North Star Audit

The short version

Three findings that sit above everything else

Each is large in scale, cheap to fix relative to its impact, and currently distorting a number someone in the business is reporting upward. They are ranked by how much damage they are doing right now.

£2.56m

Reported open pipeline — of which £188k is in an active stage with an active owner

0

Deals marked closed-lost in 2026, against 143 in 2025

0

Contacts that have ever reached Marketing Qualified Lead

87.9%

Open deals whose owner is no longer an active user

01

Critical

Nothing has been marked lost in 2026, so the pipeline has become a graveyard

Across the portal's history there are 422 closed-lost deals — 88 in 2023, 190 in 2024, 143 in 2025, and none at all in 2026. Nothing that failed this year has been closed out; it is all still sitting open, inflating every pipeline and conversion figure the portal produces.

264 of 331 open deals have a close date already in the past Exactly 1 open deal in the whole portal closes in the future 66 open deals have no close date at all

02

Critical

Nine-tenths of the open pipeline belongs to people who have left

291 of 331 open deals — worth £2,300,823 of the £2,557,250 — are owned by inactive HubSpot users. One former user, Andy Cope, holds 182 open deals worth £1,549,078 on his own. The same pattern runs through the database and the customer base: no reassignment appears to have happened when these people left.

10,286 contacts owned by inactive users 168 of 189 customer companies owned by inactive users 84% of all deals ever created

03

Critical

There is no lead qualification funnel at all — not a weak one, an absent one

20,510 contacts have entered the Lead lifecycle stage. Zero have ever entered Marketing Qualified Lead. One has ever entered Sales Qualified Lead. There is no lead scoring of any kind — HubSpot's predictive Contact Priority is unassigned across all 20,544 records and no custom score property exists. Marketing has no defensible way to say what it handed over; Sales has none to say what it received.

17,242 contacts with Lead Status "Unassigned" (83.9%) 27 contacts in any active working state 147 records on a second, parallel Leads object

Pipeline reality

£2.56m reported → £622k defensible → £188k actively owned

Two independent filters have to be applied before the pipeline number means anything: remove the deals sitting in stages that mean "not happening", then remove the deals whose owner has left.

Reported open pipelineWhat HubSpot shows today — 331 deals

£2,557,250

Excluding dead and legacy stages68 deals — removes Deffered, Lost 2026 and the Pre-JJS stages

£622,086

24.3% of the headline figure

…and excluding departed owners32 deals — James Hickson (24) and Kieran South (8)

£188,250

7.4% of the headline figure — this is the number that belongs in a forecast conversation

The stage configuration is where the damage originates

Polaris New Business 2026 holds 56% of all deals, and only three of its fifteen stages are configured as closed. Everything else counts as open pipeline — including stages whose own names say otherwise.

Stage still counted as OPEN	Pipeline	Deals	Value
Deffered (sic)	Polaris New Business	120	£770,732
Competitor client Pre JJS	Polaris New Business	66	£511,549
Deffered (sic)	Illuminator Subscription	21	£259,500
Proposal issued Pre JJS	Polaris New Business	23	£166,452
Live engagement Pre JJS	Polaris New Business	21	£132,272
Lost 2026 — not flagged as closed-lost	Polaris New Business	12	£94,659
Phantom open pipeline		263	£1,935,164

A stage literally called "Lost 2026" is not flagged closed-lost. That is a configuration bug, not a judgement call. "Deffered" is misspelled, exists twice as two separate stages with two different IDs, and between them holds £1.03m of supposedly live pipeline.

On win rate: 293 won against 422 lost reads as 41.0%, but the denominator is missing every 2026 loss. Close out the 263 dead-stage deals and the all-time win rate falls to 30.0%. Median time to close on won deals is 54 days (average 104.5).

Data hygiene

Where the records break down

Direct counts, not inference. The standout is association integrity on deals — half of all deals, and 61% of open deals, have no contact attached. A deal with no human on it cannot be sequenced, emailed, attributed to a campaign, or handed to anyone.

