If you are reading this articlein the paper version of Tech Decisions, rather than online, youknow that you received a book in addition to the regular magazinethis month. I hope and trust that you opened the book, read if fromcover to cover in one sitting, and are now basking in theintellectual glow of the insights and humor you just consumed.

The book of course is “Shop Talk,” which was recently publishedin conjunction with and its parent company, Summit BusinessMedia. The contents of the book are five-plus years worth ofarticles from this column grouped by subject area.

The production costs of the book, which were substantial, werepaid for by the following sponsors, to whom I am very grateful:CSC, Guidewire, Insurity, ISCS, MajescoMastek, Maple Technologies,Millbrook, OneShield and Unirisx. I am, of course, proud andpleased that these articles have been published in book form. Forover five years I have tried to put useful content and practicaladvice into these articles with the intent of giving back to theindustry that has provided me with a professional home for most ofmy career.

In working on the book I re-read many of these articles for thefirst time in years. During this review it struck me time and againboth how much and how little has changed over the past five years.The lack of change is partly due to the subject matter that “ShopTalk” focuses on—core insurance system legacy modernization usingthird party vendor products.

This isn't a fast-emerging field like mobile technology orsocial media. The Shop Talk domain deals with large, complexsoftware decisions and lengthy, very complex implementationprojects. The vendor market changes slowly, the software takes tensif not hundreds of person-years to build and the implementationstake multiple calendar years to deploy.

The subject area also deals with irreducibly difficult projectswhich are hard to do well: software selection and legacyreplacement. By their very nature, these are projects thatinsurance carriers do once every 15 to 20 years, so it's not likethey get better with practice.

Five years ago there were too many software vendors in theinsurance vertical. Core systems projects failed with alarmingregularity. Carriers didn't know how to select vendors. Five yearslater…could you say the same?

The Vendor Market

Five years ago there were too many vendors in the market. Thereare more now than there were then. Total vendor count in thep&c policy administration category remains above 50. There hasbeen merger and acquisition activity, but it hasn't reduced thenumber of vendors overall.

Changes that happened include: Duck Creek wasacquired by Accenture; Guidewire entered the policy space and thenwent public. Changes that didn't happen include Insurity becameChoicePoint, became LexisNexis and then became…Insurity. Claimsvendor BlueWave became part of Coverall Systems. New vendors emergelike CodeObjects and Maple Technologies.

The new breed of highly configurable vendors gained dominance inthe market while the large multi-market players continued tostruggle for relevance. One fact that was true five years ago andremains true today is that the multi-billion dollar, multi-marketsoftware behemoths remain either incapable or uninterested inpenetrating anything other than the top tier of insurance carriers.From the inside out, insurance is complex, fractured andidiosyncratic. From the outside in, it's just another part offinancial services. Vendors that hold this view will never make itin insurance. Large vendors are faced with three choices: quit themarket, status quo, or acquire expertise and market share.Accenture followed the third option.

More important than market gyrations and branding changes,vendors write better software than they did five years ago. The barhas been raised permanently as to what is an acceptable level offunctionality, configurability and architectural quality.

This positive trend certainly started more than five years agowith the arrival in the insurance vertical of professional softwaredevelopers who learned how to build insurance applications, asopposed to the legacy vendors who were largely insurance people wholearned how to build software. However, the trend has gainedmomentum and gone mainstream in recent years.

The days when vendors could sell stinky software on legacyplatforms is gone. Carriers may not be as good at selectingsoftware as we might hope, but they are good enough that vendorshave had to improve or face irrelevance. I am pleased these days tobe able to introduce client carrier companies, almost regardless ofsize, to a short list of very high-quality vendor solutions. Thiswas not always possible in the past.

Another positive trend is that carriers now have more choices asto how to contract and pay for software. Many vendors now offerboth purchase and rental options for their solutions, as well asusage-based pricing. A new generation of vendors is pushing (orpulling) the market into a cloud-based software environment whereimplementations are quicker and the total cost of ownership islower.

These vendors are, I believe, the next generation, and arestriving to wrest market dominance from the highly configurablevendors that have owned the market for the past five years. Thesenew vendors, it should be noted, also have highly configurablesolutions, on which they are building new delivery models.

The Carriers

So, the vendors have generally gotten better at building andselling software solutions, but have the carriers gotten any betterat acquiring solutions? The answer is yes, but a qualified yes.Carriers are better informed today than they were and are moreaware of their own limitations than they used to be.

What I used to tell carriers with reference to softwareselection: “This is complicated, you don't know how to do it, youhaven't done it for 20 years,” they now tell me. This is majorprogress.

Many carriers now use third party consulting services to leadthem in software searches; however it has been a personalexperience of late that this has become a reductionist situation.Now, instead of carriers not knowing how to select softwarevendors, they don't know how to select consultants to help themselect software vendors.

I am intimately and painfully aware of situations where carriershave taken six months to choose a consultant. This “reductionistproblem” aside, carriers generally make more informed decisionsabout software acquisitions than they did five years ago. Anecdotalas it may be, those occasions on which I hear of selectiondecisions that make me think “they chose who?” seem less and lessfrequent. Others in the industry concur.

Not only are carriers choosing appropriate solutions they arealso getting better at structuring appropriate business deals andcontracts for software and services. This is partly a reflection ofthe willingness of vendors to share risk and get paid forperformance, but this happy circumstance didn't arise solely fromthe goodness of the vendor community's heart. Carriers' began topay more attention to risk and cost, and vendors responded inkind.

Legacy Modernization Projects

Five years ago it was “legacy replacement,” now its “legacymodernization.” The name change didn't do a lot of good. As wenoted above, vendors now build better software and carriers arebetter at selecting vendor solutions, but this has not translatedinto an obviously improved implementation track record.

It would seem that core system replacementsstill fail as often as they succeed, and for those that succeed,“success” comes on a sliding scale. One thing that surely has notchanged over the past five years is that there are no industrystatistics on this subject. This is all insider guesswork andanecdotal evidence…but that doesn't mean it's not correct.

So why the lack of improvement in this third key area given thestated improvement in the first two? My guess is as follows: Coresystem legacy modernization (replacement) is complicated and hardto do. By definition, carriers do this once every 15 to 20 years sothey are not prepared for this once (or maybe twice) in a careerevent.

Many carriers, especially smaller ones, lack the ITinfrastructure—strong project management, formal requirementsgathering, and rigorous QA—needed for such an extensiveundertaking. Further many carriers lack the funding to employ thirdparty expertise to shore up these weaknesses, assuming they evenrecognize their existence.

The response from many carriers is to look to the softwarevendor for the expertise they lack internally. However, while thevendors have improved at software development, they have notnecessarily gotten any better at software implementation. Further,it's not the vendor's job to run client implementations.

Vendors do the parts of the project that relate toimplementation, configuration, and integration of the software.Carriers, not vendors, are responsible for requirements definition,software testing and acceptance, legacy integration, rollout andtraining, conversion and legacy retirement.

These are areas about which the carrier is expert, not thevendor. If the carrier fails significantly in one or more of theseproject domains, the project will probably fail. Not that it isalways the carrier's fault.

So there is still major room for improvement. Legacymodernization will continue apace. Many carriers have yet to makethe journey, and there are signs that the journey is less hazardousthan it was five years ago. It will be interesting to see what thereport card looks like in another five years.