14 things I have learnt moving to Microsoft Exchange Online
I’m working on a project to move mail users to Exchange Online, part of the Office 365 suite of services. So far the project has 150 users with an eventual aim of well over 10,000. Below I have listed my lessons learnt so far. For me these were not immediately obvious from the comprehensive online service descriptions and deployment guides. As such the below list may be of interest if you are considering migration to Exchange Online also.
For background we have a hybrid Exchange Online deployment as we are maintaining both on-premise and cloud Exchange environments. We have single sign-on so that a user’s on-premise Active Directory credentials provide cloud access. The on-premise deployment lets us continue to support features that are not yet supported in the cloud. The hybrid deployment means, amongst other things, that we have:
- a single shared SMTP domain between on-premise and cloud users,
- a single shared address book between on-premise and cloud,
- free/busy and calendar sharing between on-premise and cloud.
So here is what I have learnt in no particular order:
- Single sign-in – With single sign-on users still need to provide credentials to Outlook the first time after migration and whenever their Windows password is changed. In addition the username has to be in the format firstname.lastname@domain.com. This can cause user confusion (and thus support requests), no matter how much communication you do, if users have been taught to log into Windows as firstname.lastname.
- Public folders – While this is clearly stated as not supported in the Exchange Online services description you need to remember that Organisational Forms also use Public Folders. Organisational Forms may exist for anything from room booking systems to email templates to allow logging of employee leave requests. You will need a migration plan for Organisational Forms.
- FOPE – If you already use FOPE for spam protection on-premise then there will be work to merge your on-premise FOPE and cloud FOPE environments. It’s not a big issue but it needs to be done and Microsoft will need to do it for you.
- Active Directory remediation – Before you sync your on-premise directory to the cloud you need to make sure your directory is ready. The Microsoft provided tool will guide you on what is required. Plan for this early as more remediation may be needed than you think. Also note that this will be a continual exercise as there is no way to prevent non-compliant entries being added to AD.
- S/MIME mail encryption – Exchange Online clearly provides support for transport of encrypted messages. However Exchange Online will not allow storage of public keys thus you have a problem if your on-premise environment relied on Active Directory for public key storage. We resolved this by going back to basics on encrypted key exchange. Users need to exchange signed emails and use the signed email to save their counterparts contact information to their local address book. This results in an exchange of public keys using the local address book as the store. This enables the users in the cloud to send/ receive encrypted emails to their on-premise counterparts and vice versa. This is sustainable on an ad hoc basis but obviously not for a user with a significant requirement for encrypted emails.
- Directory Sync – With a large directory it can take some time to do the initial sync. In our case it took close to 1 week with the sync rate varying between 300 and 15,000 items per hour. With over 400,000 items that takes time, so my advice is allow time for the sync!
- Autodiscover – We have a multiple exchange server environment on-premise. So that Outlook knows to configure against Exchange Online for cloud users we needed Autodiscover to be configured. Not a big issue, it just needs to be done. If you currently use Outlook PRF files then these will need to be reviewed as part of this work.
- Infrastructure required – Obviously by moving to the cloud the idea is to reduce on-premise infrastructure. In a hybrid deployment you will however need to add some servers to your on-premise environment. You will need a server to sync directories between on-premise and cloud (DirSync), a server for Active Directory Federation (ADFS) for single sign-on, a server to proxy ADFS, an Exchange “co-existance” 2010 server to enable calendar free/busy lookup and depending on your data centre environment a proxy for the Exchange 2010 server. If you need highly available then double the server counts (except DirSync which does not have a HA option). Some of these server roles can be virtual.
- Outlook mode – If users have Outlook in online mode they may experience performance issues depending on the delay to the hosting centre for Exchange Online. We resolved this by changing the users to cached mode and setting download full message rather than download headers to mask download delays.
- Shared mailboxes – Mailboxes shared between on-premise and cloud users are not supported. Note that if you move a user to the cloud who has shared their mailbox then the impact may actually be seen by the users who remain on-premise (very slow Outlook performance and periods of hanging). In a similar manner if you move a user to the cloud who has access to someone else’s mailbox who remains on-premise then the cloud user may get poor performance. Interestingly poor performance is not always seen, sometime it works perfectly, either way it is not supported and when the poor performance hits you will know why.
- Scripts required – To efficiently migrate users at volume you will need some PowerShell scripting capability to set locales for non deliverable receipts etc, turn off any functionality you don’t want such as POP3, and to check/ clean-up on-premise AD attributes after migrating users to the cloud (this last item is really important, incorrect check/ cleanup of on-premise attributes can cause all sorts of issues).