Check	Volume	Share
Open deals with no forecast category set	331 / 331	100%
Products with no Tax Category set	49 / 49	100%
Support tickets with no owner	100 / 100	100%
Support tickets with no Category or Priority set	100 / 100	100%
Deals with no deal type set	1,040 / 1,046	99.4%
Open deals with no next activity scheduled	328 / 331	99.1%
Contacts with no phone or mobile number	19,471 / 20,544	94.8%
Contacts not contacted at all during 2026	19,000 / 20,544	92.5%
Companies with no associated deal	4,708 / 5,278	89.2%
Products with no Product Type set	41 / 49	83.7%
Open deals with a close date already in the past	264 / 331	79.8%
Open deals never contacted	229 / 331	69.2%
Open deals with no associated contact	203 / 331	61.3%
Deals with no associated contact	531 / 1,046	50.8%
Contacts owned by an inactive user	10,286 / 20,544	50.1%
Contacts with no job title	9,417 / 20,544	45.8%
Contacts never contacted (no logged engagement, ever)	9,358 / 20,544	45.6%
Contacts with no owner	8,961 / 20,544	43.6%
Open deals HubSpot flags as "stalled"	120 / 331	36.3%
Marketing emails with "(Clone)" in the name	43 / 129	33.3%
Companies with zero associated contacts	1,710 / 5,278	32.4%
Companies with no industry set	1,109 / 5,278	21.0%
Open deals with no close date at all	66 / 331	19.9%
Contacts with no associated company	2,525 / 20,544	12.3%
Companies with no owner	519 / 5,278	9.8%
Contacts with a hard email bounce on record	1,106 / 20,544	5.4%
Deals with no associated company	36 / 1,046	3.4%
Contacts missing a first or last name	559 / 20,544	2.7%
Contacts with no email address	134 / 20,544	0.7%

Not checked: record-level duplicate detection was not performed. At 20K contacts that needs HubSpot's native Manage Duplicates tool or an export-based match, neither available from a read-only query session. The 32.4% orphan-company rate hints that overlapping company records contribute, but treat the duplicate count as unknown rather than estimated.

Lead scoring & progression

The funnel has two stages missing from the middle

Contacts go from Lead straight to Opportunity or Customer. There is no measurable handover point between Marketing and Sales, no MQL→SQL conversion rate to improve, and no way to attribute a customer to the qualification work that produced them.

Lead

20,510

99.8% of all contacts

Marketing Qualified

0

stage never used

Sales Qualified

1

one record, ever

Opportunity

349

1.7% of Leads

Customer

297

1.4% of Leads

No scoring model exists. Predictive Contact Priority is "Unassigned" for 100% of the 20,544 contacts, and no custom score property sits on the contact object. Unlike Notify's portal — which at least had a purpose-built ICP score, however poorly tuned — North Star has nothing at all to prioritise its database with. New customers created per year: 39 (2023) · 230 (2024) · 27 (2025) · 1 (2026).

The same portal, four jobs

What this means for each team

The findings above land differently depending on which chair you sit in. These are the consequences specific to each function, and what each one currently cannot see.

Lens 01

Marketing

Acquisition has collapsed, most of what does arrive is not inbound, and there is no qualification event to prove any of it worked.

400

New contacts in 2026 (8 months), against 9,582 in 2024 — a 94% fall on an annualised basis

89.3%

Of the database originates from Offline Sources; 2026 inbound runs at roughly 19 contacts a month

3.97%

Hard bounce rate in Q1 2026 — roughly double the level at which mailbox providers throttle

Q1 2026 was a deliverability event. 178 hard bounces on 4,481 sends, 1,042 contacts suppressed, open rate halved to 18.8%, unsubscribes tripled to 1.05%. It has recovered (Q2 0.33%, Q3 0.05%), but the cause — sending to an unvalidated list — persists as 1,106 contacts carrying a hard bounce.
Around 12% of marketing contacts cannot be emailed. Of the 5,149 flagged as marketing contacts, 504 have a hard bounce and 103 have opted out of all email. They are still being counted and paid for.
The marketable flag is managed by hand. 4,311 of 5,149 (83.7%) were set to marketing status manually rather than by form submission (501) or import (337).
A third of the email library is clone debris. 43 of 129 marketing emails (33.3%) carry "(Clone)" in the name. One asset — Customer Lifecycle: Email 1: Polaris Log In's (Client N02) — has been re-cloned roughly weekly since late June and now exists in versions carrying up to seventeen consecutive "(Clone)" suffixes. Cloning instead of editing means there is no single current version of any recurring email, no version history on the asset actually being sent, and no way to compare performance across sends of "the same" email — every send is a different object with its own stats.
The website forms don't write to the CRM. They are Elementor forms collected by the tracking script, not HubSpot forms — registered under raw CSS selectors, each one duplicated. Unmapped fields never reach the contact record, so the "How did you hear about us?" and "Where are you right now?" answers are captured and discarded on every submission, despite both having a matching contact property already in the portal.
Segmentation is blocked by missing data. 45.8% of contacts have no job title and 21% of companies have no industry, so persona and vertical segments cannot be built from what is there.
The programme is small. 9,664 sends across 129 emails and 9 campaigns in the twelve months to September 2026.

