How to Evaluate a CMS Migration Partner: The Reference Questions That Matter
By the time you get to reference calls, you have usually already decided. The shortlist is down to two, procurement wants a name by Friday, and the calls get booked as a formality. So you spend thirty minutes with a happy executive sponsor who confirms what the pitch already told you, and you sign. Six months later the editors hate the authoring model and nobody can explain how the page count doubled.
Reference calls are one of the few opportunities to test a CMS migration partner’s sales narrative against a client’s experience, although the references themselves are usually curated. eight25 is a Contentful Silver-tier Solution Partner and a Contentstack partner, with enterprise migration experience that includes NetApp, Sophos, and Qlik. Across those engagements, the most useful evidence often came from the editors, platform owners, and technical leads who worked directly with the new platform, not solely from the executive sponsor selected for the call.
The list you are handed is a sales asset
Every agency keeps three or four accounts in rotation for reference calls. They are the projects that went well, with sponsors who are still in role and still friendly. That is not dishonest. It is what a reference list is for. But it means the list tells you what the agency is capable of on a good day, not what your project will look like.
Two requests fix most of this, and both are reasonable enough that refusing them is itself a signal.
Ask for two or three recent migrations from your source platform or, where confidentiality prevents that, the closest comparable engagements the partner is permitted to discuss. Do not ask only for its most successful projects. If you are leaving Sitecore, experience with Sitecore’s rendering logic, personalization rules, templates, integrations, and accumulated customization is especially relevant. Experience with other enterprise platforms can still be useful, but a partner that has recently completed a comparable source-to-target migration, using substantially the same delivery team proposed for your project, generally presents less execution risk.
Ask for one reference from a project that encountered a material challenge, such as scope growth, an integration problem, a compressed deadline, or a difficult cutover. Not every client will be available because of confidentiality restrictions, but the partner should be able to explain what changed, how it communicated the issue, and how the team resolved it. A useful request is: ‘Please connect us with a client whose scope changed significantly after kickoff, or walk us through a comparable example if a reference call is not possible.
Match the reference to the shape of your program. A 40,000-page, multilingual, multi-brand estate and a 600-page, single-locale site carry different content, governance, integration, and cutover risks. A smaller engagement can still reveal how the partner communicates and manages delivery, but it provides weaker evidence about the partner’s ability to handle your scale and complexity.
The executive sponsor should not be your only reference
An executive sponsor can explain whether the project met its business objectives, budget expectations, and leadership commitments. They may not, however, have direct visibility into daily authoring, technical delivery, or the transfer of platform ownership. Balance the sponsor’s perspective with input from people who worked in and operated the platform.
The people closest to delivery can answer questions that the executive sponsor may not be equipped to address. Where the client permits it, ask to speak with representatives from three roles not necessarily three separate reference accounts:
| Who | What only they can tell you |
|---|---|
| A content editor or publisher | Whether the authoring model works in production, how long training took, and what they still cannot do without a developer |
| The post-launch platform owner | Whether ownership actually transferred, or whether the team is still calling the agency for routine changes |
| The technical lead | Whether the integration work was scoped honestly, and what the agency’s engineers were like in launch week |
The editor conversation is easy to overlook, but it can reveal whether the migration improved everyday content operations. Migration programs often devote substantial attention to architecture and platform selection; the resulting content model may still create friction when editors begin using it at scale. Publishers can tell you whether common tasks became faster, whether previews and approvals work as expected, and where developer assistance is still required. In eight25’s published account of NetApp’s migration from Tridion to the headless Contentstack platform, publishing a PDF had previously required 30 to 40 minutes of manual CMS work. After migration, it became an upload workflow, with metadata pulled automatically from the DAM when the relevant fields were blank. That type of operational improvement is best validated with the publishers who use the workflow. It also illustrates an important principle of headless migration: the value comes from redesigning content operations and integrations, not simply replacing one CMS with another.
If an agency will not connect you with anyone below the sponsor, treat that as an answer.
What to ask: the three things a demo cannot show
Most reference-call scripts online target software buyers and ask about support tickets. Migrations fail somewhere else.
Scope, after kickoff
Migration scope frequently changes as discovery progresses. Source content may be less consistent than expected, previously undocumented properties may surface, and locale or integration requirements may prove more complex. In eight25’s account of the NetApp program, the initial estimate was 40,000 pages, a later audit put the estate near 12,000, and approximately 14,000 pages were tracked at launch. Variance of this kind is not unusual in a large content estate. The important question is whether the partner has a transparent method for validating scope, managing change, and protecting the launch plan.
Ask the reference:
- How much did the scope change between kickoff and launch, and how did you find out?
- Was there a change-order process you understood in advance, or did costs appear as surprises?
- When an item fell into a genuine scope gap, how did the partner and client decide whether its cost or schedule impact would be absorbed, shared, or handled through a change request?
A strong answer should be specific and procedural. According to eight25’s account of the Sophos migration to Contentstack, a formal, priced change-request log protected the Phase 1 scope against a fixed December 11, 2025 launch. The partner portal, ecommerce work, and partner-blog migration were managed through separate signed change orders, with three running simultaneously at peak. On a reference call, ask the client to confirm whether that process was transparent, predictable, and understood before the additional work began.
