How much of your IT infrastructure and operations is outsourced to an external provider?


Great Question

But, plan a follow-up.

What's the plan for three, or five, years from now?

wildly optimistic and self-deluded

The road to the vendors' wallets is paved with good intentions.

How many times have you heard the demise of the mainframe forecast? Sure its gradually happening... over 20-30 years not 2-3. I pay next to no attention to what IT people say they're gonna do: they're generally wildly optimistic and self-deluded.

More outsourcing in Europe

It is quite unusual to find a company in Finland which has not outsourced a major part of infrastructure. A configuration might be this. Fujitsu runs desktops and SD. Logica takes care of Servers, Tieto manages applications and Elisa telecommunication. There is a lot of variation in the model but the traditional inhouse IT has nearly vanished. Outsourced SD's are quite common.

I believe that it is a mistake to outsource SD's and it is based on a popular misconception of the SD role.


wanted the service desk business

the respondents so far must be non-European.

yes I recently totally outsourced a client to five or more providers. You gotta believe they wanted the service desk business but that's the only bit I told them to keep in house.

Wait... um... I thought service desks were gone away?

Not to go away but change

SPOC is history but not Service Desk.

Here is a link to the presentation James and I did on Brighttalk:



Repurposed might be a better way of thinking of SD 2.0 , and the drivers for that aren't just social media, in fact the prime driver is probably a maturing approach to self service in various forms. I could see level 1 support splitting up into two sub levels, one very people focused dealing with queries that are too complex or novel for self service support, and the other servicing the aspects of the self service respects that require manual intervention. It is possibly simplistic to envisage that in terms of an on-shore/off-shore division of labour.

James Finister

Some of it… but...

We have carved up our infrastructure team pretty well at this point, with all the non-core competencies and most of the level 1/2 stuff either outsourced or offshored to captives. For the outsourced components, it's a best-of-breed approach; a cabling company handles the structured cabling, a maintenance company handles the hands-on in the data center, a carrier handles the network, a help desk company handles the help desk, etc.

Vendor management can be painful sometimes, and you do have to deal with pointed fingers, but you get experts in their domain for a reduced cost (and the flexibility of being able to grow/shrink with the current economics).

CIO Magazines - State of CIO Survey

So, I am in a stats class and just last week we had to go out and find a survey that has run at least 2 different times that asked the same question.

I found the CIO Magazine State of the CIO Survey and used the 2010 and 2011 versions.

I actually picked the question "What percent of your organization’s IT services are currently provided by 3rd parties including outsourcers and could service providers?"

The point of the exercise in my Stats class was to do a two proportion z-test and the results. The H(0) hypothesis is that 2010 results = 2011 results and the H(A) is that they don't statistically match.

Well, I did get full credit so, I assume that means my professor actually read over my work and didn't just give me credit for turning it in. The results I have are that there is statistically significant evidence that CIO's are increasingly outsourcing IT services. The question in particular I was using was the "0%" - so how many had 0% outsourced (so really, that went "down"). The P-value I had was 2.6E-4 - or...0.00026...which would say..given H(0) is true (that there is no change in the real proportions of CIOs that outsource 0% of their IT) then the probability that a random poll of CIO's would produce the 2011 results would be 0.026%. The 95% confidence interval I found was there was an increase between 2.7% and 9.2% from 2010 to 2011 in CIO's that outsourced more than 0% of their IT services.

It relates to this poll here anyway. You can of course go look up those surveys yourself and do whatever analysis you'd like on them.

the actual percentage outsourced

Hi Stephen

I'm not sure why you measured the not-0% population. I don't know anyone with 0% outsourced. In this country I can count on one hand the companies doing their own telecoms.

I thought >50% was a more meaningful measure.

And although increases are very exciting for those in the industry, most of us want to know the actual proportion.

For those busting to hear what the ACTUAL figures were:

In 2010 the percentage with more than 50% outsourced: 12%
Company size distribution is as follows: Small companies: 27 percent; midsize companies: 37 percent; large companies: 27 percent; government or nonprofit: 5 percent (3 percent of respondents did not provide an answer).

A broad range of industries is represented, including government and education (19 percent); manufacturing (16 percent); financial services (14 percent); high tech, telecom and utilities (11 percent); legal, consulting and real estate services (11 percent); healthcare (10 percent); and retail, wholesale and distribution (9 percent).

Fifty-six percent of respondents are from North America.

I can't find the 2011 survey yet, despite giving away my identity (sorry Chris) on a misleadingly-titled link that led me back to 2010.

Well, I was a bit lazy

I could have added categories together and then compared those proportions - but I honestly I didn't know how my professor would react to me taking some liberties with the data and was just as easy to not concern myself with any rounding issues and how that may impact the results.

All that being said, I have the two surveys if you want I can email them to you - you can email me at stephen.alexander AT gmail DOT com

The reason I picked 0% is that it was the easiest to show that there is a trend to outsourcing given the buckets available, without combing any of the buckets myself to create something like ">50%".

I can do the math though and share the results with you if you'd like. It shouldn't be that difficult. Just know the proportions will be maybe a little off since I don't have the actual data (just %) so I don't know the rounding - perhaps it could be off by 1 or 2% (after adding them together) - which may not be all that much, but it could be enough to make it statistically significant (or not). Once I run the numbers we can see how close it is to say..0.05 or .025. The further away, the more confident we can be that there is a significant trend even with the fudge factor we'd have to deal with.