Conversion path

The website enquiry form is rendering unstyled

The "Need some help?" form is displaying with browser-default styling rather than any site or brand CSS. For a business taking roughly 19 inbound contacts a month, this form is close to the whole digital acquisition path — and it currently looks broken enough to cost submissions on trust alone.

No CSS is reaching the form. Square 1px default input borders, default select chevrons, a violet heading that is the browser's default link colour, and a second unrelated pink on the privacy link. Two accent colours, neither of them North Star's.
Zero left gutter — labels and fields sit flush against the container edge — and labels collide with the field above them, with no vertical rhythm.
Field widths don't align. First and Last Name are half-width; Email, Brand/Company, both selects and the textarea are full-width to a different right edge, so nothing lines up.
The consent block is malformed. The opt-in checkbox and the privacy line render as list items — the privacy line has a bullet and no checkbox. The required asterisk lands after "privacy policy here", reading as a footnote on the link rather than a required-field marker.
No visible submit button in the rendered view, and the textarea shows a default drag handle.

Worse than the styling: the fields aren't connected to anything. The site forms are not HubSpot forms at all — they are Elementor forms picked up by the tracking script's non-HubSpot form collection. You can see it in the names HubSpot has given them, which are raw CSS selectors rather than form names: #NS_Website_Contact_Form .elementor-form, #NS_Website_Footer_Form .elementor-form, #NS_Website_Partners_Hub_Form .elementor-form, and an older #gform_3. Each one is also registered twice, under two slightly different selectors, so three website forms occupy six of the portal's 33 form records.

A collected form only writes to a contact record where its fields are explicitly mapped to properties. Unmapped answers are kept inside the submission payload and never reach the record — so they are invisible to lists, workflows, scoring, segmentation and every report. Nothing maps to the company object at all.

The two questions the form asks that matter most are exactly the ones being thrown away. how_did_you_hear_about_us and location both exist as contact properties in this portal — so "How did you hear about us?" and "Where are you right now?" have somewhere to land, and aren't landing there. That is the attribution data missing from a database that is 89.3% offline-sourced, being collected and discarded on every submission.

Every enquiry since the forms were connected has lost its qualification and attribution answers

Probable cause of the styling, not confirmed: the total absence of styling — rather than bad styling — almost always means the HubSpot form's own stylesheet isn't loading: the embed set to "raw HTML form", the site CSS not targeting HubSpot's .hs-form classes, or the form script being blocked. The broken list markup in the consent block points the same way — that is HubSpot's default <ul class="inputs-list"> showing through with no CSS on it. Assessed from the rendered page, not its source, so check the embed settings first; this is likely a ten-minute fix rather than a rebuild.

The highest-value fix costs nothing: define MQL and start setting it. Without a qualification event in the data there is no conversion rate to improve and no way to demonstrate that any of the above matters. Note that the form already collects "Where are you right now?" and "How did you hear about us?" — exactly the qualification and attribution data missing everywhere else in the portal, so submissions lost to the broken form are lost twice.

Lens 02

Sales

The pipeline a rep opens in the morning is four-fifths fiction, and the working set that remains mostly belongs to people who have left.

263 / 331

Open deals sitting in stages that mean parked, deferred, lost or legacy

229

Open deals that have never had a single logged contact

32

Deals in a live stage with an active owner — the genuine working set

Knowing which stages to ignore is tribal knowledge. A rep has to discard "Deffered", "Pre JJS" and "Lost 2026" from memory — exactly the kind of knowledge that stops working when someone leaves, which is what has happened here.
Andy Cope's 182 open deals include 17 in genuinely active stages worth £278,000. Those are the ones to triage first, this week.
Activity discipline has stopped. Only 66 open deals have been contacted at any point in 2026. 328 of 331 have no next activity scheduled, 186 have no next step written, and HubSpot's stalled flag is on 120 — including 23 of the 68 genuinely-live deals.
Losses are never learned from, and never followed up. 93 of the 422 closed-lost deals (22%) carry no reason at all, and the single biggest recorded reason is "Other" at 151 (35.8%) — a taxonomy whose catch-all is its largest bucket describes nothing. Capture is getting worse, not better: 1.6% of 2024 losses had no reason against 53.1% in 2025. And once a deal is lost, nothing happens to it — there is no re-engagement of any kind.
Reps cannot work the database even if they wanted to. 94.8% of contacts have no phone number, 61.3% of open deals have no contact attached, 43.6% of contacts have no owner, and there is no score or lead status to say which of 20,544 records to open first.

