I am constantly amazed at my bipolar expectations of the Internet. On one end, my online and mobile banking apps give me exactly what I want the first time. I do not need to scroll through dozens of pages to find the info I’m looking for. I can easily download relevant content like statements, and there is little or no extra content that is irrelevant to me.

I feel like the app is or can be customized to my needs. Yet at the other extreme, when I search the Internet for business needs, I feel like I’m trying to find a needle in a haystack!

Googling for Business Needs

Here’s a simple example of why traditional Internet search engines are not the best at meeting business needs. Suppose I manufacture portable GPS devices for hikers, and I need to find a lithium battery manufacturer to power my units. So I Google “lithium battery manufacturer” and, to my surprise, Google has found about 4,170,000 results. That’s more than I need, and much more than I could ever sort through.

I look at the first 10 results, figuring they should be the most relevant. Here’s what I found (note: Google results vary from user to user, so your results for the same search term could vary):

• One battery retailer
• A link to 10 manufacturers of lithium batteries for electric cars
• A Zacks investment article on lithium stocks
• An article on the stock value of lithium battery manufacturers
• A lithium battery manufacturer
• A glossary of battery technology terms
• Four press releases on the lithium battery manufacturer industry

So far, I am one for 10.

As I peruse a few more pages of results, I do find a few more lithium battery manufactures, but they are sprinkled throughout a mix of other, unrelated topics. And the deeper I go into the results list the harder it is to find what I’m looking for.

Exact Google Search

Considering my general Google search didn’t yield me very helpful results, I try Googling an exact phrase. This limits results because they “must contain” my search term word for word.

This time, two of my first 10 results are lithium battery manufacturers, while the rest are mere directory links (e.g. Alibaba) or news items (not helpful).

All in all, the accuracy on the first (presumably the best) 10 results of each of my general and exact phrase search is between 10-20%.

Of course, if I wanted to further leverage my search results, I could analyze a few hundred of them to determine which were most relevant to my lithium battery needs. To do this efficiently, I would love to be able to export the first few hundred results and filter out the irrelevant ones. Unfortunately, I can’t easily export my Google search results. So I patiently, and manually, plow through the results and cut and paste interesting possibilities into a spreadsheet. Later I’ll have to research every company to determine its potential compatibility.

This manual approach works, but it requires a lot of labor. And even then, I suspect that there would be “nuggets” I would miss during my manual processing.

Powerlinx: Building an Accurate Business Search

This labor-intensive approach to finding relevant businesses is not practical for Powerlinx members. Each of our members has unique needs that call for an efficient and accurate business search system. In other words, our members need an automated process that will do the research and sorting for them. This is where the Powerlinx matching engine comes in.

How the Powerlinx matching engine works is proprietary and complex, but here is a simplified version.

We start by building and maintaining a huge custom database of company information that we compile from publicly available information. It’s important that we compile this database ourselves instead of licensing a traditional business list like Dunn & Bradstreet’s because our matching algorithms have specific, unique and demanding data requirements. Here are a few examples of the information we compile:

Company website

We use this as a basis for collecting company information and partially for validating the company’s existence.

Company description

Once we find a company’s website, the matching engine uses specialized algorithms to decipher properties like the company’s core business and business type (manufacturer, distributor, etc.). The algorithms even take synonyms into consideration, so phrases like “manufacturer” and “producer” are treated similarly. So if a member is looking for lithium battery manufacturers, the matching algorithm can find appropriate partner opportunities with extraordinary accuracy.

Business tags

The matching algorithm incorporates tags in order to supplement and corroborate company descriptions. Since tags are usually put in place for SEO purposes, however, they can be overly broad. Take the example of a mattress manufacturer, which might use tags like “furniture manufacturer,” “bedding manufacturer,” and “futon manufacturer” among dozens of others. The matching engine’s task is to remove broad terms like “furniture manufacturer” and reduce the remaining tags into singular terms. The engine’s tag refinery system sorts out synonyms so tags like “bedding” and “mattresses” are not counted twice.

Tiebreakers

In the case that two businesses’ cores, descriptions and tags are similar, the matching engine uses data like awards, associations and news events to distinguish which company is the better match.

Naturally, there are many more parameters we incorporate into the Powerlinx matching engine in addition to those listed above. As important as this information is to proper business matching, there are few or no huge global business databases that list this kind of information, which is why we decided to create a web-based data mining solution to find and process company information to match businesses with relevant prospects. Some of the matching parameters are what we call free text (such as company descriptions), some are categorical (such as geographic region), some are quantitative (how does one define a “medium” sized company?), some are expert-defined (startups in biomedicine have different needs than startups in apparel manufacturing even though they might be looking for the same kind of business partner) – imagine a Google search that could process all of that!

Welcome to Powerlinx, where we help you grow your business by finding you the right partner.

Featured image by epSos.de