AI Readiness Audit — Northbank Communications
How to use this report: Start with the 5-minute version, then go to whichever question matters most to you today. Part 4 is the plan — it opens with your numbered action list, and every other part refers back to those numbers.
Everything else in this report expands on what’s below. Read this, do the first move, come back for the rest when you’re ready.
Cover approved tools, client-data handling rules and disclosure obligations on a single page. This is the unlock for everything else — the prompt library, the reporting automation and any client-facing AI output all depend on it existing first.
The full set of quick wins recovers an estimated 60.3 hrs/week across the agency — that is your 86.2 automatable hrs/week after a workflow-maturity reduction; the working is in Section 04.
Doing nothing costs roughly £13,568/month in unrecovered capacity.
Do the first move above this week — that’s it, everything else can wait until it’s done. Your Action Plan spreadsheet is a separate Google Sheet, linked in the email that delivered this report. When you want the full plan, go straight to Part 4: Roadmap & Recovery.
A high-level view of where your agency stands today — overall score, benchmark position, and the key findings that matter most.
You scored 51/100 — the “Established” tier.
Strongest: Tools & Stack (69). Biggest gap: Measurement & ROI (28).
The headline view: your overall score, the shape of your readiness across eight dimensions, and the findings that matter most.
Northbank Communications scored 51/100 on the Nexolve AI Readiness Index, placing the agency in the "Established" tier and slightly ahead of the 47 average for agencies of similar size. The stack is your strength: modern, API-capable and largely already paid for. The constraint is everything wrapped around it — no written AI policy, no measurement, and 60 recoverable hours a week still being done by hand. Close the governance gap first, because it unblocks everything else in this plan.
The information you provided, reflected back. Spot an error? Reply to your delivery email (or write to hello@nexolve.co.uk) within 7 days and we will regenerate your report on the corrected answers, free. Appendix E plays back every questionnaire answer.
Your score on each of the eight scored dimensions — what the data shows, where the gaps are, and what to do about each one.
Your biggest gap is Measurement & ROI (28) — this part explains why, and what to do about it first.
Just want the plan? Skip ahead to Part 4 — the roadmap stands on its own.
Your score across all eight dimensions, with the weight each carries. The dashed marker shows the score you can realistically reach by completing the roadmap.
Each target’s “via R…” note names the numbered actions that move that score. Every R-number is stated in full in your action list (Section 08).
Leadership is engaged and AI is discussed at board level, but intent has not yet been converted into a written plan with owners and milestones. That gap is what separates your 58 from the 80 this dimension could reach.
This is your lowest score of the eight and your most urgent. ChatGPT and Claude are already running across client accounts without a written policy, which means every team member is making their own judgement call about what client data can go in.
Your data hygiene fundamentals are sound, but client data lives across five disconnected platforms. That fragmentation is the single biggest blocker to the highest-value automation opportunities in your roadmap.
This is your strongest dimension and a genuine asset. The stack is modern and API-capable — the opportunity is not buying more tools but switching on and connecting what you already pay for.
Processes are understood but almost entirely manual. With 169 hours a week across the tracked task areas and 86.2 of those automatable, this dimension carries the largest single financial opportunity in the report.
Appetite is high and roughly 45% of the team already uses AI weekly, but usage is self-taught and inconsistent. Without shared standards, output quality varies by individual rather than by intent.
This is your lowest score and it constrains everything else. Without evidence of what AI is returning, phase 3 and phase 4 investment will be difficult to justify internally, however sound the plan is.
Experimentation happens, but informally and without a way to capture what was learned. Good ideas surface and then evaporate when the person who tried them moves on to the next deadline.
Financial modelling, compliance posture, and the commercial levers available to your agency right now.
Waiting has a price: roughly £13,568/month in unrecovered capacity.
What automating the identified tasks is worth to you in hours and pounds, modelled at a blended rate (the average hourly cost of the people doing the work, senior and junior together) of £50–£100/hr (reported).
As a monthly retainer agency, time savings can be re-deployed into margin improvement or capacity expansion.