Before anyone reads performance into this: per-rep win rates are not meaningful in this portal. 84% of all deals sit with departed users and 2026 has no recorded losses, so any rep comparison says more about who inherited what than about how anyone is selling.

Lens 03

Customer Success

The customer base is real. The record of it is not — the renewal book is almost entirely invisible, and the support pipeline is being used as a shared to-do list rather than a service desk.

11

Deals in the Renewals 2026 pipeline (£63,840) against roughly 175 live engagements

168 / 189

Customer companies owned by an inactive user; 3 more have no owner at all

177 / 297

Customer contacts with no deal attached — no record of what they bought

Roughly 94% of the live customer base has no renewal deal, no renewal date and no renewal owner anywhere in the system.
If a customer emails in tomorrow, HubSpot routes them to someone who has left. Only 18 customer companies have an active owner — Kieran South (17) and Victoria Kudawoo (1).
"Live" is being used as a permanent resting place. Of 158 Polaris engagements marked Live, 36 closed in 2023 and 71 in 2024 — 107 have been sitting there for two years or more. Either they are still running, in which case renewals should exist against them, or they ended and nobody moved them.
Contact cadence is the one bright spot. All 189 customer companies have been contacted at some point and 140 during 2026; the remaining 49 form a small, workable at-risk list.

Critical

The Support pipeline has no owners and no triage data at all

There are 100 tickets in the portal — 91 created during 2026, 77 still open. Every single one of them is missing the three fields that make a ticket a ticket.

Field	Tickets missing it	Consequence
Owner	100 / 100	Nobody is accountable for a single support ticket
Category	100 / 100	Nine well-designed options exist on the property and none is ever used, so recurring problems can't be spotted
Priority	100 / 100	Nothing can be triaged as urgent

And the tickets aren't really tickets — they're a shared to-do list. All 100 have a source of Email, and the subjects are raw reply threads and internal queries rather than support requests: "RE: Applied Nutrition Renewal IDs Query", "RE: RTD Spirits Asda NAD Quote – Served Drinks", "The Curators – Asda range", "RE: Novomins on Polaris – #USU3012". Several are commercial — a quote, a renewal query, an upsell — rather than service issues at all. A shared inbox is piping every inbound email into the Support pipeline, and the team is working the resulting list as tasks, leaving the structured fields empty because those fields aren't what the list is being used for.

No first-response or resolution-time metric can be produced, so no service-level commitment can be evidenced.
No volume trend by category can be produced, so recurring product or onboarding problems stay invisible.
Support requests and commercial opportunities queue in the same undifferentiated list — and the commercial ones are not reaching a deal pipeline.
A second ticket pipeline, "Reengagement" (Monitoring → Reached Out → Moved To Deal → Lost), exists and is completely empty — scaffolding for a sales motion built on the ticket object, never used.

Not visible in this audit: the ticket records above are readable and their counts exact, but conversation content and response-time analytics are blocked by the portal's AI access settings ("Customer conversation data" is disabled) — so first-response time, resolution time and CSAT/NPS could not be measured. There are also no subscription records and no invoices.

Lens 04

Finance

HubSpot cannot currently produce a forecast, and the pipeline number that would reach a board pack is roughly four times too big.

331 / 331

Open deals with an empty forecast category — no weighting, no commit discipline

£0

Closed-won in both July and August 2026, against a 2025 average of ~£52k a month

99.4%

Deals with no deal type set — new business, renewal and upsell cannot be separated

Any time-phased forecast places £2.5m of pipeline in the past, because only one open deal has a close date in the future.
Net revenue retention, ARR and churn are uncomputable from this data. There are no subscription records, no invoices, 5 quotes and 19 line items in total — every deal is a single lump-sum amount with no product or term structure behind it.
Currency is clean. 1,042 GBP deals, 4 unset, no mixed-currency aggregation risk, and only 4 deals with no amount.
The marketing contact tier carries roughly 600 unmailable contacts inside the 5,149 being counted — a small but recurring cost with no return.

High

The product library would block any move to quoting or invoicing

49 products exist — Polaris Subscription and Set-up tiers, created in May 2023 and untouched since May 2024. The fields that make them usable for billing are empty.

Field	Products missing it	Consequence
Tax Category	49 / 49 · 100%	The field HubSpot uses to apply tax automatically to line items. Every quote or invoice raised from these products carries no tax treatment and must be corrected by hand
Product Type	41 / 49 · 83.7%	Subscription products can't be separated from one-off set-up fees
Term	22 / 49 · 44.9%	Recurring value can't be annualised — the split needed to compute ARR

Why this matters now rather than later: for a GBP business with UK VAT obligations, Tax Category is the one product field that cannot be left blank — if the portal is ever connected to HubSpot Payments, Stripe or an accounting sync, tax will either fail to calculate or default incorrectly. Only 19 line items exist against 1,046 deals, so the library is barely used today. That makes it cheap to fix now and expensive to fix later: the moment quoting or invoicing goes live on top of it, every gap becomes a finance correction.

