CRM, automation and the stack
Most marketing stacks have grown by accretion: an ad account here, a form tool there, an email platform chosen years ago, a CRM the sales team picked. The result is that the same person exists in several places under slightly different names, and nobody can say with confidence which campaign produced which customer. The fix is not another tool. It is a clear statement of which system holds which record, and which direction data flows between them.
| Layer | What it holds or does | Usual owner | Where it is covered |
|---|---|---|---|
| Ad platforms and search | Generate visits and leads; receive conversion outcomes back | Paid media and SEO | Google Ads measurement, Meta Ads measurement |
| Website, forms and tags | Capture the lead and the identifiers that arrived with it | Marketing operations | Conversion tracking expertise |
| Analytics | Behaviour before and after the form; key events; reporting | Analytics | Analytics hub |
| CRM | Contacts, companies, deals, activities; stage, owner and source | Sales operations, with marketing | CRM for marketers |
| Automation workflows | Triggers, conditions and actions that run on CRM records | Marketing operations | Automation workflows |
| Email and messaging | The channels the workflows send through | Email and lifecycle marketing | Email automation |
| Customer data platform | Unifies identities where several systems hold the same person | Data or engineering | Introduced below |
The table is a map of responsibilities, not a shopping list. A small business runs the whole thing with a CRM that includes forms, email and workflows; a large one separates every layer and adds a CDPUnified first-party customer profiles. to reconcile identities across them. Either way the questions are the same: where does the record of a person live, who may change it, and how do the other systems learn that it changed.
The CRM as the system of record
A system of record is the place a fact is true. If the CRM says a lead is at proposal stage and the email platform says it is still a new subscriber, the CRM wins, the email platform is wrong, and the integration that should have told it is broken. Making that rule explicit, and building every workflow to respect it, is the single decision that separates automation that helps from automation that sends the wrong message to the wrong person. Where a separate marketing platform holds the same people, the guide to syncing the CRM and the marketing platform writes that rule down field by field.
For that to work, the CRM has to be designed for marketing as well as sales. The CRM fundamentals guide covers the objects, the properties that must be filled in, and the source fields that let marketing trace a customer back to a campaign. The lead stages and pipeline guide covers the part that most often goes wrong: written definitions for each stage of the FunnelModel of the customer journey stages. and the exit criteria that move a record from one to the next. Without those definitions a KPIA metric tied directly to an objective. such as lead-to-opportunity rate means something different to each person reading it, and every report built on the pipeline is an argument waiting to happen.
The system of record also has to be hygienic. Duplicates, blank source fields and records nobody owns do not just make reports untidy; they cause workflows to fire twice, to skip people, or to route a hot lead to someone who left the company. Hygiene is a routine, not a project, and the CRM section treats it as one. Its further guides cover auditing and documenting CRM fields, importing contact lists, matching contacts to company records, reassigning records when someone leaves and moving to a new CRM without losing history.
Automation on top
Once the record is trustworthy, automation is the act of writing down what should happen when a record changes and letting the system do it. Every workflow has the same anatomy: a trigger (a form submitted, a stage changed, a page visited, a date reached), conditions that narrow who it applies to, actions that change records or send messages, and exit rules that stop it. The workflows section builds each guide on that anatomy.
Two workflows carry most of the value in a lead-generation business. Lead scoring turns fit and behaviour signals into a number that decides when marketing hands a lead to sales, and it only works when its thresholds are tied to the stage definitions in the CRM. Nurture sequences carry a lead that is not ready yet through content that moves it forward, with entry and exit rules that respect what sales is doing. Both send through the email channel, which has its own flows for welcome and abandoned-cart moments in the email automation section, and increasingly through SMS and WhatsApp, where consent and template rules are stricter.
Automation also fails quietly. A workflow written for last year's pricing keeps running; two workflows email the same person on the same morning; a suppression rule is removed in a hurry and never restored. Automation governance is the inventory, naming, change-control and safety-rail routine that keeps the library honest, and it is the guide to read before the library grows. When a workflow has quietly stopped, a merge queue refills, a value reverts after a sync, a lead lands with the wrong seller or sales stops updating its records, the CRM and automation troubleshooting section starts from the symptom and routes each cause to the guide that owns it.
Connecting marketing and sales data
The reason to join the CRM to advertising and analytics is closed-loop reporting: knowing not just which campaign produced a form fill, but which produced a customer, and what that customer cost and is worth. Cost per acquisition (CACCost to acquire one new customer.) measured at the form is a guess; measured at the won deal it is a fact, and set against lifetime value (LTVTotal value a customer brings over time.) it tells the business which channels deserve more budget.