Each row’s Annual figure is that task’s share of the £163k headline — the workflow-maturity multiplier and the 25% conservative reduction are already applied, so multiplying one row’s hours by the rate and 48 weeks will give a higher, un-reduced number.
| Task | Current | Automatable | % of recoverable | Annual |
|---|---|---|---|---|
|
Client reporting & analytics
Automated data pulls + AI-drafted commentary
|
34h | 23.8h | 28% | £45k/yr |
|
Content creation & first drafts
Prompt library + AI first-draft workflow
|
30h | 13.5h | 16% | £25k/yr |
|
Meetings & follow-ups
AI transcription and action-point extraction
|
26h | 3.9h | 5% | £7k/yr |
|
Client communications
Templated AI drafting with human review
|
22h | 8.8h | 10% | £17k/yr |
|
Media monitoring
Webhook-fed AI digest replacing manual scanning
|
20h | 15h | 17% | £28k/yr |
|
Admin & operations
Workflow automation across PM and finance tools
|
16h | 8h | 9% | £15k/yr |
|
Data collection & cleaning
Scheduled API integrations into one data layer
|
12h | 7.8h | 9% | £15k/yr |
|
Competitor research
AI-assisted landscape briefs on a weekly cadence
|
9h | 5.4h | 6% | £10k/yr |
| Total
86.2h raw → 60.3h after the workflow-maturity multiplier — the formula above
|
169h | 86.2h | — | £163k/yr |
The £163k headline is a run-rate — what a full year looks like once the changes are embedded. These scenarios move the two things the model is least certain about: which end of the published estimate range holds, and how quickly the Section 11 sequence lands. First-12-months figures assume value ramps linearly from zero to the run-rate; the ramp assumptions are listed with the rest of the model in Appendix B.
| Scenario | Run-rate value | Full run-rate by | First 12 months |
|---|---|---|---|
|
Conservative
The Section 11 sequence slips by half (run-rate at month 18) and recovery runs at the low end of the published estimate range.
|
£130k/yr | Month 18 | £43k |
|
Expected (the published figure)
The Section 11 sequence lands on schedule (run-rate at month 12) at the published central estimate.
|
£163k/yr | Month 12 | £81k |
|
Upside
Quick wins land in the first quarter, the sequence completes early (month 9), and the high end of the estimate range holds.
|
£195k/yr | Month 9 | £122k |
A scenario, not a forecast. The £163k direct saving above is what the audit supports on its own; the figures below assume 60% of the freed hours are re-sold as billable work.
Net benefit is stated after the £16,560/yr (£1,380/mo) of plan upgrades recommended in Appendix A, on top of the core-stack tooling budget. Those upgrades are not a contradiction of the “activate what you already pay for” finding: they are a phase-4 dependency, bought for the API access the integration work needs, not to unlock features you already own.
Payback above is how long the recovered hours take to cover that tooling spend — licences only. It is not the implementation timeline: the work itself runs across the phases in Section 11, Your sequence, and is your team’s time, which this model does not price.
Blended rate: £50–£100/hr (reported) — modelled at £75/hr, the midpoint of the band you reported (at £50/hr the direct saving is £109k/yr; at £100/hr, £217k/yr) · Recoverable: 60.3 hrs/week · All figures are conservative estimates, which is why the headline saving is shown as a range. Every assumption behind these numbers — recovery rates, working weeks, the 25% reduction — is listed in full in Appendix B. The direct saving and billable uplift scale with the hours you reported — if your team's real week on these tasks is a fifth higher than your estimate, so are they; the tooling costs stay fixed.
↪ You reported 169 hrs/week across the eight tracked task areas.
↪ Your Workflow Automation score of 39/100 sets the ×0.7 maturity multiplier.
Your self-reported governance and data privacy posture. Not a legal assessment — a map of where action may be needed under GDPR, the EU AI Act, and ISO/IEC 27001.
3 of 8 governance checks confirmed. 5 items require attention before client-facing AI deployments at scale. The Data governance depth items further down are graded on the 0–5 scale and sit outside this count.