The number to quote: £188,250 across 32 deals. That is the open pipeline that is both in a live stage and assigned to someone who still works here.

Trend

Every input metric fell at the same time

Closed-won revenue, contact creation, deal creation and new customers all turned down together. That uniformity is the strongest signal in the audit — and the thing most worth confirming with the team.

Closed-won revenue by close date

2026 covers eight months to 31 August

£314,847

2023

£602,945

2024

£629,559

2025

£100,167

2026 YTD

2026 by month: Jan £0 · Feb £36,367 · Mar £15,000 · Apr £0 · May £12,000 · Jun £36,800 · Jul £0 · Aug £0.

New contacts created

Deal creation followed the same curve: 303 · 402 · 288 · 53

6,623

2023

9,582

2024

3,939

2025

400

2026 YTD

Of the 400 created in 2026, 249 came from Offline Sources and 151 from any digital channel — roughly 19 inbound contacts a month.

The fixes, broken down

Changes by function

Every improvement is placed in one of three lanes: what you could do yourselves (mostly hygiene and settings — set, reassign, remove), somewhere in between (needs a data clean-up plus a decision on the target design), and where RevOps Hub can help (needs new process, scoring logic, or build work implemented). Click into any function below for the full breakdown.

28

Improvements identified across the portal

11

Do-it-yourself fixes — settings and hygiene

8

Hybrid fixes (data + decision)

9

Where RevOps Hub can lead

Each fix is tagged with the type of benefit it delivers — click any item to expand it and see the reasoning:

Cost Saving

Reduces spend, licence tier, or write-off risk

Efficiency Gains

Saves time, reduces admin, improves reporting accuracy

Revenue Generating

Grows or protects pipeline, conversion, or retained revenue

General Cleanup

Standardising properties, fixing or tidying existing data

Marketing — full breakdown

What you could do yourselves

3 items

▸Fix the website enquiry formRevenue GeneratingEfficiency Gains

The "Need some help?" form renders with browser-default styling — no gutter, misaligned field widths, a consent block breaking into bullets, and no visible submit button. For a business taking roughly 19 inbound contacts a month, this form is close to the whole digital acquisition path, and it currently looks broken enough to cost submissions on trust alone.

Check whether the embed is set to raw HTML and whether the site stylesheet targets HubSpot's .hs-form classes. Likely a ten-minute fix rather than a rebuild.

Assessed from the rendered page, not the source — confirm the cause before rebuilding anything

▸Adopt an email naming convention and stop cloningGeneral CleanupEfficiency Gains

43 of 129 marketing emails (33.3%) carry "(Clone)" in the name. One asset has been re-cloned roughly weekly since late June and now exists in versions carrying seventeen consecutive "(Clone)" suffixes.

The cost isn't cosmetic: there is no single current version of any recurring email, no version history on the asset actually being sent, and no way to compare performance across sends of "the same" email — every send is a different object with its own stats. Archive or rename the clones and move the recurring send to a template.

▸Suppress the hard-bounced contactsCost SavingGeneral Cleanup

1,106 contacts carry a hard bounce. Around 600 of them sit inside the 5,149 records being counted and paid for as marketing contacts, alongside 103 who have opted out of all email.

Q1 2026 saw a 3.97% hard bounce rate — roughly double the level at which mailbox providers start throttling — with 1,042 suppressions and the open rate halving to 18.8%. Suppressing and unmarking these protects sender reputation and recovers tier spend.

Somewhere in between

2 items

▸Rebuild the website forms so submitted data reaches the recordRevenue GeneratingGeneral Cleanup

Replace the collected Elementor forms with native HubSpot forms — or, if the site must keep its own forms, map every field explicitly to a contact or company property. Either way the fields have to be connected, because today the answers are captured and discarded.

Deduplicate while doing it: three website forms currently exist as six records under near-identical CSS selectors, which also splits submission reporting across duplicate entries.

Which answer maps to which property is your call — particularly whether "Brand / Company name" should create or match a company record rather than sitting as text on the contact. We build and test the mapping once that's agreed, and confirm submissions land end to end before it goes live.

