Project : some thoughts

Update:

OK, I see Shri Sreenath has ridden to my rescue again and closed the time window for Phase 3 revals.

Am receiving mails from folks saying "could you release the best 1 or 2 projects so that we can see and learn how the project should have been done", types.

Let me clarify - that it would be a stretch to impute normative traits to the highest scoring projects. The highest scoring projects were simply the ones that were the most diligent in adhering to project guidelines, structure and requirements the most.

It in now way means that the highest scoring project teams worked harder or came up with clearly superior solutions to the client problem. It just means that their solutions were logically consistent and well-executed on the PPT.

Also, because nobody exceeded 13 marks, I have allotted upto about 1 point (between 0.5 and 1) to the teams that worked hardest in securing the largest number of phase II responses. The final scores you see actually reflects this additional score. And the 2 highest scoring teams had both, not coincidentally, also done well in phase II.

Original post Continues below:

Chalo,

Shall soon be done with the finalization of phase III marks and then with the grades.

Some more thoughts on the project execution:

1. It is vital to think through the analysis process before actually doing anything.

Some groups had quite clearly gotten the thinking-through messed up. Hence the specter where the target group (prospective buyers, as defined by the gateway question, #7 was it?) was not filtered in. Result - the analysis now contains a lot of noise throwing doubt on result validity.

Had some groups argue that since we have the info anyway, why not use it; the respondents filtered out because they aren't looking to buy in the next 12 months might still buy a car in the next 2 or 5 yrs maybe, so their input should be considered, etc. Well, maybe, maybe not. POint is (and this learning carries through to industry projects you'll face at work), you have to draw a line and stick to project boundaries - else rationalizations (which are part of human nature, btw, nothing wrong with them per se) can lead you onto a never-ending runaround.

2. After the thinking-through is done, ensure the execution is meticulous.

Some groups had quite apparently thought through the analysis and knew what they wanted to do. But they seemed to have lost the plot in the execution phase. Can recall at least 2 groups that went neck deep into analysis, graphs and all but in the end failed to integrate the analyses into a coherent set of recommendations.

The point of the analysis was to select an option (which submarket) based on well-defined consumer segments (which in turn arose from attribute preferences among other criteria).

3. Ensure slide communication is complete and self-explanatory - don't assume the reader will know what you are talking about.

Quite a few groups claimed they meant something else than what was written on the slides. Such an argument cannot stand because the slides are all the space there was to tell the story. What didn't find its way onto the slides could not be considered. Besides, aajkal, in industry also, many projects ask for deliverables in PPT terms, not in word doc reports. PPT communication is a useful skill to master.

4. The market assessment task was a washout.

Market assessment, the way the consultants do it - should have merited a half-lecture at least, admittedly. It will surely find place next year.

Mkt assessment is a very logical and seemingly common-sensical sequential process of deducing the market size or potential mkt size using a series of extrapolations and assumptions on secondary, known data.

So the potential demand for the proposed car could be --> (Total # households in india) * (% in the target income group) * (% in the target age range and demographics) * (% in urbans and metros) * (% already own a car)* (% willing to pay in teh target price range)*(% looking to buy in the next year)* (etc etc*) types.

*Just indicative but you get the idea.

You were asked to do a mkt assessment of the consumer segment you finally selected (after the cluster analysis stage) - by projecting sample share onto the target population share. Very few groups attempted this at all.

5. Be diligent when catering to client requests that have been explicitly made.

In this case, I had quite clearly asked for a data section (half slide worth perhaps) for each analysis method that clearly specified the dimensions of the input matrices to each analysis procedure.

NOBODY did this or bothered to write this down.

Now, the data section would have immediately helped place what went into the analysis proc, so I would have a good idea of what came out.

That did not happen.

There was little to show what went into factor analysis, for example. Was it filtered or not? What was the coding scheme like - were missing values replaced by mean or deleted? Basic things like this. Not even the appendices contained a data section. Not good!

Folks, the client is no fool and woe befalls those who make such an assumption. Typically, in consulting projects(aajkal at least) clients have their own panel of experts go over the recommendations the consults dish out with a fine toothcomb to ensure the consults have been rigorous and proper. Nowadays, clients also playoff one consult firm against another by comparing bits and pieces of the same project spread across the two. It really pays to be diligent in such matters.

Added later:
6. Let the data speak.

Many groups had decided on submarket a priori and then proceeded to do factor and cluster analysis on a chosen submarket. I can't say the approach is wrong but I wasn't too happy with the approach to start with. I would rather the attribute prefs decide segmentation- then targetting - then positioning (under the classic STP framework) simply because it lets data speak quite freely. The submarket preferences of respondents could conceivably have beena ccommodated as an additional clustering variable, perhaps.

Anyway, heated debates followed on why the first-submkt then factor-cluster approach is better with a lot of groups. Let me clarify that if you did factor cluster outside the classic STP framework, it is not necessarily wrong, just that the input logivcal flow is quite a little skewy now and that is why you got partial marks for the 'factor/cluster input quality' component. The 'factor/cluster interpretation' grade component is quite independent of the input quality, though.

OK, so the grading in phase III was deliberately kept slightly non-easy because of a variety of reasons. Perhaps the other grade components save phase 3 were quite easygoing and could do with some balancing. In any case, the class had ample and repeated notice about the tough-grading prospect in phase 3.

Chalo bhai, signing off for now. Sayonara.