The decision you are actually making
A CRM selection looks like a software comparison and behaves like an operating change programme with a licence attached. The platforms on any credible shortlist overlap heavily on features, which is why organisations that start with demos end up deciding on presentation quality and price. The sequence in this playbook exists to prevent that: requirements, data and partner questions are settled before any vendor is contacted, so the demos answer your questions instead of setting them. It complements the Before Choosing a CRM guide, which covers the risks to pressure-test before commitment; this playbook walks the selection itself, from first framing to rollout sequencing. Plan for twelve to sixteen weeks from kick-off to signature for a mid-market commercial organisation, run by a commercial operations lead with explicit sponsorship from the CRO or CEO. The steps assume you may conclude, part-way through, that the current platform configured properly is the right answer. That is a legitimate outcome of the process, not a failure of it.
Step 1: Write down what the CRM must change
Before any vendor contact, produce one page stating the three commercial outcomes the system must move and the behaviours leadership is prepared to make mandatory. Outcomes need current baselines: if forecast accuracy is the target, record how far the last four quarterly forecasts landed from actuals and how that number is produced today. If pipeline visibility is the target, record how long it takes to assemble a reliable pipeline view and how many spreadsheets are involved. Then the harder half: which behaviours become non-negotiable. Opportunities not in the system do not exist for commission purposes. Pipeline reviews run from live screens, not exported decks. Stage definitions are enforced by sales management, not suggested by the admin team. Circulate the page to the CRO, the CFO and the regional heads, and have them sign it. Resistance at this step is cheap information: if leadership will not commit to the behaviours on paper, no platform will supply the discipline later, and you have saved the organisation a seven-figure programme.
Step 2: Audit the data estate and the integration map
Two audits determine more of the eventual cost than the licence price, and both can be finished before a shortlist exists. The data audit counts what the new platform inherits: how many account records exist, what percentage are duplicates or dormant, how many have disputed ownership between teams or regions, and which fields are populated reliably enough to migrate. Assign a named data owner with authority to settle ownership disputes, because those disputes are commercial arguments and they take calendar time regardless of tooling. The integration audit lists every system the CRM must exchange data with: billing, quoting, marketing automation, service desk, data warehouse. For each, record the direction of data flow, the owner, and whether a supported API exists today. This map becomes the honest core of the cost model in Step 6 and the demo script in Step 4. Selections that skip it discover the real project in month two of implementation, when the integrator prices what the vendor's proposal politely left out.
Step 3: Build the weighted requirements model before any demo
Agree how the decision will be scored before anyone has a favourite. Run a half-day workshop with sales, marketing, service, finance and IT to separate gate requirements (anything answered yes or no, such as data residency or a mandatory integration) from weighted criteria, then fix the weights using the Weighted Scoring Model. Weights agreed after demos drift towards whichever vendor performed best, which defeats the purpose. Keep the criteria few and evidence-based.
| Criterion group |
Typical weight |
Evidence that counts |
| Fit to core commercial workflow |
25–35% |
Scripted demo on your pipeline structure, scored independently |
| Integration effort and openness |
15–25% |
Technical review of the Step 2 map against vendor APIs |
| Usability and adoption effort |
15–20% |
A real sales team completing set tasks unassisted |
| Four-year total cost |
15–20% |
Your own cost model, not the vendor quotation |
| Vendor and partner strength |
10–15% |
References at your size and stage, taken without the vendor present |
Step 4: Shortlist three platforms and script identical demos
Use the requirements model to cut the market to three candidates; the CRM Platforms landscape maps the categories the shortlist will come from. Resist adding a fourth or fifth platform out of politeness, because every additional candidate dilutes evaluator attention and adds weeks. Send each vendor the same demo script two weeks in advance: your sales stages, your product structure, your discount approval thresholds, your renewal motion, and two integration scenarios lifted from the Step 2 map. State plainly that the showroom demo will score zero. During each session, evaluators score against the published model in silence and submit before the group debrief, so the loudest voice in the room does not become the methodology. Ask each vendor to leave a sandbox configured to the script for a week afterwards, and give a working sales team a set of tasks to complete in it unassisted. The gap between demo polish and sandbox usability is one of the most predictive numbers in the whole selection.
