FoodBanter.com - View Single Post

at Thu, 29 Dec 2005 23:27:53 GMT in
.com>,
(Chembake) wrote :

>at Wed, 28 Dec 2005 06:12:38 GMT in
s.com>,
(Chembake) wrote :
>>at Tue, 27 Dec 2005 10:41:10 GMT in
ps.com>,
(Chembake) wrote :
>>...
>.
....
>> By contrast even the slickest marketing can rarely salvage an
>>honest-to-goodness dud. ...
>A product that is destined not to succeed is because of the failure of
>the new product developer to assess the product quality that it will
>conform with the customer needs which should have been already
>anticipated .

In theory, that's how it's supposed to work. But in practice, even the
best-researched and thoroughly tested products bomb in the marketplace, and
I've seen plenty of cases where they bombed because of reasons that could
have been caught had the product been subjected to consumer trials. Usually
in those situations it seems obvious after the fact, but this is because
some key factor that had never before been noted becomes visibly self-
evident. No company will make the same mistake again, because now whatever
that mystery factor is will have been determined and integrated into
testing, database, and other systems. However, before the release of the
fatal product, no company knew or perhaps cared about this factor.

>But it seldom happens as its a team effort

I agree that a total failure is rare.

>Besides a failed product is not only something of sensory quality. It
>may fail even if it satisfies many the criteria what the consumer
>wants from that product..€¦.but if the developer expectation or much
>more the company behind his team expect so much for that particular
>product and they have their own projection for its performance and it
>it happens it does not reach the standard of performance

That's part of what I was referring to in describing the assumptions of the
developer. On paper the product may look as if it's going to be a smash hit
while in practice it might turn out to be only a modest success. Companies
inevitably feel a bit deflated when this happens.

>best selling products they will just stop selling , bring it back for
>more study and see if they can improve it further before they can
>relaunch it the same or as a different product name

Another thing that can happen is that the product was great, but its cost
structure just was too high for the actual market they were able to
capture. This is one of the most common causes of customer mystery - it
becomes a "whatever happened to that great xxx product..." when the truth
is the cost they'd factored in assumed a larger market than actually came
to pass.

>>>It sounds like you misinterpreted the analogy,...
>>Nope
>>What, exactly, are you disagreeing with here?
>
>Its the comparison of subjects for ( drug use and confectionery
>consumption evaluation).

Don't be overly distracted by the fact that both of these items are
comestibles. That wasn't really the point. The point was to illustrate
industries where the market research process could justifiably invoke
similar themes.
....

>>If you tried to extract the same level of detail out of the consumers
>>for the same product, you could expect larger variances - i.e. a wider
>>overall statistical distribution of the results, but this would be a
>>reflection of the level of detail sought, not the accuracy of the
>>analysis that can be performed. ... If, on the other hand, you restricted
>>your questions to simple like/dislike response - the sort of level a
>>consumer would probably respond on, then you would get probably equally
>>accurate results ...
>
>Yes and no responses, like and dislike €¦. Its not just not accurate
>enough to describe the attributes of the food product.

I would say that it's not a matter of *accuracy*, it's a matter of
*resolution* - how many separate features are you going to try to extract?
With any statistical data set taken with any sample, the resolution of your
data is going to be inversely proportional to the accuracy - so the broader
the conclusion you want to draw, the more accurate, in the sense of being a
good predictor, you can expect your results to be. OTOH sometimes perfectly
accurate results in a very broad classification don't tell you very much.
Depending on the sample taken you can manage different levels of tradeoff
between accuracy and resolution - in the case of the test panel you're
getting excellent resolution at the sacrifice of some accuracy, while in
the case of the consumer group you get good accuracy but lose resolution.
To a certain extent using accuracy and resolution is quibbling over
terminology, but the underlying property - of a tradeoff between 2
different desirable features of the analysis - is what I want to emphasize.

>Yes the result can also be statistically evaluated but it will never be
>used as the major factor that the product fits the expectation of the
>customer.

Why not? If the consumer likes it, we may assume fairly well that it fits
their expectation. The direct evidence of actual response is more solid
than the indirect inference that you could draw based upon an idea that if
a product fell within a certain profile it could be expected to be well-
received.

>There is a wide variety of confectionery products and even me I dont
>like many of them; so are the customers; there is a certain target
>client for a certain confectionery item and that will be expected to
>patronize them if all their needs for that certain item is filled up.
>
>Therefore Target market is the keyword here
>Every food designer have it in their mind before they embark on such
>particular food product development
>So when customer who is interested on that particular item will buy
>the product they are optimistic that they would appreciate it.
>Only people who has no affection for that certain confectionery line
>will dislike it..