Do this before the top-of-funnel engine — intent scoring is worth less if inbound answers never reach the record

▸Close the contact and company data gapsEfficiency GainsRevenue Generating

45.8% of contacts have no job title, 21% of companies have no industry, and 94.8% have no phone number. Persona and vertical segmentation can't be built from what's there today, and no outbound motion that assumes calling is viable.

The clean-up is mechanical; the decision isn't. Which fields matter, which enrichment source, and what gets enriched at the point of entry versus retrospectively — those are yours. We scope, source and run it once agreed. Duplicate records were never quantified and should be measured in the same pass.

Where RevOps Hub can help

3 items

▸Define MQL stages, responsibilities, handoffs and SLAsRevenue GeneratingEfficiency Gains

The single action the whole audit points at. Zero contacts have ever entered Marketing Qualified Lead; one has ever entered Sales Qualified Lead. Contacts go from Lead straight to Opportunity or Customer, so there is no measurable handover point between Marketing and Sales at all.

Workshop to agree what qualifies a lead and who owns it at each step, then build the lifecycle automation, routing rules, SLA timers and the reporting that proves the handover happened. Until a qualification event exists in the data, no conversion rate exists to measure and no scoring model has anything to predict.

Affects Sales as much as Marketing — both teams in the room

▸Build a top-of-funnel engine that surfaces warm leads from intent dataRevenue Generating

Intent and web-visit signals scored against ICP fit, surfaced to sales as a prioritised daily list rather than a database of 20,544 undifferentiated records.

This is what replaces the absent lead scoring and gives the team something to work in the morning. It matters more here than in most portals because inbound is running at roughly 19 contacts a month and 89.3% of the database originated from offline sources — there is no volume to wait for, so the warm signals in the existing base have to be found rather than generated.

▸Build the lead scoring modelRevenue GeneratingEfficiency Gains

There is no scoring of any kind today: HubSpot's predictive Contact Priority is "Unassigned" across all 20,544 contacts, and no custom score property exists on the contact object.

Fit plus engagement scoring, feeding the MQL threshold agreed above — so it sequences after the MQL definition, not before it. 17,242 contacts (83.9%) also sit at Lead Status "Unassigned", leaving 27 records in any active working state.

Sales — full breakdown

What you could do yourselves

5 items

▸Flag the dead stages as closedGeneral CleanupEfficiency Gains

In Polaris New Business only three of fifteen stages are configured as closed. Everything else counts as open pipeline — including a stage literally called "Lost 2026", which is a configuration bug rather than a judgement call.

Set "Lost 2026", both "Deffered" stages, "Competitor client Pre JJS" and the Pre-JJS legacy stages to a closed state. Around thirty minutes in pipeline settings, and it removes £1,935,164 of phantom pipeline across 263 deals while restoring every win-rate and velocity metric in the portal at once.

Highest return per minute spent of anything in this audit

▸Merge the duplicate "Deffered" stages and fix the spellingGeneral Cleanup

"Deffered" is misspelled and exists twice, as two separate stages with two different IDs — one in Polaris New Business (120 deals, £770,732) and one in Illuminator (21 deals, £259,500). Between them they hold £1.03m of supposedly live pipeline.

Merge to one correctly-spelled stage as part of the same settings pass as the action above.

▸Put a real close date on every open dealEfficiency Gains

66 open deals have no close date and 264 have one already in the past. Exactly one open deal in the entire portal closes in the future. Any time-phased forecast therefore places £2.5m of pipeline in the past.

Set them, then keep them current at every stage change. This is a habit as much as a task — the discipline is what makes it stick.

▸Standardise on one closed-lost reason field and fix the taxonomyGeneral CleanupEfficiency Gains

Two competing fields exist. The freetext closed_lost_reason is empty on 420 of 422 closed-lost deals — effectively dead. The live one is the dropdown, and it has only four options, of which the catch-all is the largest: Other 151 (35.8%), No Budget 146 (34.6%), no reason at all 93 (22.0%), Nielsen Client 26, Circana Client 6.

Capture has also degraded badly — deals closed with no reason recorded: 15.9% in 2023, 1.6% in 2024, then 53.1% in 2025, and no losses recorded at all in 2026.

Retire the freetext field, expand the taxonomy so "Other" stops absorbing a third of all losses (timing, no decision / went silent, price, lost to internal build, feature gap), and make the field required on entry to a closed-lost stage.

Do this before clearing the deal backlog — closing 263 deals without reasons destroys the data the nurture below segments on

▸Clear the backlog of aged open dealsGeneral CleanupEfficiency Gains

Qualify, disqualify or close out. 146 of the 331 open deals were created in 2023 or 2024 and are still open; 229 have never had a single logged contact; HubSpot's stalled flag is on 120.

