The CRM nobody filled in, now filling itself in.
Every sales system of the last twenty years asked the same thing of sellers: stop selling for a moment and tell us what happened. Most of them never got an honest answer, which is why so many forecasts were fiction. Sales Cloud’s current pitch is to invert that — capture the calls, emails and meetings automatically, keep the account history beside them, and let agents do the research and the chasing so the seller does the selling.
Everything here comes from what Salesforce publishes — its product pages and its newsroom. Its customer figures are labelled as its own, because that is what they are.
Truth, context, action.
Salesforce organises its own Sales Cloud pitch around three words. They are worth taking seriously, because each one is a different kind of problem and most CRM projects only ever solve the first.
What actually happened, captured without being typed
Calls, emails and meetings captured automatically rather than entered after the fact. This is the foundation and the one most organisations have never had: a record that reflects reality instead of reflecting whoever was diligent about data entry on a Friday afternoon.
Account history beside live signals
Thirty years of account history, as Salesforce puts it, with what is happening now set against it. A renewal looks different when you can see that this customer has churned once before, or that the person who championed you has just changed jobs.
Handoffs between sellers and agents that do not drop
Autonomous agents that research prospects, prepare deals and monitor accounts around the clock, flagging risks and opportunities — and handing them to a person at the point where judgement is actually needed. Salesforce calls the arrangement agentic selling: every seller with a small team of agents doing the operational work.
What a seller stops doing.
The test of any of this is not the dashboard. It is whether a working seller’s Tuesday is different.
An hour of admin, then selling
- Writing up yesterday’s calls from memory, badly.
- Re-reading an account history scattered over email, a shared drive and someone else’s head.
- Hunting for the last quote, the signed contract and the current price list.
- Updating close dates the night before the forecast call.
- Finding out a champion left two months after they left.
- Building the same proposal for the fourth time this quarter.
The admin happens around them
- Calls, emails and meetings captured as they happen.
- Account history in one place, with the live signals against it.
- Research and preparation done overnight by an agent, ready to read.
- Risks flagged when they appear rather than at quarter end.
- Quoting, approvals and contracts on the same record.
- Deal updates in Slack or Teams, without opening the CRM at all.
That right-hand column is the promise, not a report of what you will get. Which of those six you actually land depends almost entirely on the section below.
An agent cannot read what nobody wrote down.
Here is the thing we say in the first meeting, because it saves everyone a year. Agents read the same fields a person would. Where those are stale, blank or contradictory, an experienced rep quietly works around it — they know the close date is optimistic and the owner is wrong. An agent does not know that, so it reports what the record says, confidently.
This is why AI makes bad CRM data feel worse before it feels better. It is also the best thing that can happen to you, because for the first time the cost of the mess is visible to the people who could fix it.
So the order of work is not negotiable for us: hygiene, then automation, then agents. Required fields that are genuinely required. One definition of a qualified lead, agreed by the people who disagree about it. Owners on accounts. Close-date discipline with a consequence. And automatic activity capture, so the record stops depending on anybody’s memory.
Every organisation that told us its CRM data was fine had not looked. Every organisation that told us it was a disaster was about three weeks from fixing it.
From the discovery stage of more projects than we would like to admitClose dates in the past
The single most common finding. It makes every forecast wrong and every agent wrong with it.
Accounts with no owner
Nobody is accountable, so nothing is maintained, so nobody trusts the segment it sits in.
Four definitions of qualified
One per team. Every conversion-rate argument traces back to this, and none of them is a data problem.
Contacts who left in 2023
Still receiving campaigns, still counted in reach, still the named champion on a live renewal.
Notes in a private inbox
The real account history, invisible to everyone and gone the day that person resigns.
Stages that mean nothing
Six stages, all advanced by optimism rather than evidence. Fix the exit criteria, not the stage names.
Six things are sold under this one name.
“Sales Cloud” in a proposal can mean any of these. Knowing which you are being quoted for, and which you actually need, is worth doing before the commercial conversation rather than during it.
| What | What it is | When you need it |
|---|---|---|
| Sales Cloud | The core CRM: leads, accounts, contacts and opportunities, automatic capture of calls, emails and meetings, deal and pipeline tracking. | Always. This is the thing itself. |
| Headless Sales | Sales intelligence and deal updates inside Slack, Teams, ChatGPT and Claude, under enterprise guardrails, rather than inside the CRM interface. | When adoption is your problem, not capability. See below. |
| Sales Performance Management | Sales Programs, Sales Planning and Spiff: real-time earnings, quota tracking, incentives and the performance data behind coaching. Spiff is the commissions product. | When commissions are run in a spreadsheet nobody fully trusts. |
| Partner Cloud | Partner relationship management: a shared account view with partners, co-sold deal tracking and market development fund claims. | When a meaningful share of revenue arrives through resellers or distributors. |
| Revenue Cloud | Quoting, configure-price-quote, contracts, pricing and billing on the same platform and the same data as the CRM. | When quoting is slow, inconsistent or produces prices nobody can defend. |
| Agentforce for Sales | The agents: prospecting, pipeline management and sales actions, built and governed on the Agentforce platform. | After the hygiene work, not before. It is also the largest line on the invoice. |
The first credible answer to “nobody uses the CRM”.