A quick count

I have here that in 2011 it was 12% had over 50% of the IT was outsourced and in 2010 it was 13% had over 50% of the IT outsourced.

Be careful, the 2011 title page says "2010" but if you read it, it is actually the 2011 survey.

Anyway, without doing the math, it seems quite unlikely that 1% change is anything significant at all with these surveys (the population sizes are 550 or so and 700 or so - not nearly big enough that a 1% delta would mean too much). I highly doubt a 2 proportion z test will reject H(0) in this case. I suppose I could be wrong though.

For 2010 there was 13% that 0% of their IT outsourced and in 2011 it was 7%.

Again, the 95% confidence interval was (about) -9 to -2. Which is to say, based on these two surveys, if we took a random sample of CIO's we should have a 95% confidence that the proportion that have 0% of their IT outsourced would be 4% to 11%.

I don't know how many businesses you know in New Zealand but if it is 1 out of 20 that doesn't outsource anything then it is within the 95% confidence interval. If it is 1 out of 50 or 100 or something...well...maybe there is something going on in New Zealand that isn't happening at quite the same rate in the rest of the world or these surveys are not 'random' enough?

Or maybe those CIO's don't know what they are talking about! (gasp, imagine that)

global rumping

Thanks Stephen

What i'm trying to get at is what proportion of my readership are in environments where most or all of the IT is outsourced, where - as Jim puts it - IT is a "rump".

That's very different to the situation where only some specialist components are outsourced, such as telecoms, network monitoring, procurement, or field support. My point was that worldwide I bet 0% outsourced is quite rare, but that doesn't tell us anything about 80-100% outsourced, which is what I'd call a rump.

I'm assured that rumping is common in Europe. In fact I'm given the impression that it is the case for most European organisations. Readership of this blog is roughly 1/3 North America, 1/3 Europe (with UK over-represented), 1/3 "other". So i'm trying to get a more global picture

50% of what

The catch in trying to come up with a reliable figure is trying to work out a sensible measure. You can't go on just IT spend, because that includes things like hardware which is probably leased or purchased semi independently of the supply of services and won't appear on the outsourcer's books, even though they are managing them on behalf of the customer. Depending on how a company looks at the world outsourced software development can also get hidden under another heading. Is headcount a valid alternative?

The CIO figures are clearly dominated by the US figures, and of course the US has not embraced off-shoring as much as Europe.

The Harvey Nash 2010 CIO survey gives global figures of

- Software application development 62% outsourced
- Application maintenance and infrastructure maintenance both 53%

As far as I can see those percentage of respondent figures, not percentage of budget

The 2009 Oxford Economics survey identified 340k outsourced workers in the UK (that is, they don't include off shore outsourced workers supporting companies in the UK) , A figure from another source gives the total number of people working in IT for non IT companies as 663k. I don't the data is sufficently compatible to make a direct read across though, since many, if not all, of the 340k might be included in the 860k working for IT companies

James Finister

It would be great

To see that survey. I went to the site, and requested to have an electronic copy. What was the sample size? (62% of how many?). Also are those numbers for "fully" outsourced or just partially outsourced? If it is fully that is pretty interesting. The rule for conditional probability for two independent variables (I don't know for certain, but I believe the two variables you listed are independent of each other) is to multiply them - so we can then figure that .62 * .53 = .33 or 33% have both software development and application maintenance/infrastructure maintenance outsourced.

That is certainly bigger than the CIO Magazine would (generally) indicate and is slightly larger than what could be approximated by this poll here (so far - I see 15% all, 9% most = 24% all or most - which I would approximate with having software development outsourced/application maintenance/infrastructure outsourced).

However, if it is only to mean they have some component of software development or maintenance outsourced then we need to first see what is the conditional probability that a company does neither = .38 * .47 = .178 (or 18%), then 1 - .18 = .82 or 82% of companies outsource software development or application/infrastructure maintenance. That number is a bit smaller than what we see in CIO Magazine (which I would take to mean anything above 0% would have some area of IT being outsourced) - so CIO Magazine is saying a full 89% to 96% have 'something' outsourced. But it is close to what we see here (84% of respondents say at least 'something' is being outsourced).

So a quick table:

CIO Magazine: ALL/MOST -- 7-9% SOMETHING -- 89-96% NONE: 4 - 11%
Harvey Nash: ALL/MOST -- 33% SOMETHING -- 82% NONE: 18%
ITSkeptic: ALL/MOST -- 24% SOMETHING -- 84% NONE: 16%

For whatever that is worth. Different people answering the different questions with different interpretations as to what the question means and me grouping them together in a way that may, or may not be appropriate. : - )

I thin the questions are close enough to group though. The numbers seem close in all areas except "All/Most" with CIO Magazine being the oddball.

from the last millennium

Great analysis Stephen, thank-you!

What would be interesting now would be a comparative stat from last millennium to see how much most/all has changed. I suspect it has gone up but not as much as folk might think. Going on memory alone I'd guess say 10% had mostly/fully outsourced to the likes of EDS.

Syndicate content