My wife and I love to take vacations. However, we approach our shared trips very differently. My wife likes to ‘explore’ new places and without planning the vacation in advance. Instead of doing research, she likes to get there first and then ask locals where we should go, what are the ‘must sees’ for our visit. I, on the other hand, am a confirmed Planner! I look up the ‘highlights’ on the Internet and then plan our travel in advance. Why? I don’t want to miss out on anything.
Most of us can relate to this example. However, as a leader of a large team of IP and R&D searchers, I like to try to get the best out of the ‘Explorers’ and the ‘Planners’ in my team. This means knowing the advantages of each approach. Or more importantly: how to achieve the best results for our clients by getting these ‘partners in search’ working together.
Let’s explore!
No different from my ‘holiday’ example, Explorers don’t follow a set methodology to find results in a search project. Instead, they use their knowledge to predict ‘hotspots’ with a high probability of finding results. For example, while doing a novelty search for an invention, Explorers would usually define the ‘universe of results’ based on their knowledge of the field, and then try to make multiple narrow search queries within that universe to find best results in shortest possible time. The universe of results means a set of documents which includes all possible relevant results. Simply put: if someone analyzes all the documents in the Universe of results, recall1 of the final novelty search report would be 100%.
Explorers focus their energy on quickly finding the first relevant result. Once they find one or two good results, they look for ‘clues’ (e.g. citations, references, company names, inventor names, etc.), branching out their search if required to find more relevant results. Free from a defined methodology, they can serendipitously change course on a case-by-case basis.
What’s the plan?
Planners leave nothing to chance. They follow a defined search methodology. For example: they will do much research before meticulously defining the universe of results for their search. (Let us call this the ‘big query’, and assume it contains most of the relevant documents; the big query will result in high recall but low precision). Then they follow a structured approach to narrow down the number of documents in the universe to the ‘affordable level’ at which the Planner will be capable of analyzing all the documents. They then analyze each document to identify the relevant ones. They may also look for ‘clues’ later to find additional results, but only after analyzing ALL the documents. For Planners, it’s critical that no possible result is missed. They follow a checklist of steps for each search type – often an identical process for each search.
Planners put extra effort into analyzing higher numbers of documents than the Explorers. They are willing to invest resources into going through a large number of documents, but they won’t compromise on recall.
The pros and cons of each approach
So which type of searcher is better? The table below summarizes our discussion so far and can help us answer this question:
| Explorer | Planner | |
| Where do the searchers search? | Hotspots within Universe of results (or big query) | ‘Affordable’ document set within big query |
| Recall | Potentially Medium – Low | Relatively higher |
| Speed of search* | Fast | Relatively slower |
| Effort investment | Low | Relatively higher |
*time to complete the search project
Risk and resources make all the difference. I’ve realized over the years that in some use cases, Explorers work better, and in others, Planners do. Specifically, it depends on the search recall level required and the available budget (money or time). Where risks are high, and greater recall is required, I recommend using Planners; however, for searches where quick results are required, Explorers are better suited.
Assume that we are working on a novelty search, where the client wants to find out whether his invention is novel or not. A single result already may give the client what they need – the answer to the question ‘is this invention novel or not’. So, in this case recall isn’t important. My searcher can stop the search as soon as he/she finds a single result that precedes the client’s invention. Even if we miss ten or twenty similar results, it doesn’t matter as we’ve already answered the client’s question. In such a case, my Explorer is the right choice as they will work on a novelty search for 4-5 hours and find me one preceding-invention result and then move onto another project, as opposed to a Planner spending 12-15 hours – with a more methodological approach – and finding 4-5 good, but unnecessary, additional results.
This changes completely for a Freedom to Operate search. I wouldn’t want an Explorer anywhere near an FTO search! I need 100% recall in an FTO search, as even one result missed can lead to a million-dollar lawsuit for my client.
How we decide
Based on this discussion, the following table helps us decide which type of searcher works best for different IP searches:
| Type of search | Recall required | Explorer | Planner |
| Novelty Search | Low | ✓ | |
| Validity or Opposition search | Medium | ✓ | ✓ |
| State of the Art or Intelligence Studies | Medium | ✓ | ✓ |
| Freedom to Operate study | High | ✓ |
So, my secret is out – this is how I plan staffing for the searches that come into my team. The Explorers provide greater return on investment in novelty searches, and so are better used here, while Planners are a must for FTOs. You will notice that for a few types of searches, either Explorers and Planners can work. We use this insight internally at Evalueserve. Perhaps you read one of Urs’ earlier blogs about parallel searching. For high-stake searches of the two kinds mentioned above, we usually staff two searchers – one Explorer and one Planner – on the same search to get best of both the worlds.
This ensures superior search quality and client gets the best results!
1 For our definition of recall, see our previous post