The L number is the maturity level you answered on our 0–5 scale, from L0 (nothing in place) to L5 (running and reviewed). L4 and above is counted as in place.
Important: This compliance check is based on your self-reported answers. It is not legal advice. We recommend consulting a data protection officer or legal counsel to validate your GDPR and EU AI Act obligations.
This report shows what you gain by acting. This section shows what you lose by waiting. Every month without action carries the same cost again.
The competitive window is an illustration of pace, not market intelligence: it is how far ahead a similar agency starting this report’s roadmap today would be by the time you begin — on the same phase timings as Section 11, Your sequence.
These costs recur every month you wait. Over 6 months that is £81,408 in unrecovered capacity. Over 12 months, £162,816 — the monthly figure repeated, not compounded. Meanwhile, agencies that act now will have established AI workflows, trained teams, and measurable efficiency gains — widening the gap with every passing month. The question isn't whether to adopt AI, but how much the delay will cost you.
↪ Based on your audit: Derived from 60.3 recoverable hrs/week at your reported blended rate.
How your readiness compares to agencies of similar size, revenue and service mix, drawn from the Nexolve benchmark dataset for 31–50 person agencies.
How each of your scores compares to the average for agencies of similar size.
| Dimension | Your Score | Peer Average | Gap | Position |
|---|---|---|---|---|
| Strategy & Vision | 58 | 52 | +6 | Ahead of peers |
| Governance & Risk | 32 | 35 | -3 | In line |
| Data & Privacy | 48 | 42 | +6 | Ahead of peers |
| Tools & Stack | 69 | 56 | +13 | Ahead of peers |
| Workflow Automation | 39 | 46 | -7 | Behind peers |
| People & Skills | 52 | 50 | +2 | In line |
| Measurement & ROI | 28 | 54 | -26 | Well behind |
| Innovation & Experimentation | 44 | 40 | +4 | In line |
↪ Based on your audit: Benchmarked against agencies in your size band (31-50 people) from the Nexolve sector peer set, April 2026. Each row is that band's average for the dimension; the cohort (the group of comparable agencies your scores are measured against) average uses the same dimension weights as your own score.
Peer averages are modelled from Nexolve's benchmark dataset for your size band; positions are directional, not percentiles.
26% of agencies share your Established tier — 60% sit in the tiers below yours, 14% in the tiers above.
Agencies scoring 70+ share four capabilities you can realistically build within this roadmap.
Top competitors deploy AI for all first-draft content, cutting production time by around 60%.
AI-powered sentiment and coverage tracking across 10,000+ sources, delivered as a morning brief.
AI identifies the best journalist matches for a story, measurably improving open rates.
Dashboards refresh automatically, reducing reporting turnaround to near-zero.
Your Tools & Stack score of 69 is well ahead of the peer average of 53 — you already own most of the capability required. The work is activation, not acquisition.
Where Northbank Communications sits: at a readiness score of 51, you are in line with the average of 49 for agencies your size (Nexolve benchmark data, April 2026). The gap between agencies acting on AI strategically and those still experimenting is widening.
Exactly what to do, in what order. It opens with your numbered action list — every recommendation in this report, stated once — then what to do first, the quick wins, and the phased sequence.
Completing this list recovers an estimated 60.3 hrs/week in total.
Start with: Publish a one-page AI usage policy.
Everything this report recommends, in one list. There are 16 distinct actions — each one is stated in full once, and referred to by its number everywhere else.