The mechanics are a pattern rather than a product. Click identifiers and campaign parameters are captured on the form and stored on the contact; when the contact reaches a stage worth paying for, the CRM sends that outcome back to the platform that produced the click. Google documents this through offline conversion imports and enhanced conversions for leads; Meta documents it through the Conversions API. Before any lead data is sent back, check that the business has the consent or other lawful basis to share it with the platform and that identifiers are hashed the way the platform requires; the privacy and consent operations playbook puts that check ahead of every upload. The CRM integration guide explains the pattern end to end and leaves tag setup to the platform guides, and the conversion tracking expertise page describes how the author approaches the whole measurement layer. A site move or rebuild can cut that capture off, and the CRM tutorial Moving a Website Without Breaking the CRM and Its Workflows puts the forms, rules and links in a fixed order.
The same join serves retention. Once purchase and renewal data sit on the contact, lifecycle marketing can trigger on real customer events rather than on a calendar, and the customer analytics work of cohorts and churn has a reliable source.
The sections
CRM for marketers. What marketers need from the CRM: the data model and the fields that matter, written stage definitions and a pipeline both teams trust, and the integrations that send outcomes back to ads and analytics. Core guides on each, further guides on migration, syncing, field audits, list imports, company matching, campaign records, handing over a leaver's records and a website move, and a hub that is the foundation for everything else here.
Automation workflows. How workflows are designed and governed: lead scoring, nurture sequencesKeeping in useful contact with leads who are not yet ready to buy until they are., SMS and WhatsApp automation, and the governance routine that keeps a growing library safe. Core guides on each, further guides on sales alerts, the closed-won workflow, quote follow-ups, standardising field values, lifecycle stage changes and pausing marketing to a customer with an open issue, and a hub that starts with the anatomy of a workflow.
The sections are ordered deliberately. Readers who arrive wanting to build a nurture sequence are usually better served by first checking that their stages are defined and their source fields are filled, because a sequence built on an undefined pipeline nurtures the wrong people towards an exit nobody has agreed. The See-Think-Do-Care framing is a useful lens for both sections: the CRM records which stage a person is in, and automation decides what each stage should receive. The email marketing hub covers the channel most workflows send through, and belongs alongside this one rather than beneath it.
A first implementation plan
A first implementation should produce a trustworthy record and a small number of workflows that are obviously useful, in that order. The sequence below is the one most teams end up following whether or not they planned to.
- Define the objects and required fields. Decide what a contact, a company and a deal are in this business, which properties are mandatory at creation, and who owns each record. Write it down where both teams can see it. If the CRM already exists, the field audit tutorial starts from what is there.
- Write the stage definitions. Every stage gets a definition, an exit criterion and an owner, agreed by marketing and sales together. Thresholds for marketing-qualified and sales-qualified leads are set here, not in the scoring model. The tutorial on automating lifecycle stage changes then sets each stage by rule, with the date it was entered.
- Capture source at the form. Campaign parameters and click identifiers are stored on the contact at creation, so attribution and the later integration have something to work with. Campaign records named like the ads give those parameters something to join to.
- Build the hand-off workflow. When a lead crosses the agreed threshold, the system assigns an owner, creates a task and notifies the salesperson. This is the workflow that proves the value of the rest.
- Connect outcomes back to the platforms. Won and qualified stages are sent to the ad platforms and analytics so bidding and reporting see real outcomes. The closed-won workflow tutorial sends the won outcome as part of everything else that should change when a deal is won.
- Add nurture and scoring. With the record and the hand-off in place, nurture sequences and a calibrated scoring model have something sound to run on.
- Govern from the start. Name workflows consistently, keep an inventory, and schedule the first audit before the library grows.
A software consultancy's first quarter
A consultancy with 6 salespeople and one marketing manager inherits a CRM holding 14,000 contacts, of which 3,900 are duplicates and 5,200 have no source recorded. Month one is spent on the record: merging duplicates, making source and owner mandatory on creation, and agreeing five pipeline stages with written exit criteria. Month two adds click identifier capture on the three enquiry forms and a single hand-off workflow that assigns new qualified leads within 15 minutes and creates a follow-up task. Month three sends won deals back to Google Ads through an offline import and switches on a four-email nurture for leads that stall at the first stage. At the end of the quarter the team can report, for the first time, that 41 of the 120 deals won that year came from paid search at an average cost of £310 per won deal, while the 28 that came from a partner webinar cost £95 each. The budget conversation that follows is the first one based on customers rather than form fills.
The plan deliberately leaves the clever parts, predictive scoring, multi-channel orchestration and a customer data platform, until the record has earned them. Most businesses that struggle with automation did not lack features; they automated on top of a record that could not bear the weight.
