+44(0)1689 258 210 | enquiries@e-focusmarketing.com
- CRM Strategy Solutions
- Our CRM Methodology
- CRM Resources
- Supper Club
- Contact
+44(0)1689 258 210 | enquiries@e-focusmarketing.com
Optimisation is the sixth of the six capabilities in the Customer Relationship Maturity Model (CRMM), developed by Kate Barrett of eFocus Marketing, and the second of two in the Evolve stage. It measures whether an organisation tests and proves what drives CRM performance, isolating cause from correlation through genuine experiments, and whether what it learns changes what happens next.
Most teams run tests. The CRMM asks a narrower question: did the last three tests change a decision?
Every time I have looked at a testing programme in an audit, the volume of tests has told me almost nothing about the level, because the common pattern is a run of subject-line splits with no hypothesis, no minimum sample, and a winner declared on a Tuesday afternoon because someone needed to send. That is activity that looks like optimisation, and the CRMM scores it as Level 2.
How do you find out what is actually driving results, rather than guessing? When you made your last significant change, did half the audience not get it, so you could compare? What is the minimum bar (sample, duration, significance) a result has to clear before it is allowed to change anything? And when a test wins, what happens on Monday?
The last question is the one that matters. A test that wins and changes nothing has cost you the send it ran on and taught you nothing you will use, which is worse than not testing at all because it feels like rigour.
The capability also covers the discipline of testing assumptions formally rather than inferring them from reporting: the difference between “customers who get the welcome series spend more” (a correlation, and probably a selection effect) and “customers randomly assigned to the welcome series spent 12% more over 90 days than those held out” (a cause you can spend money on).
The CRM Priority Check asks one question about this capability, with five statements describing how organisations actually test and prove what works. The levels below describe the evidence at each point.
Changes are made on instinct or opinion, and at most results are compared before and after. No tests are run, so nothing the team believes about what works has been checked.
Some A/B tests run occasionally, but results are not rigorously analysed or consistently acted on. Testing is a nice-to-have when there is time rather than an embedded practice.
The organisation has the discipline to run genuine tests, isolating cause from correlation, and tests assumptions formally rather than reading them off a report. A result has to clear a defined bar before it counts.
Experimentation is used systematically to establish what is actually driving performance, and results feed directly into what happens next. The testing roadmap and the activity plan are the same document.
Proof from testing continuously reshapes how the organisation operates, not just the specific thing that was tested, and it does so as a permanent way of working rather than a periodic process.
At Level 1, do not try to build a testing programme. Pick one change you were going to make anyway and split it instead: half get the change, half do not, then compare properly. One genuine test, however small, teaches your team what real looks like before you try to scale the discipline.
At Level 2, set a minimum bar for what counts as a real result before it is allowed to change anything (sample size, duration, significance, whatever suits your volumes), then pick your last three tests and check whether the result actually changed a decision. If none did, the gap is in using what testing tells you rather than in testing.
At Level 3, make sure the discipline does not silently depend on one person’s rigour. Document the standard you hold tests to, so it survives someone being on leave or leaving, and so a less rigorous test cannot slip through under the same name.
At Level 4, the risk becomes volume outpacing rigour, with more tests than you can properly analyse. Keep enforcing the minimum bar as volume grows, so speed does not quietly erode the proof standard that got you here.
At Level 5, protect the habit deliberately with a standing slot in your team rhythm that does not get cancelled for being “just” a test review, because testing discipline is usually the first thing that slips under pressure and nothing visibly breaks when it does.
Optimisation is capability six of six in the Customer Relationship Maturity Model and the second of two in the Evolve stage. It proves what Measurement reports, and it feeds what it learns back into CRM Strategy, Execution Excellence and, at the highest levels, into the definitions inside Customer Progression Design. Whether a clean test can be run at all is often a Technology and Enablement question.
When this scores lowest in a CRM Priority Check it is flagged as the capability to fix first. The eFocus Testing Framework (Prioritise, Hypothesise, Design, Execute, Evaluate, Apply, Learn) is the method behind it, and CRM Optimisation is the engagement that builds it.
Usually Level 2. Subject-line splits are tests, but if there is no hypothesis, no minimum sample and no record of what the winner changed beyond that one send, the discipline is not there yet. The level is set by rigour and by what happens after the result, not by how often you split.
It depends on your volumes, which is why the CRMM asks whether you have one rather than what it is. For most consumer brands a workable starting point is a pre-registered hypothesis, a sample sized in advance, a fixed duration, and a significance threshold you do not move after seeing the data. The last of those is where most programmes fail.
Not to reach Level 3. A hold-out group, a spreadsheet and a written standard get you there. Platforms help at Level 4, when volume and speed make manual analysis the bottleneck, and the risk at that point is that the tool runs more tests than anyone can interpret.
Because everything else changed too: the week, the weather, the trading calendar, the competitor’s sale. A before-and-after tells you what happened around the change; a controlled split tells you what happened because of it. The CRMM scores the former as Level 1 for that reason.
Optimisation is where the loop closes. Discover designs the relationship, Grow delivers it, Measurement reports it, and Optimisation proves which parts of it are working so the next cycle of design and delivery starts from evidence rather than opinion. A Level 4 or 5 score here with weak capabilities upstream is rare, because there is little worth testing until the others exist.
Free Peak Readiness Assessment
Answer 22 questions to get your readiness score, your sharpest gap with three actions to take this week, and a to-do list for before, during and after the peak.
Start my free assessmentFree. A few minutes. Black Friday is on 27 November 2026, so the sooner you know, the more you can fix.