Most importantly, no deal has been marked closed-lost at any point in 2026 — against 143 in 2025. Nothing that failed this year has been closed out, which is the root of the inflated pipeline.

Somewhere in between

3 items

▸Reassign all records owned by inactive usersRevenue GeneratingEfficiency Gains

291 of 331 open deals — worth £2,300,823 of the £2,557,250 — are owned by inactive users. Andy Cope alone holds 182 open deals worth £1,549,078. The same pattern runs through 10,286 contacts and 168 of the 189 customer companies.

The bulk tooling and matching rules are ours; who inherits what is yours. Start with the 17 live-stage deals worth £278,000 under Andy Cope, then the customer companies — that is where the revenue risk sits.

▸Repair deal-to-contact associationsGeneral CleanupEfficiency Gains

531 deals (50.8%) have no contact attached, and 203 of the 331 open deals (61.3%) have none. A deal with no human on it cannot be sequenced, emailed, attributed to a campaign, or handed to anyone — it breaks marketing attribution and sales follow-up at the same time.

We build and run the matching logic; your team resolves the ambiguous cases the match can't decide on its own.

▸Introduce next-step and activity disciplineRevenue Generating

328 of 331 open deals (99.1%) have no next activity scheduled, 186 have no next step written, and only 66 have been contacted at any point during 2026.

We build the tasks, nudges and manager dashboards. Your managers run the cadence — that half genuinely can't be outsourced, and the tooling won't hold without it.

Where RevOps Hub can help

3 items

▸Review and rebuild the pipelines with relevant stagesEfficiency GainsRevenue Generating

Seven pipelines and 36 stages, with the main one carrying legacy "Pre JJS" stages, a duplicated misspelling, and a "Lost 2026" stage that isn't closed. One pipeline ("Upsell Opps") holds a single deal; another ("Renewals 2026") holds 11 against roughly 175 live engagements.

Rationalise down to the pipelines you actually sell through, with correct closed-won and closed-lost flags, sensible stage probabilities, and written entry and exit criteria per stage. This goes well beyond the thirty-minute settings fix in the first lane — that one stops the bleeding, this one fixes the structure.

▸Build a proper closed-lost process with reason-based nurture and follow-upRevenue GeneratingEfficiency Gains

There is no closed-lost process at all today. A deal is lost, the record stops moving, and nothing else happens to it — ever. Between the 422 closed-lost deals and the 263 sitting in dead-but-open stages, roughly 685 records represent lost or parked revenue with no follow-up mechanism against any of them.

Design a defined closed-lost stage with a mandatory reason, then automated re-engagement sequenced by what that reason actually implies:

No Budget — 146 deals (34.6%). The largest genuine reason. A timed re-approach against budget cycles, with light-touch nurture in between rather than a single reminder task.

Competitor losses — Nielsen (26) and Circana (6), plus the 66 deals parked in "Competitor client Pre JJS" worth £511,549, of which 19 already carry a named competitor. Reactivation timed to the likely incumbent renewal window rather than a fixed interval. This is the highest-value reactivation segment in the portal, and it is currently sitting in a stage that isn't even flagged as closed.

Deferred — 141 deals across the two "Deffered" stages, £1.03m. These aren't losses, they're deferrals with no revisit date and no owner. Each one needs a date and an owner or it should be closed properly.

Timing and went-silent losses. Currently invisible inside "Other" — a shorter cadence, since these convert fastest once the taxonomy can identify them.

Re-engaged accounts should feed back into the lead scoring and intent engine, so a closed-lost account showing renewed activity resurfaces as a warm lead rather than staying dormant. Add closed-lost reporting alongside it — loss reason by pipeline, owner and quarter — which is impossible today and is exactly the input the pipeline rebuild needs.

▸Data hygiene automation and quality alertingEfficiency GainsGeneral Cleanup

Validation rules, required fields, stale-record alerts, and an offboarding checklist that reassigns records automatically when someone leaves.

Without this the portal drifts straight back: the inactive-owner concentration exists precisely because nothing caught it when people left. Fixing the data once without fixing the process means auditing it again in eighteen months.

Customer Success — full breakdown

What you could do yourselves

2 items

▸Make Owner, Category and Priority required on ticketsEfficiency GainsGeneral Cleanup

All 100 tickets in the portal are missing all three fields. Nobody is accountable for a single support ticket, nothing can be triaged as urgent, and no volume trend by category can be produced.

The nine category options already exist on the property — Access & Logins, Renewals, Data Query, Custom Report Query, Platform Training, New Data Quote / Upsell, Onboarding & Setup, Billing & Admin, Other — and have never once been used. Turn the fields on as required, then backfill the 77 open tickets.

