CRM integration
CRM integration decides whether a CRM programme pays back. We design and assure the flows between your CRM and your finance system, marketing tools, support desk, telephony, web shop and data platform, so one customer record runs from first enquiry to cash collected.
Delivered for and alongside
































Every CRM integration uses one of four patterns, and the usual mistake is picking the pattern before the flow is understood.
Salesforce's architecture guidance names the same choices in its own words: request and reply, fire and forget, batch data synchronisation, remote call in, user interface updates on data change and data virtualisation. We pick a pattern per flow, not per project, and write down why. Our ERP integration page covers the same decisions from the finance system's side.
Four patterns, chosen per flow
Most CRM integration failures are ownership failures dressed up as technical ones. Two systems both allowed to edit the same field will disagree, and the sync will faithfully spread the disagreement.
We settle ownership object by object before any mapping is written. The finance system normally owns the billing account: legal name, invoice address, VAT number, credit limit, payment terms and the customer number that becomes the shared key. The CRM owns the relationship: leads, contacts, opportunities, activities, consent and the commercial story. Products and prices come from the ERP, because that is where cost, stock and tax live. Business Central makes the point for you: when items synchronise to Dynamics 365 Sales it creates the price list automatically and Microsoft warns against editing that list by hand, and currency flows one way, from Business Central out.
Matching rules need the same care. HubSpot's data sync matches contacts on email and companies on name or domain, ignoring capitalisation and punctuation, and treats the second record the partner app returns as a duplicate. That is sensible for a marketing database and dangerous for a customer master with two trading entities called the same thing. We define the survivorship rules, the keys that travel between systems and the fields each side may write, then enforce them in the connector or middleware. The wider discipline is set out in our data strategy and governance service, and the one time clean up before go live in our data migration consultancy.
Who owns what
Lead to cash is the full journey: an enquiry becomes a lead, a qualified lead becomes an opportunity, the opportunity gets a quote, the accepted quote becomes an order, and the order is fulfilled, invoiced, paid and recognised as revenue. Quote to cash is the back half of that journey, and it crosses the CRM to ERP boundary at least four times. Each crossing needs a decision about trigger, direction and what happens when it fails.
Quote. Built in the CRM, as HubSpot's Revenue Hub and Salesforce allow, or in the ERP where pricing rules are complex. Business Central can process a quote automatically when a user activates it in Dynamics 365 Sales.
Order release. Orders submitted from Dynamics 365 Sales arrive in Business Central as Released and cannot be edited until reverted to Open on one side or Active on the other. Decide who may do that and how often it will happen.
Credit and fulfilment. Credit holds, stock allocation and delivery dates must flow back so sales can answer the customer without ringing finance.
Invoice and payment. Posted invoices should appear against the CRM account, as Business Central sends them to Sales and as HubSpot's NetSuite integration does for invoices and orders. Overdue status should drive the next conversation, which HubSpot's revenue agent now attempts to automate.
Around that spine sit the other integrations: marketing platforms, where consent and unsubscribes must flow both ways or you will breach your own privacy policy; the support desk, where ticket status and entitlements belong on the account; telephony, logging calls against the right contact; e commerce, where web orders and new customer accounts must not create duplicates; and the data platform receiving a nightly extract. We map all of it on one page before build, and the same thinking shapes our HubSpot and Salesforce work.
Lead to cash, and the four boundary crossings
We have seen the same integration failures often enough to test for them by default:
Duplicates created on both sides because the match key was email in one system and account number in the other.
Sync loops, where a timestamp update on one side triggers an update on the other, forever.
Mapping drift when someone renames a property or adds a pipeline stage without telling the integration owner.
Rate limits hit by a bulk import, so a month end run fails halfway and nobody notices until the reconciliation.
Filters that only governed the initial load, as HubSpot's do, so records you excluded start syncing once they change.
Changes made under the integration user that never synchronise, which Microsoft warns of in Business Central.
Currency, VAT and rounding differences that make the CRM total and the invoice total disagree by pennies, then by thousands.
Testing therefore runs on real data shapes, not clean samples: end to end scenarios through every crossing, volume tests at peak load, deliberate failures to prove the retry and alerting work, and a reconciliation report comparing counts and values on both sides. Monitoring continues after go live with a failing records queue that someone owns, alerts on error rates, and change control on properties and stages. On BP Pulse, getting the flows between systems right cut billing errors and improved the financial reporting that regulators saw, and that is the standard we hold integrations to. Our team has implemented Oracle, Dynamics, NetSuite, Sage, SAP S/4HANA, Unit4 and Access, so the finance side of the boundary holds no surprises. You can engage us for Part A, independent assurance of your partner's integration design and testing, or Part B, delivery of the integration by our own practitioners. Fees are shown plainly on our pricing page, and our testing and go live service describes the cutover routine.
Ran the procurement of every software provider on the National Grid separation, more than 22 million assets, through to implementation including Oracle to SAP S/4HANA.
Delivered the Oracle to Microsoft Dynamics migration at BP Pulse 39 days ahead of forecast, with a £25m saving, the largest on a programme where PEF held the main contract.
Delivered the £9bn Energy Bills Support Scheme for Ofgem, BEIS and HMRC.
Implemented and delivered most major finance systems over 19 years: Oracle, Microsoft Dynamics, NetSuite, Sage, SAP S/4HANA, Unit4 and Access.
A suite of consultants who have led and analysed pre, mid and post migration decisions and implementations across ERP and CRM platforms, independent of every vendor.
CRM integration connects your customer relationship system to the other systems that hold customer data, most importantly the finance or ERP system, plus marketing platforms, the support desk, telephony, e commerce and reporting. It uses native connectors, integration platforms, direct APIs or batch files to move or synchronise records such as accounts, contacts, products, prices, quotes, orders and invoices, so each team works from the same facts.
Split it by object rather than picking one system. The ERP or finance system should own the billing account, credit terms, tax details and the customer number, because invoices depend on them. The CRM should own leads, contacts, opportunities, activities and consent. Products and prices belong in the ERP. Write the ownership down, make the integration enforce it, and never let both systems edit the same field.
A native connector is enough when the vendor's mappings, timings and conflict rules match your process, for example syncing contacts, companies and deals between HubSpot and Dynamics. You need middleware or an integration platform when flows cross three or more systems, need transformation or orchestration, must run in seconds, or involve volumes that would hit API limits. Decide per flow, and check the connector's rules before you commit.
Test with real data shapes and end to end scenarios: create a lead, quote it, release the order, post the invoice and check every record on both sides. Add volume tests at peak load, deliberate failures to prove retries and alerts work, and a reconciliation comparing record counts and money values. Keep the reconciliation running after go live, because integrations fail quietly weeks later, not loudly on day one.
Usually because nobody owns them. Properties get renamed, pipeline stages added, users bulk import spreadsheets that trip rate limits, and filters set at launch stop applying to changed records. Without a named owner watching the failing records queue and a change control step for CRM configuration, errors accumulate until finance spots invoices that do not match the CRM. Monitoring and governance are part of the design, not an afterthought.
Where next
Tell us which systems you run and where the data breaks. We will map the flows and show you the fix.