to search for information in old tender responses. This still requires some manual work, but searching for the right information in large volumes of unstructured data like PDF documents, images and Excel files can be quite a pain. Secondly, repetitive tasks of the tender response can be automated by suggesting the most relevant offering, descriptions or topics and assisting in the writing of the tenders.
In order to make sure assessing tender documents can be done automatically and is of high quality, tracking the performance and analytics of a tender once finished is vital. It means that tracking the won, partially won, and lost rates, lost reasons, competitive intelligence, and sales process efficiency for future analyses and predictions is essential information to feed the algorithms. Analytics can be added like geographical information, product/product category, and customer views. It can be enhanced with tender market overview information, supply vs demand comparisons, competitive analyses and performance KPI’s.
The 5 amazing examples of AI applied in tendering results in 3 positive business outcomes:
Author: Fedor Klinkenberg is the co-founder and chief executive officer at Brainial. He leads the company’s overall commercial strategy and teams. Fedor has over 11 years experience working at one of the fastest growing tech companies in the Netherlands: Mendix. Fedor holds a Master of Science degree from RSM Erasmus University Rotterdam.