Progress on IFRS 17 Implementation Journeys: A Maturity Map

With the reporting deadline now visible on the horizon, the question I get most from clients is no longer “what does IFRS 17 require?” — it is “are we behind?”. That is a useful question but a hard one to answer in the abstract. Every insurer I have worked with sits at a different point on the implementation journey, and the gap between the front runners and the middle of the pack is wider than most boards realise.

This post is a map of that journey. It sketches the phases every programme goes through, describes where the industry actually sits right now based on projects I have touched in South Africa, Malaysia, Hong Kong, and Saudi, and flags the phases that tend to drag. If you are trying to benchmark honestly, this is the yardstick I use.

1. The maturity scale

Implementation is not a single workstream. It is a chain of six phases, and you only get to the next one by closing the previous one cleanly:

flowchart LR
    A[Gap analysis] --> B[Financial Impact<br/>Assessment]
    B --> C[Implementation<br/>roadmap]
    C --> D[Vendor selection]
    D --> E[Configuration<br/>& testing]
    E --> F[Go-live &<br/>parallel runs]

Most of the industry sits somewhere between phase 3 and phase 5. The front runners are in phase 5 working on parallel runs. The tail is still finishing phase 2. Everyone tends to slightly over-state where they are, because the phases overlap and it is always possible to point at a document that suggests a later phase is “in progress”. The honest test is whether the deliverables of the previous phase have actually been signed off and are being used downstream. Until then, the previous phase is still open, even if the next one has started.

2. Gap analysis — mostly done, but not always honest

Nearly every insurer has completed some form of gap analysis. The quality of those assessments varies enormously. A good one identifies, per system, what exists, what needs changing, and what is missing altogether — broken down into actuarial models, accounting engines, data staging, disclosures, and general ledger. A weak one identifies the standard’s requirements and notes “further work required” against most of them.

The tell is whether the gap analysis fed a concrete work plan with owners, dates, and cost estimates, or whether it became a PowerPoint that was read once and shelved. If it is the latter, the programme is effectively still at phase 1 regardless of the label it uses.

3. Financial Impact Assessment — where half the industry still is

The FIA phase is where most of the middle of the pack sits. The goal is to run the new standard through a representative sample of the book and produce a first pass at the income statement, balance sheet, and projected profit impact. Done well, it surfaces three things: which measurement models apply where, what the working assumptions look like, and which data items you simply do not have yet.

The reason this phase drags is that FIA results raise more questions than they answer. A first run typically generates a list of methodology decisions — discount rate choice, risk adjustment approach, CSM amortisation pattern, transition method by portfolio — that the insurer had not finalised. Each decision loops back into methodology, and the FIA has to be rerun. The better run programmes treat the FIA as iterative from the start and budget for three to four passes. The ones that struggle try to treat it as a single deliverable.

4. Implementation roadmap — the phase nobody admits is hard

Once the FIA is stable, the insurer is in a position to plan the remaining work end to end. This sounds like the easy phase. It is not. The roadmap has to reconcile four moving parts: the vendor selection timeline, the internal system changes, the actuarial model rebuild, and the finance and disclosure readiness. Each of those has dependencies on the others, and the critical path is almost never where management expects it to be.

The common failure mode is planning the workstreams in parallel without sequencing them against the vendor. The vendor’s implementation approach constrains when data staging can be built, when the actuarial model needs to deliver cash flows in a specific format, and when the subledger mapping has to be finalised. Roadmaps that do not start from the vendor’s plan tend to get re-written once the vendor is chosen.

5. Vendor selection — where the bulk of the market is right now

A very large share of insurers I have seen are currently in vendor selection. RFPs are out, shortlists are being scored, and proof-of-concepts are running. This is the phase where the advantage of being behind the front runners shows up: the vendor products are more mature, release cycles have stabilised, reference installations exist to visit, and the consulting community has real implementation experience to draw on.

The risk is that vendor selection drags into a second half of the year it was not meant to occupy. The drivers are usually (a) internal alignment on the selection criteria, (b) procurement cycles, and (c) the insurer realising mid-RFP that the components they actually need are broader than the accounting engine alone — staging, transition calculation, profitability testing, sometimes actuarial modelling too. Programmes that front-loaded the component scoping in the roadmap phase move through selection much faster.

6. Configuration and testing — the front runners’ territory

The front runners have selected a vendor, signed the implementation contract, and are configuring the solution against their book. This phase is where the workload becomes concrete: mapping the chart of accounts, defining the business events, tagging contracts to groups, running test cases, reconciling shadow-model outputs against vendor outputs, and fixing the inevitable differences.

What the front runners discovered — and what the middle of the pack will discover soon — is that configuration is not a vendor deliverable. It is a joint exercise, and the insurer carries most of the weight. Internal resources have to translate the methodology decisions into vendor parameters, sign off on the posting logic, and own the validation. A vendor that is “doing the configuration” usually means the vendor is setting defaults and the insurer is rubber-stamping them, which surfaces as reconciliation problems three months later.

7. Go-live and parallel runs — very few are here

A small number of insurers are already in parallel run, producing IFRS 17 results alongside their current GAAP reporting, reconciling the two, and working on the transition message for the market. The challenges here shift from building to running — shortening the close cycle, tightening controls, industrialising the data pipeline, and preparing the transition calculations and disclosures for the opening balance sheet. Parallel runs expose process weaknesses that configuration testing never caught: timing issues in data arrival, manual reconciliation steps that do not scale, and governance gaps around sign-off.

8. What being behind the front runners actually costs

Being in the middle of the pack has real advantages. Vendor products are more mature. A pool of IFRS 17 specialists now exists that the front runners had to build themselves. Tools and accelerators are available that did not exist two years ago. Virtual working has normalised, and international specialists can plug in at competitive rates.

The costs are also real. Scarce specialist capacity is booked out by programmes that started earlier. Time pressure means less ability to iterate on methodology decisions — you have to commit earlier with less evidence. And if something breaks late in the programme, there is less runway to fix it before go-live.

The honest conclusion for most insurers is: the scale above is a useful map, but the specific phase matters less than the question of whether the programme is moving. Programmes that close phases cleanly, even slowly, reach go-live. Programmes that keep every phase half-open run out of runway. If your map tells you everything is in progress and nothing is closed, that is the signal to act on, not the phase label.