▸Retire the empty "Reengagement" ticket pipelineGeneral Cleanup

A second ticket pipeline with stages Monitoring → Reached Out → Moved To Deal → Lost — scaffolding for a sales motion built on the ticket object, holding zero records.

Delete it or commit to it. Leaving it in place adds noise to every ticket view and every pipeline dropdown.

Somewhere in between

2 items

▸Build renewal coverage for the live customer baseRevenue Generating

Roughly 175 engagements sit in won "Live" stages, against 11 deals worth £63,840 in the Renewals 2026 pipeline. About 94% of the live customer base has no renewal deal, no renewal date and no renewal owner anywhere in the system.

"Live" is also being used as a permanent resting place — 107 of the 158 Polaris engagements marked Live closed in 2023 or 2024. We design the renewal pipeline, dates and early-warning triggers; you confirm which of those engagements are genuinely still running.

▸Agree the customer ownership and routing modelRevenue GeneratingEfficiency Gains

168 of 189 customer companies are owned by an inactive user and 3 have no owner at all. Only 18 have an active owner — Kieran South (17) and Victoria Kudawoo (1). If a customer emails in tomorrow, HubSpot routes them to someone who has left.

Decide who owns which customer and where an inbound email should land; we build the routing behind it. 177 of 297 customer contacts also have no deal attached, so there is no record of what they bought.

Where RevOps Hub can help

1 item

▸Rebuild the service desk as a service deskEfficiency GainsRevenue Generating

The tickets aren't really tickets — they're a shared to-do list. All 100 have a source of Email and subjects that read as reply threads and internal queries: "RE: Applied Nutrition Renewal IDs Query", "The Curators – Asda range", "RE: Novomins on Polaris – #USU3012". A shared inbox is piping every inbound email into the Support pipeline and the team is working the list as tasks.

Several of those are commercial — a quote, a renewal query, an upsell — rather than service issues, and none of them reach a deal pipeline. Separate genuine support from commercial enquiries, with categories, priorities, SLAs, routing, and a defined path from ticket to deal. The shared inbox stays; the to-do list becomes a service function.

Response-time analytics stay blocked until "Customer conversation data" is enabled in the portal's AI settings

Finance — full breakdown

What you could do yourselves

1 item

▸Set a Tax Category on all 49 productsCost SavingGeneral Cleanup

Tax Category is unset on 49 of 49 products. This is the field HubSpot uses to apply tax automatically to line items created from a product, so every quote or invoice raised from these carries no tax treatment and has to be corrected by hand.

For a GBP business with UK VAT obligations this is the one product field that cannot be left blank — and if the portal is ever connected to HubSpot Payments, Stripe or an accounting sync, tax will either fail to calculate or default incorrectly. A bulk edit, half a day at most.

Somewhere in between

1 item

▸Backfill deal type and forecast categoryEfficiency GainsRevenue Generating

99.4% of deals have no deal type set and 100% have no forecast category — so new business, renewal and upsell cannot be separated, and HubSpot cannot produce a forecast at all. There is no weighting, no commit or best-case discipline, and no probability signal beyond default stage percentages.

We define the values and the automation that maintains them going forward; your team validates the historical classification, which is the part that needs someone who was there.

Where RevOps Hub can help

2 items

▸Product, quoting and billing readinessCost SavingEfficiency Gains

Beyond setting tax categories: Product Type is unset on 41 of 49 products and Term on 22, so subscription products can't be reliably separated from one-off set-up fees — exactly the split needed to compute ARR.

There are no subscriptions, no invoices, 5 quotes and 19 line items against 1,046 deals, so every deal is a single lump-sum amount with no product or term structure behind it. The library is barely used today, which makes this cheap now and expensive later: the moment quoting or invoicing goes live on top of it, every gap becomes a finance correction.

▸Build the reporting and dashboard layerEfficiency Gains

Pipeline and forecast for Sales and Finance, funnel and attribution for Marketing, renewal risk and service volumes for Customer Success.

Deliberately sequenced last. Reports built on the data as it stands today would mislead more confidently than no reports at all — the £2.56m pipeline figure is the clearest example of a number that is technically correct and completely wrong.

★ marks the five priority actions. Three are where to start, one from each lane: the stage settings fix costs nothing and repairs the most reporting, the reassignment protects revenue that currently has nobody attached to it, and the MQL definition unblocks every metric that follows. The other two — the pipeline rebuild and the intent engine — are the larger builds those first three make possible, and shouldn't start before them.