Step 5: Run the implementation partner competition in parallel
The partner who configures and migrates the platform influences the outcome at least as much as the platform choice, so run partner evaluation as its own competition rather than accepting the vendor's introduction. For each shortlisted platform, meet two candidate partners. Interview the people who would deliver, not the people selling: ask for named consultants with CVs, what they are staffed on now, and when they roll off it. Ask each partner for their last three projects of comparable scope, the original contract value against the final invoiced value, and a reference from a project that went badly, which tells you more than the happy references will. Probe the mundane mechanics: who arbitrates when your data owner and their migration lead disagree, what a change request costs to raise, how much partner time is senior versus offshore delivery. Score partners with the same discipline as platforms. Platform and partner are one combined bet, and the scoring should treat them that way.
Step 6: Price the deal and the risk before negotiation starts
Build a four-year total cost model per finalist from your own inputs: licences at realistic seat growth and tier mix, partner implementation fees with a change allowance, each integration from the Step 2 map, data cleansing effort, an internal administrator, and training with refresher cycles. Then score each finalist with the Procurement Risk Calculator before commercial negotiation starts, so concentration, dependency and exit risks are priced rather than discovered. Complete a Vendor Scorecard Template per finalist to hold capability scores, cost and risk in one view the steering group can compare. The point of this step is negotiating posture: a buyer who can show a vendor exactly where its proposal is expensive or risky extracts better terms than one negotiating from a discount request. It also surfaces the uncomfortable cases early, such as the cheapest licence proposal carrying the most expensive integration bill, which is a pattern the sales process is designed to obscure.
Step 7: Negotiate for year three, not day one
CRM contracts are priced for the organisation signing them and exercised by the organisation you become. Negotiate seat flexibility in both directions, because commercial teams shrink as well as grow, and per-seat deals rarely contemplate reduction. Cap renewal uplifts now, while you have alternatives; at renewal you will have migration costs holding you in place and no negotiating power. Fix the partner rate card for two years and add key-person clauses for the named consultants from Step 5. Secure your data exit in the contract: format, completeness, cost and timescale of a full export, tested before final payment rather than asserted in a schedule. Tie a meaningful payment tranche to acceptance criteria drawn from the Step 1 outcomes, not to go-live alone, because go-live is a vendor milestone and adoption is yours. None of these terms is exotic. All of them are cheaper on the day you have three interested vendors than on any day afterwards.
Step 8: Sequence the rollout for adoption, not coverage
Resist the instinct to switch everyone on at once. Deploy first to one commercial team with its integrations working from day one, because a rep who sees billing history and quote status in the account record has a reason to log in, and a rep who sees an empty screen has a reason never to return. Have that team's leader run every pipeline meeting from the live system from the first week. Set expansion gates on evidence, not calendar: field completion rates, opportunity ageing hygiene, and whether the team's forecast is genuinely built from stage data. Hire or designate the platform administrator before go-live and route every report request and field change through them, with a published turnaround. Keep a visible fix log in the first month; early complaints handled fast convert sceptics better than any training programme. Only when the first team's data quality holds for a full cycle does the second wave start.
Where independent challenge earns its cost
Three moments in this playbook reward outside challenge disproportionately. The first is the end of Step 3, before the shortlist is fixed, when requirements quietly shaped by the incumbent platform or by one loud function are cheapest to correct. The second is the end of Step 6, when the scores are close and the difference between finalists is really a judgement about partner risk and adoption effort, which is exactly the judgement internal teams find hardest to make about themselves. The third is the week before signature, when momentum is strongest and dissent is at its most expensive to voice. Digital Advisory puts a CRM selection in front of selected senior operators who have run commercial organisations through platform replacements: people who can read your scorecard and tell you which line they would not trust. Set against the four-year cost model from Step 6, a confidential pressure-test is one of the cheapest items in the programme.