Yes, you have hit the nail on the head when you identify the key issue as
target market. Third-party analysis can be much more accurate when
you have a narrow target market, because in that case there's been some
attempt to pre-qualify your audience. Nonetheless, I think in this case a
consumer testing round is valid - you just need to screen your consumers
who are going to participate. Clearly it's futile to survey a broad
spectrum of consumers for a product that's only going to appeal to a
certain clientele - e.g. your licorice example. What's done is that with a
quick round of preliminary questions taken from a broad canvassing, you can
qualify your target consumers - who then form the basis of your consumer
test panel. This is commonly called a "focus group", and although focus
groups, like anything else, are only part of the marketing picture they
have been remarkably successful when used intelligently. I can see how you
might have thought what I was advocating would be foolish if you thought I
meant that one should just randomly pick consumers out of a crowd.

....
>As most consumer panel are just randomly selected how can the
>evaluators see a reliability that they have amassed the right target
>customer for that particular product line?
>A lot of consumer panel loves freebies.... and they have nothing to
>lose but something to gain.
>They may not like the product but out gratitude for the freebies and
>compensation for their time and effort they will gladly lie in the
>sensory evaluation to please the leader of the consumer panel
>evaluation team.

When a company devises their consumer panels like this they have only
themselves to blame for their own poor results. Such a panel will usually
have been assembled to justify a pre-formed conclusion, exactly the
behaviour I was warning against. Yes, a blind consumer panel conducted as
you outline is useless. That's why you don't conduct consumer panels like
that. But if you imagine that all consumer panels must of necessity be this
way then you are missing out on an important market-evaluation tool.

>ïƒ˜ Again, most people don't follow an algorithm when making a
>>purchase decision.
>
>That justifies the reasoning that its not wise to trust the customers
>judgement as they are capricious . and most of the time unreliable.

IMHO it justifies rather the reasoning that it is not wise to treat
marketing as an algorithmic, deterministic process that you can just follow
procedurally. People are somewhat unpredictable and thus an attempt to
reduce things down to a rigidly determined outcome will inevitably result
in the occasional perplexed surprise when things don't go according to
plan.

>.
>>I would emphasize that the trained panel see the elephant in many
>>angles so its more exacting than feeling only a part of it.
>>Except that it is the customers who are buying the product. So if the
>>goal
>i>s to sell an elephant, then you have to appeal to the blind customer,
>not
>>to the sighted panel. The panel may be able to expound on qualities
>>that would be important to those who can see, but to the blind man such
>>qualities might well be immaterial. Yes, the panel has a richer
>>description, but all that richness of description means little when the
>>buying decision is being made by somebody with a more impoverished
>>perspective.
>
>This sums up that if a certain consumer only see a part of the whole
>picture then how can their decision be taken into account as reliable
>basis that the product is good or bad?

Because it doesn't matter what *you* think. What matters is what the
*customer* thinks. This is true to the extent that even if the customer
were blind and attempting to buy an elephant, and the sighted experts could
aver that what he was feeling was, indeed, an elephant, if that customer
were to think that he was feeling a giraffe, then from the POV of the sale
it would be a giraffe. That's what's meant by the aphorism "the customer is
always right". People tend to take that statement as a policy directive for
customer service, when in fact the meaning is much deeper - it means that a
company must always follow what the customer says he wants, even if from
the company's POV that seems absurd. It's futile to try to second-guess the
customer.

....
>i>t's the customer who has the more holistic viewpoint, refusing to see
>an
>>object as a collection of distinct subobjects.
>That is what I mean€¦.to see things as a whole€¦.but if you judge it
>by the term or like and dislike (which is common in consumer panel )
>which are half truths ...it does not say anything to be taken
>seriously by a competent evaluator as it does not say anything valid(
>if not solid) descriptors that can be used to relate to the
>technically trained panel.

To the technically trained panel what the customer says may be impossibly
vague and meaningless, but the important point is - the trained panel is
not the group who is going to be *buying* the product. So what they think
in terms of what meaning the customer's description has is totally
irrelevant. To give another example: in the film industry it's common for a
film to get great reviews by the critics, who we may assume to have
excellent knowledge. But then it bombs at the box-office, often because
what the critics saw in the film was too obscure and/or inaccessible to
make any sense to the viewing public. Meanwhile, all the critics could
roundly pan a movie which then is a blockbuster, because even though it
contains nothing that the critics see as commendable, it has appeal to the
common man. I would argue that it is the critics who have the wrong
perspective in these cases, not the audience. From the POV of the film's
producers critical acclaim is only valuable insofar as it increases box-
office returns, and likewise a high gross more or less negates any issues
of poor review. If the audience likes it, I feel, the film should be
considered "good" regardless of what the critics say, and it's in fact the
critics who need to adjust their criteria of excellence based on the
popular response - at least insofar as their aim is to provide a service
that indicates to the readership what films they should see if they wish to
be entertained.