The oldest complaint in this category is that CRM takes time away from selling in order to serve a system nobody sells from. Every previous answer was a better interface. This one is different: stop requiring the interface at all.
Salesforce names Slack, Microsoft Teams, ChatGPT and Claude as places where deal intelligence and updates can live, under enterprise guardrails. A seller asks about an account where they already are, and updates a deal in the same place, and the CRM stays current as a by-product.
What to check: which guardrails, exactly. An assistant that can read your pipeline in a third-party chat tool is a governance question as much as a productivity one, and it belongs in the security review rather than in the sales demo.
What Salesforce claims, and who said it.
We reproduce these because you will meet them in the sales cycle, and it is better to know whose figures they are. Every one is Salesforce’s own, published without a method.
Also on the record and more checkable: Salesforce has been named a Leader in the Gartner Magic Quadrant for CRM Sales Platforms for the twentieth consecutive year. An analyst placement is not a promise about your project, but unlike a customer-success metric it is independently produced.
The only figure that should move your decision is the one from your own pipeline. Two teams, one quarter, one changed process. Six weeks of work, and the argument is settled for good.
What we would rather you did than believe any of the aboveWhere this leads next.
The five are read in any order, but these two are the ones this page leans on most.
© 2026 Mawkish Technologies. This guide, its diagrams and illustrations are protected by copyright. You’re welcome to learn from it and to quote it with credit and a link. Reproducing it, by hand or through AI, and publishing it as your own is not allowed. Read our copyright notice
What sales leaders ask us.
Our CRM data is a mess. Will agents make it worse?
They will make it visible, which feels worse for a fortnight and is the best thing that could happen to you. An agent researching an account reads the same fields a person would; where those are stale, blank or contradictory, the agent's output shows it plainly rather than quietly working around it the way an experienced rep does.
So the honest sequence is hygiene first: required fields that are genuinely required, a close-date discipline, owners on accounts, and automatic activity capture so the record reflects what happened rather than what someone remembered to type.
What is Headless Sales?
Salesforce's name for putting deal intelligence and updates where the seller already is — it names Slack, Teams, ChatGPT and Claude — under enterprise guardrails, instead of requiring a trip into the CRM interface.
It matters more than it sounds. The oldest complaint about CRM is that it takes time from selling to serve a system nobody sells from. Moving the interaction to the channel people live in is the first credible answer we have seen to that.
Can we believe the 30% more revenue figure?
Treat it as what it is: a vendor headline with no published method, alongside 25 hours a week back for sellers. The customer figures beside it are the same — The 1916 Company closing 63% faster, Crexi saving reps five hours a day, Wonolo cutting handle time 20%, BACA doubling productivity.
These tell you the shape of what companies report, not what you would get. The number worth having is from your own pipeline: pick two teams, run one on the new process for a quarter, and compare. That is six weeks of work and it settles the argument permanently.
How long does an implementation take?
For a single sales team with clean-ish data and standard processes, a first useful release is weeks rather than months. What stretches it is rarely the software: it is agreeing one definition of a qualified lead, one set of stages, and one owner per account, across teams who have quietly run their own versions for years.
We would rather spend the first fortnight on that agreement than on configuration, because configuration built on an unresolved disagreement gets rebuilt.
Checked against Salesforce's own pages in November 2026. Product names and packaging change; check salesforce.com before you commit to anything.
Show us your pipeline report, not your requirements.
Bring last quarter's forecast and what actually closed. The gap between those two numbers tells us more about what you need than any requirements document, and it usually points at process and hygiene rather than at software.
- Your forecast against your actuals
- The hygiene work that has to come first
- Which of the six products you actually need
- A first release in weeks, not quarters
Salesforce, Agentforce, Data 360, Slack, Tableau, MuleSoft and other Salesforce product names are trademarks of Salesforce, Inc. Other names are the property of their owners. Screens shown are illustrations with sample data. Product descriptions follow Salesforce's own published material and can change, so check salesforce.com for current details.
The Darlane Check is our shorthand for looking after the details nobody sees in Salesforce Sales Cloud.