- Exchange Online IP addresses – The addresses for Exchange Online are expanded regularly so if you are not using DNS references on your internet facing firewalls then use an “any” IP firewall rule to your ADFS proxy and Exchange “co-existence” servers. You wont be able to keep up with the rapid expansion of IP addresses otherwise.
- Existing email vault – On-premise vaulting functionality, if used, will be lost for users moved to the cloud. Existing vaulted emails should still work – as these are just shortcuts that point to an on-premise archive server. But functions within Outlook will stop – i.e. ability to un-vault individual emails. You may just want to plan to un-vault emails and migrated these to the cloud also.
- It just works – The above items can all be managed. Once up and running Exchange Online just works. And if you never used ActiveSync to mail enable your iPhone and iPad devices then you will wonder why you never had it!
There are still a few things that we are working through; Exchange Unified Messaging to provide voicemail and Blackberry functionality being the top priorities.
If you are also migrating to Exchange Online please add your lessons to the comments thread 
20 checks when reviewing a technology project business case
Most organisations require some form of business case for any new project. Developing a business case can be especially difficult when new technology is involved because accurate and reliable reference information can be challenging to find. While technology vendors can assist this may create a conflict of interest. If you are developing or reviewing a business case for a technology project then the following checks may prove useful.
Check the benefits:
1. Are the benefits measurable? If not, are you confident that the benefits will be delivered? is the business case valid without them? Consider claims such as “increased efficiency” and “increased collaboration”.
2. Have benefits secondary to the technology investment been claimed? If so, have these been identified as secondary benefits? For example, could cost savings delivered by a technology project have been achieved solely through existing price negotiation independent to the technology investment? This is important to understand in context of the business case alternatives covered next.
Check the alternatives:
3. Have alternative solutions to meeting the business case objectives been considered? If not, how can you be confident that better solutions are not readily available?
4. Are business objectives the baseline for considering alternatives? Make sure there is no confusion with technology objectives (which are a means to delivering business objectives but not an end in themselves).
5. Are there ”do nothing”, “replace” and “upgrade” options? Perhaps the cost of ”do nothing” is acceptable? Perhaps an upgrade is too short-sighted? or maybe replacing the existing system is an over investment? Either way conscious consideration of the options should be demonstrated.
6. Is there an option to deliver part of the business objectives? This is relevant if some objectives can be delivered separately to the technology investment. It may also be relevant if specific objectives have a disproportionately high cost attached.
7. Have the options been presented without bias? Is it worth having the alternatives reviewed by someone with a completely neutral view on the approach taken?
Check the assumptions:
8. Are assumptions correctly identified? The assumptions omitted may be more important that those identified. The assumptions for any project should cover the expected benefits, the time it will take to deliver the benefits and finally the associated cost to deliver the benefits.
9. Is the risk to the business case of each assumption stated? Do the assumptions have justified confidence levels and a potential impact if incorrect? If not, how will you determine the risk to the business case if the assumption is wrong?
10. Was it possible to increase the confidence level of the assumption? For example, confidence on existing environment readiness for a technology upgrade may be increased through an environment audit.
11. Was it possible to remove the assumption? Is existing information available to turn the assumption into a fact? For example, assumptions on integration with existing systems can often be proven with test systems.
12. Is there a plan to address the assumptions? All assumptions should have actions identified to validate or remove them. In particular, high risk assumptions will ideally be addressed early in the project life cycle so that potential impact to time and cost can be minimised.
Check the timelines:
13. Is a period of initial stabilisation provided? Technical implementations always require a period of time to resolve functional issues. Is this reflected in the business case timings for when benefits are expected to start accruing?
14. Is a period to allow the solution to scale provided? Once initial teething issues are resolved there will be a phase during which new issues arise as the solutions scales. This will be related to system capacity, system stability and related items such as scaling processes and procedures. Is this reflected in the business case timings for when benefits are expected to start accruing?
Check the budget:
15. Does the budget have an accompanying conceptual design? Can you be confident in an equipment list and labour estimates without a conceptual design, or should a budget contingency be appropriately allocated?
16. Does the budget have an accompanying project plan? Without a high level set of tasks, effort and their interdependency can you be confident that labour costs are accurate?
17. Are the operational costs accounted for? Are the true costs to maintain, operate and manage the life cycle of the technology included? For example, consider the need for feature upgrades, test environments, staff training, maintenance and break/ fix.