Whether the editors actually use it
A migration that launches on time and leaves the content team dependent on developers has moved the problem rather than solved it. Ask:
- How long after launch before your team was publishing without help?
- What training did you get, and was it built for your content types or generic?
- What do editors still complain about?
The last question is the useful one. Every platform has friction, and a reference who names a specific irritation is telling you the truth. One who says everything is fine has either not asked their team or is the wrong reference.
The first month after go-live
Launch is a date. Operating the platform is the actual deliverable. Ask what the first forty-eight hours looked like, what broke, who was awake, and how long support lasted. Ask whether the agency’s post-launch commitment sits in the contract or only in the pitch deck.
Then compare the original timeline with the actual result: What launch date was initially committed, when did the site go live, and did the delivered scope change along the way? A short, transparently communicated delay may indicate less risk than an apparently on-time launch achieved through uncommunicated scope reduction. eight25 reports that NetApp’s go-live occurred four days after the original target; use the reference call to verify how the delay was communicated and whether it affected the business.
How to read what you hear
Reference feedback is often positive because the partner selects clients willing to participate. Rather than trying to interpret tone or hesitation, test general praise with requests for specific examples. If a reference says the team was responsive, ask what happened when an urgent problem arose, who responded, and how quickly it was resolved. Specific events are more reliable evidence than vocal cues.
Then there is praise with no texture. “Great team, very responsive” describes nothing. Someone who lived through a migration can tell you the name of the project manager and one specific thing that went wrong. If they cannot, they were not close enough to the work.
A reference who can discuss business outcomes but not delivery details may simply have held an executive role rather than having been coached. Clarify the person’s involvement, then ask for another contact who can speak to training, technical delivery, content operations, and the first weeks after launch.
One question compresses most of this: would you hire them again for the next one, and is that a yes or a yes-with-conditions? Almost nobody says no. The conditions carry the information.
What we tell buyers to ask our own references
eight25 hands prospects a list of questions to ask our references rather than a list of quotes, because the questions above are the ones we would want asked about our own work. Ask NetApp’s team what happened when the page count moved. Ask Sophos’s publishers what author training looked like across content types in the weeks before a fixed December launch. Ask Qlik what enabled its Sitecore XP-to-Contentful migration to finish in under five months, how well the new content model supports its global publishing operation, and whether eight25’s migration tooling delivered the expected efficiency. Contentful reports that page-migration work that might otherwise have taken an hour or more per page was reduced to approximately five minutes, including quality assurance. Confirm any region, language, or production-time figures directly with Qlik before adding them.
We would also point out where references are the weaker signal. If your migration has no close analogue in a partner’s portfolio, no reference call will tell you much, and a paid discovery sprint or a proof-of-concept built on your own content is stronger evidence than a conversation about someone else’s project. Reference calls work best when you can find a near-twin engagement and reach the people who did the work. They are weakest when a happy stranger vouches for a project shape they have never run.
Frequently Asked Questions
What should I ask a CMS migration agency’s references?
Ask about three things a demo cannot show: how much the scope changed after kickoff and how the partner handled it, how long after launch the client’s editors could publish without developer help, and what the first month after go-live actually looked like. Then ask whether they would hire the partner again, and whether that is an unconditional yes. The conditions attached to a yes carry more information than the yes itself.
Who at a reference company should I actually speak to?
Ask for three people rather than one: a content editor or publisher, the person who owned the platform on day 31, and the technical lead. The executive sponsor approved the project and has an interest in it having gone well. Editors and platform owners live with the outcome and will tell you whether the authoring model and the ownership transfer actually worked.
How many references should I ask a CMS migration partner for?
Ask for their last three migrations off your specific source platform, in chronological order rather than curated, plus one engagement where the scope moved significantly after kickoff. Recency and platform match matter more than volume. Three relevant references beat ten flattering ones from projects that look nothing like yours.
What are the warning signs in an agency reference call?
Warning signs include praise unsupported by specific examples, an inability to provide any perspective beyond the executive sponsor, vague answers about scope changes or post-launch support, and no willingness to discuss a project that encountered difficulty. Confidentiality may limit which clients or individuals can participate, so assess whether the partner offers credible alternative evidence rather than treating an NDA alone as a red flag.
Are reference calls enough to choose a CMS migration partner?
No. They are one input, and least reliable when your migration has no close analogue in the partner’s portfolio. A partner willing to build a proof-of-concept on your own content, or run a paid discovery sprint that produces real scope before you commit, gives you better evidence than a conversation about an unlike project.
Before you book the calls
Before booking reference calls, define the risks you need them to test: platform fit, content-model usability, integration complexity, delivery transparency, ownership transfer, and post-launch performance. Prioritize references that resemble your source platform, target architecture, scale, and content operation, and seek perspectives from editors, platform owners, and technical leads wherever possible.”
“eight25 welcomes that level of scrutiny. Our work across Sitecore, Tridion, Contentful, and Contentstack has shown that successful headless migration depends on more than moving content: it requires a usable editorial model, a well-designed integration architecture, disciplined delivery, and a clear path to client ownership. We can help you develop a reference-call scorecard, even if you use it to evaluate multiple partners.
Talk to our team about scoping a migration
How to evaluate a CMS migration partner — the questions to ask before you sign