Putting Evidence Behind a Sales Pipeline in HubSpot
A digital PR agency ran client health in Google Sheets, scored leads on a system nobody could explain, and had a HubSpot deal pipeline with probabilities set by feel. This is an ongoing engagement, so it is written as what has been done so far: pipeline stages with win probabilities calculated from 202 historical deals, intent-based lead scoring with ownership rules, and the client scorecard moved into HubSpot where it can be reported on.
The Challenge: A CRM that recorded outcomes but not the road to them
HubSpot was in place and paid for. Deals were logged. What was missing was everything between a first conversation and a signed contract. More than half of won deals had been moved straight to Won with no stage history, so there was no way to know where deals actually stalled. No lost deal had a reason recorded. Stage probabilities had been typed in on instinct, which meant the weighted pipeline number the managing director looked at each week was a guess dressed as a forecast.
Marketing had a lead score, but it rewarded page views and email opens rather than intent, so a contact who had opened six newsletters outranked one who had booked a call. Nobody could say when a lead was ready to hand to sales, or whose it was once handed.
Client services ran a scorecard and a contract audit in Google Sheets: renewal dates, health indicators, upsell targets, the conversations account managers had actually had. Useful, and entirely dependent on one person's spreadsheet.
Underneath all of that, the HubSpot subscription carried seats and marketing contacts nobody was using.
What Has Been Done So Far: Evidence first, then the build
Pipeline stages with probabilities from the data. Every one of the agency's 202 historical deals was analysed for how long it sat in each stage, how many activities it took, and whether it closed. The pipeline was restructured into seven stages from Marketing Handover to Closed Won, each with a win probability calculated from the deals that had tracking history rather than from instinct, and each expected to sharpen as new deals move through. The analysis also surfaced things worth knowing: won deals average 26 sales activities against 12 for disqualified ones, marketing-sourced deals are worth roughly three times partner-sourced ones, and retainers are worth four and a half times a project per deal and make up 82% of won revenue. Closed-lost reasons are now mandatory.
Lead scoring based on intent. The old score was replaced with a tiered model: high-intent actions such as booking a meeting or being met at an event score in the tens, engagement scores in single figures, and fit alone cannot push a contact over the line. At 100 a contact becomes marketing-qualified and is assigned to the sales lead, with a notification only if there is no live deal already, so sales are not told about people they are already talking to. Under 100 the contact is flagged as marketing-owned. Below-100 ownership is handled with a custom property because HubSpot will not let non-sales seats own contacts formally.
Reporting that answers the weekly questions. A cold, warm and hot dashboard for marketing; a touches-per-stage report so the team can see how much effort each stage really takes; the weighted pipeline now resting on calculated probabilities.
Client services out of the spreadsheet. Custom properties for contract dates, client health, meeting counts and upsell tracking, with the scorecard history imported, so the account managers' view lives in the same system as the sales view and can be reported on the same way.
Subscription cleanup. Unused seats and marketing contacts identified for removal, worth roughly £4,000 to £6,000 a year, which is more than the first phase of the work cost.
The engagement continues on a monthly retainer. The next pieces are the Monday.com to HubSpot sync so delivery data reaches the client record, and nurture workflows for the leads the scoring model now identifies.
This is an ongoing engagement, so the results below are what is measurable now rather than a final tally.
202 historical deals analysed to produce stage probabilities and five data quality findings the agency did not know it had, including zero recorded loss reasons and six open deals over 200 days old.
£4,000 to £6,000 a year in licence savings identified from unused seats and marketing contacts, covering the cost of the first phase.
82% of won revenue from retainers, a figure that was in the CRM all along and had never been surfaced, and which now shapes what the pipeline prioritises.
The weighted pipeline the managing director reads each week is now a calculation rather than an opinion, marketing knows when a lead is ready and who owns it, and the client scorecard no longer walks out of the building with the person who maintained it.
The client is anonymised. The deal, revenue and savings figures come from the agency's own HubSpot data; the licence saving is an identified figure rather than a realised one at the time of writing.