18. Is there a budget contingency that recognises risk? For example, a contingency amount if any assumptions in the business case turn out to be incorrect.
Check the financials:
19. Is the payback time reasonable given technological change? Consider how quickly the technology you are investing in will be obsolete or considered ineffective. Also consider the rate of price drops in the industry that will make a good price now look expensive later.
20. When is the project not worth doing? This is important to define so that new information either external or internal to the project can enable validation that continuing the investment is worthwhile.
Many of the above items are standard when reviewing any business case and there are also many other considerations not mentioned such as reviewing the resources selected for the project team. We hope the above provides a useful start.
Negotiate based on cost over a total useful life
When I speak to companies negotiating technology purchases I often observe a focus on the percentage discount on equipment (obviously this only applies to product based purchases). When you take into account the total costs of a solution, over a useful life of say 5 years, it is often apparent that annual services for ongoing management and support represent the largest cost component. Therefore if you really want to save money focus on the annual services and don’t get overly distracted on upfront discounts on equipment. Let me illustrate with an example.
The above chart may provide some insight on why customers often focus on equipment discounts as a primary negotiation objective. The pricing shows equipment as 60%, implementation as 40% and annual services as an additional 20%. Given this breakdown it makes sense to focus on equipment discounts as it represents the majority of the cost.
A quick re-organisation of the original pricing to provide a view over a useful life of say 5 years is also presented with annual services included rather than stated as an addition. Over 5 years the 20% annual services cost from the original proposal is now 100%, or the same, as the equipment and implementation cost. Annual services is now the majority of cost at 50% of the total pricing with equipment and implementation 30% and 20% respectively. Obviously if the supplier provided this pricing the negotiation of annual services would become more significant than equipment.
The lesson learned here is probably quite simple – prioritise your pricing negotiations by considering the total solution useful life. Equipment discounts are an easy measure to discuss with colleagues but don’t let them overly distract you if the key costs are elsewhere.
Technology traits to help you evaluate product maturity
There is readily available information that can assist assessment of a technology’s maturity such as the Gartner Hype Cycle. Information is also readily available that enables comparison between providers of those technologies. Again using Gartner as an example, a view on a providers completeness of vision and ability to execute can be found via their Magic Quadrants. However, objective information on the maturity of individual vendor products significantly less available is especially when many product lines are available for a given technology. The traits identified in the below table will hopefully enable you to make your own evaluation to determine this information.
| Higher Maturity | Lower Maturity | |
| Traits |
|
|
| Advantages |
|
|
| Risks |
|
|
When reviewing the traits it is important to recognise that high product maturity is not necessarily better. For this reason advantages and disadvantages that tend to accompany different maturity levels are also identified. What is most important is having a clear understanding of the maturity of a product you are considering using, being comfortable with the risks, and having confidence that these are outweighed by the advantages.
Eight ways that technology can make your contact centre provide a better customer experience
There are many ways that technology enables contact centres to be more efficient. Common examples include using an Interactive Voice Response (IVR) to ask customers why they are calling, an Automatic Call Distribution (ACD) to distribute calls to agents based on skills or availability and outbound diallers to keep agents busy when inbound calls are low. In general none of these technologies improve the customer experience they may even detract.
So what about the technologies that improve customer service? Well here are a few ideas, with some surprisingly simple:
- Provide direct numbers – IVRs waste time for repeat customers and often annoy new customers. If you provide direct numbers for regular customers and dedicated numbers for new campaigns then you can receive the same efficiency result (customers are filtered by the reason for their call) but you don’t waste their time on an IVRs.
- Personalise the contact – By using the callers identity (telephone number, email address etc) and integration to your customer management system your agents should know the difference between new customers and existing customers (obviously there are exceptions). Rather than using this information to annoy customers by executing a captive sales pitch an alternative is to follow up on previous transactions. This can be enabled by automatic prompts highlighting relevant historical information. Use technology to delight customers and then they will want to give you business.
- Don’t force customers to wait – Every business has peak inbound inquiries. Give customers the option to leave their number. Even better integrate this with an outbound dialler and let their message move through the queue and then automatically initiate a call back. Or best of all just let customers enter their number on your website and automatically initiate a call back as the next agent becomes available. See some examples at Fonolo and LucyPhone.
- If they must wait then value their time – A customer in a call queue who is given updates on the expected wait time will at least have an expectation on the delay. And if they chose to wait don’t have canned music use the opportunity for smart marketing – see On Hold Company for some examples.
