Imagine you are being asked to give an estimate for the completion of a piece of work. Imagine that this group includes Testers, Product Owners, Programmers and representatives from different departments. Imagine that you are going to have to haggle, negotiate and barter to get the time you think you need to do a good job.
Now imagine that I could give you a little trick that *could* help you get the number of days you actually want.
The above scenario is a common occurrence for many teams. We are often asked to give estimates for work.
Some teams use techniques like the Delphi method, rubbing the goat (don’t worry, it’s safe for work), like for like (same estimate as programmers) and a variety of other techniques. These are all valid and I’ve seen each one work; including the goat.
Now imagine this. As you are in the process of estimating you have the direct ability to dramatically influence what others agree on simply by shouting out a number (or holding up a number in Planning Poker (first round)). This number could be anything, but the more extreme the better (at least for this imaginary situation).
Simply by offering a number you have the ability to create an “anchor”.
I’ve been studying a lot about “anchoring” and how this relates to price and numbers. Anchoring is a bias (cognitive one) that occurs when people place too much emphasis on a piece of information. Michele Smith wrote a good blog post on anchoring and testing here.
So let’s look at this from a generic example point of view.
You are looking to sell something to someone. In this instance the price is not listed. You think this “something” is worth about £2000; that’s what you want to get for it.
So you throw in a crazy high price of £5000. The buyer laughs and nearly chokes. So you smile and say comforting words, then come back with a price of £3000.
According to much of the research in anchoring the above scenario would likely end in you selling this “something” for a higher price than if you had opened the pricing at £2500.
According to studies on pricing you have thrown in the anchor of £5000. It is now acting as a reference point in the buyers mind. Your offer to sell for £3000 now seems like a real bargain compared to the £5000. The same applies when buying. Go in low and then move up from there. This offer that no-one is ever likely to accept is curiously known as a “dead dog”.
When I read more around price anchoring I see this in everyday life.
You’ve surely seen the adverts: “Was £100 – now only £25” or “Reduced for a short time only from £100 to £25”. In these instances the £100 is an anchor. The fact that the product might only be worth £25 anyway is why anchoring is used in the first place.
Another technique is to offer your products (with slight variants in deals) for different prices. One incredibly high price which no-one would likely buy and a lower price which is really what you want to sell your services/product for. In this example the research suggests that you will sell more of your “lower price” service because people are using the high price as an anchor. High end designer shops go mad with this technique.
Every day you are exposed to anchors in pricing.
So what has this got to do with Testing?
A lot really.
Going back to the example earlier, let’s say you want to get an estimate of 10 days to be accepted. So you throw in an estimate of 25 days and give a list of reasons why this might be a good estimate.
Lots of smirking, muttering and breathing in air through teeth. You sense the discomfort.
So you now go for an estimate of 14 days and give some reasons why you could achieve this and why this might still work for you. According to the anchoring theory it seems likely that this estimate could very well be accepted. In comparison to 25 it’s a good deal.
In a sense this is the interesting side of Planning Poker. The first round is intended to be called at the same time, for the very reason that anchoring plays a large influencing effect in estimation. Yet the first round can often throw in curve balls. Three people give 5 points, one person gives 1 point and another person gives 10 points. The second round often has people gravitating from their original number to either the higher or lower; a lot of this gravitation depends on the negotiating skills of the outliers.
A counter to anchoring is something called an Adjustment Heuristic which is essentially a mechanism we use to adjust to the anchor to get closer to our estimates/outcomes.
Anchoring is complex, complicated and very intriguing. I haven’t done it justice here, but it’s an area that I believe warrants a lot more reading about (and experiments). Over the next few months I’ll hopefully share some of what I am learning with you and no doubt fine tune my own thoughts on how anchoring plays a part in Testing.
If you’re interested then here are some resources on anchoring: http://en.wikipedia.org/wiki/Anchoring
Photo by David Feltkamp on Flickr.