>>Unknown products?....how can that be a confectionery product is
>>supposed to be known and its not as dangerous or risky if compared to
>>drug use
>>It's not that the reaction might be dangerous, it's that it's
>>unpredictable, precisely because, as you point out, customers are
>>notoriously subjective. ...
>
>That is one of the major reason that I dont want to compare drug
>evaluation to confectionery assessment . They are different : a food
>item is never comparable to a medicinal product..
>It does not give any sense or even logic at all for an equivalent
>comparison.

No analogy can be "perfect", retaining all the properties of the thing
analogised, for if it did, it would be that very object. The only way one
can, indeed, distinguish separate objects is that they have different
properties. Thus when making an analogy it is necessary to restrict one's
assessment of the similarities to that domain where the analogy is presumed
to apply - the overlap of similar qualities - and not get caught up in how
things are different from one another. Otherwise you'd find there were no
good analogies for anything.

....
>If your statistical thinking is correct and just focusing on the taste
>aspect alone, then what have been found in the laboratory and pilot
>scale studies including sensory analysis already produced positive
>result that the majority of the attributes as what the customer
>wants ( for that certain product)then why would the consumer will be
>expected to say grossly the opposite that is not what they want?
>Where is the positive correlation statistically speaking?

It could be different because, as we have both pointed out exhaustively,
the average person doesn't deconstruct a taste into its respective
components. Rather, he sees it as a total object that he "likes" or
"dislikes" subjectively. If you break down a taste into components you
assume that these components are in one sense or another independent - or
at least that you can reduce the taste to a minimal set of independent
attributes that can be considered the "dimensions" of the taste as far as
your analysis is concerned. Unfortunately, *real* taste tends to have
inseparable variables: it's a case of "everything depends on everything".
This kind of problem foils a database because DB's are designed on the
relational model that assumes a 1-to-many hierarchichal relationship - that
you can break down your attributes into that set of independent variables.
Any time you've got the kind of many-to-many mapping that characterises
real taste, your DB will go haywire.

In actual fact, it's not quite a hopeless picture, because in spite of the
fact that taste is so intertwined, you can approximate the picture with a
series of more-or-less independent variables which give at least a
reasonably complete representation of the taste. For many situations this
gives good results, but since it's an approximation, it can't catch every
case. The analysis might then indicate that such-and-such a combination
will be a success, but in fact it's either not so successful as expected,
or in the worst cases a total failure. Nothing was wrong with the analysis,
it's just that if this result comes as a total shock then the people
involved never really fully grasped that they were dealing with
approximations and not with absolute fact.

>Another thing is
>Why would the developer rely on the outside feedback to dictate them in
>their jobs when they are already aware what the consumer want and they
>are developing the products in that direction?

If they're not receptive to outside feedback, then I think the developers
are running the risk of becoming closed-minded, convinced of their own
knowledge. Part of being a first-class developer is being able to listen to
and heed external input.

>The big decision if the product will fail or succeed in the market does
>not come from the consumers but within the producers ranks; These
>people are not crazy to waste resources without having a forethought if
>that particular product is doomed to fail .They are certain the know
>what the customer wants and they are going in that direction.

Being absolutely certain you know what someone else wants sets you up for
the biggest possible embarrassment when you discover that what you thought
they wanted was not what they actually did want. Sometimes you can predict
effectively, sometimes not. Better not to assume that you know, but rather
to believe that you have some ability to predict.
>Therefore this validates my earlier statement that the consumer panel
>is just SUPPORTIVE or CONFIRMATORY in any confectionery related
>development.

Once a consumer panel has been set up with the assumption that you know
exactly how they will respond and are merely trying to verify your
knowledge, you have committed the error I described above as "Such a panel
will usually have been assembled to justify a pre-formed conclusion".

>>Except that in the specific case of lecithin a customer has some reason
>>to believe the confection *might* be fatal and no hard data to assure
>>him that it won't. It's for that reason that some people in the
>>soy-allergic group are scrupulous about avoiding soy lecithin.
>
>That is dubious thought....lecithin to be fatal....when similar lipids
>exist in the human body?

It's not the lecithin in particular that could be fatal, it's the
derivation from soy, which for those truly allergic might be a cause for
concern, which could be fatal.

>And it was proven time and time again that is safe...regardless if
>comes from soybean or other plant material.
>If there is somebody who is really allergic to it is very rare and not
>a cause of concern for the confectionery manufacturer.

No, the confectionery manufacturer can't worry about that sort of thing
explicitly, because then you are desiging for an extreme exception.
However, in order to avoid possible legal entanglements, you may need in
today's litigous society to print a disclaimer on your label. Meanwhile if
a confectioner did choose to avoid lecithin it might be because he had
other objections and thus being able to assuage the concerns of the soy-
allergic would merely be a small bonus.

--
Alex Rast

(remove d., .7, not, and .NOSPAM to reply)