- Prioritise important customer calls – Some calls are more important than others. You will convert more new customers if they never have to wait for example.
- Provide alternative channels – Think beyond email and fax to Instant Messaging and smart phone or tablet applications.
- Maximise your voice quality – Perhaps very simple, but good quality headsets with background noise reduction together with a contact center environment designed to reduce noise and echo will directly improve your customers experience. If this sounds obvious how many contact centres have you called where the agent is hard to hear?
- Have targeted surveys – Forget surveying all customers after a contact. This results in survey overload with a lower customer completion rate. You can get a better result with less surveys that are targeted. For example, survey all first time customers, or the first 10% of respondents to a specific offer, but don’t survey any single customer more than once a year.
Hopefully the above stimulates some ideas on how to use technology to provide your customers a better contact centre experience.
Six tips when doing supplier reference checks
Supplier reference checks are often used as a formality at the end of a selection process rather than as an input during the selection process. It is our observation that this occurs because references provided by suppliers are usually careful managed and thus provide little opportunities for differentiation between suppliers. In this post we outline a few approaches for making the most out of supplier reference checks.
- Validate the references – Perhaps this is obvious, but make sure any proposed references are from a comparable company with a comparable supplier solution. If you are planning a complex global solution then a reference for a local company with a simple solution will be of little value. It can sometimes take a bit of digging to make this validation. If the references are not relevant then do not accept them and ask for new ones. If the supplier cannot provide relevant references then be concerned.
- Ask for more references – Every supplier will have well groomed references, therefore ask for two additional references to those offered willingly. The objective here is to gain access to references that have not been heavily managed. If the additional references result in glowing endorsements then perhaps the supplier is indeed glowing! If two additional references cannot be supplied then find out why.
- Ask for new references – Take note of the companies rattled off as customers in sales presentations and then single out appropriate companies and ask to see a reference specifically from them. Even better ask for this during the sales presentation to reduce unforeseen issues making the reference “unavailable”. If the supplier avoids providing specific references even with ample notice then you should be concerned. Again the objective here is to get representative rather than heavily managed references.
- Find alternative references – User groups are a great source of unbiased references. For best results attend a user group meeting and ask the group president if they could introduce you to members who use the supplier that you are interested in. This approach can be extremely effective but plan for this early especially if user group events are only a few times a year.
- Extract the most from references supplied – Don’t allow the supplier to attend the reference call. Do your research and have questions prepared that focus on specifics and thus voiding high level groomed responses. Make it your objective to be granted follow up calls with others within the reference company such as operational staff, especially those without a vested interest with the supplier.
- Ask the manufacturer/ developer – If the supplier is delivering a solution that is manufactured/ developed by others then ask the manufacturer/ developer for a reference. When making this request understand if the manufacturer/ developer representative has incentives attached to recommending specific suppliers over others as this may reduce the effectiveness of this approach.
Hopefully the above tips can help gather useful information from reference checks that actually assists in differentiating between suppliers and their proposed solutions.
10 considerations when selecting a telepresence solution
Here are 10 considerations, in no particular order, that may be useful when selecting a telepresence solution:
- Do you want to see everyone in a conference or just the active speaker? Consider the balance between life-size images and limited screen space to display participants. With more participants the main images will need to be smaller or voice activated switching is required.
- How important is it to connect to other telepresence, video or telephone systems? Consider traveling executives, external company directors or even companies you work closely with.
- How many people need to use the system? Consider Board meetings that may have larger groups than normal.
- How important is it to share PC content? Consider the type of content (static PowerPoint versus detailed Excel versus movie files) and the type of content displays (dedicated individual or larger common displays).
- How intuitive is the system to find other locations and establish a conference? Consider if you expect users to self-serve or not.
- Do you want to manage the solution or outsource the management? Consider if you can either build/ hire the technical and operational capability to manage the solution or if a truly mature capability exists with outsourced providers.
- How much space do you have? Consider if you have the physical room space required without the need for renovations… it can be very surprising in 6+ seat rooms.
- Do you want dedicated or shared use rooms? Consider if there are trade offs between the telepresence experience and room flexibility.
- Which cities and countries might you expand to in the future? Consider the capability of solution supply and support in those regions.
- Have you held a real meeting in the solution to fully appreciate the experience? To appreciate the difference between the various telepresence experiences available you need to run a real meeting for 1-2 hours rather than attend a demonstration.