| # | Action | Effort | Impact | Dimension | Phase |
|---|---|---|---|---|---|
| R1 |
Publish and socialise a one-page AI usage policy across the agency
Quick wins · Priority matrix · Your sequence
|
Low
Half a day
|
High | Governance & Risk | 1 |
| R2 |
Automate daily media monitoring digests with AI summarisation
Quick wins · Priority matrix · Your sequence
|
Medium
1 day
|
High | Workflow Automation | 2 |
| R3 |
Automate first-draft client reports using AI templates
Priority matrix · Your sequence
|
Medium | High | Workflow Automation | 2 |
| R4 |
Create reusable client AI consent and data-handling templates
Quick wins · Priority matrix
|
Low
2–3 hours
|
High | Data & Privacy | 1 |
| R5 |
Build an AI ROI tracking dashboard
Priority matrix · Your sequence
|
Medium | High | Measurement & ROI | 3 |
| R6 |
Build a shared prompt library and AI writing style guide
Quick wins · Priority matrix
|
Low
Half a day
|
Medium | People & Skills | 1 |
| R7 |
Launch monthly AI training drop-ins for all staff
Priority matrix
|
Low | Medium | People & Skills | 2 |
| R8 |
Build a RAG knowledge base for credentials and case studies
Priority matrix · Your sequence
|
High | High | Tools & Stack | 4 |
| R9 |
Integrate client analytics into a unified reporting data layer
Priority matrix · Your sequence
|
High | Medium | Data & Privacy | 4 |
| R10 |
Quantify manual workflow cost for the investment case
Quick wins
|
Low
2–3 hours
|
Medium | Measurement & ROI | — |
| R11 |
Switch on the AI features you already pay for
Quick wins · Your sequence
|
Low
2 hours
|
High | Tools & Stack | 2 |
| R12 |
Complete vendor DPA and data-residency audit
Your sequence
|
Low | High | Governance & Risk | 1 |
| R13 |
Appoint an AI champion per function
Your sequence
|
Low | Medium | Strategy & Vision | 1 |
| R14 |
Develop a two-year AI capability roadmap
Your sequence
|
Medium | Medium | Strategy & Vision | 3 |
| R15 |
Consolidate Cision and Meltwater onto one platform
Your sequence
|
Medium | Medium | Tools & Stack | 3 |
| R16 |
Package AI capability as a client offering
Your sequence
|
High | Medium | Innovation & Experimentation | 4 |
How to use these numbers. They are stable references, not a ranking of effort or a schedule — the order follows priority, and the Phase column tells you when each one starts. Use them to assign owners: “who has R3?” means the same thing in this report, in your Action Plan spreadsheet, and in a conversation with your team. Impact and effort are bands, not scores: actions ranked on the priority matrix are banded from their computed impact, and actions that sit only in your sequence are banded from the priority the sequence gives them. Effort throughout this report is hands-on working time — approval, legal review and sign-off cycles add elapsed days on top, so a “half a day” action may take a fortnight to fully land.
Your AI and automation opportunities, mapped by impact versus effort. Each dot is one action from your action list — dot 3 is R3.
How to read this: violet dots are your recommended starting points; dark dots are your next phase. Effort, impact and phase for every one of these are in your action list. A quick win can also be a prerequisite — some low-effort actions here (a policy, a named owner) unblock the rest, which is why the phase order in Your sequence is the dependency order, not this chart’s.
Actions you can take this month with minimal investment. Each one is scoped to land inside a month and requires no new tools or infrastructure. Lines marked ↪ show the audit answer each recommendation is based on.
Cover approved tools, client-data handling rules and disclosure obligations on a single page. This is the unlock for everything else — the prompt library, the reporting automation and any client-facing AI output all depend on it existing first.
↪ Based on your audit: Your Governance & Risk score is 32/100.
Your media team spends 20 hours a week scanning feeds. Route alert output into a single AI-assembled morning brief covering coverage hits, sentiment shifts and stories to watch. One day of configuration replaces a daily manual grind.
↪ Based on your audit: You reported ~20 hrs/week on Media monitoring.
Curate 15–20 tested prompts for press releases, pitch angles and boilerplate rewrites, paired with a one-page style guide. Store it somewhere the whole team can reach by the end of the day.
↪ Based on your audit: Your People & Skills score of 52/100 leaves clear headroom.
A short disclosure paragraph and a data-processing addendum, drafted once and inserted into new contracts immediately. This protects both you and your clients, and is a prerequisite before any AI-generated output goes to a client audience.
↪ Based on your audit: You answered "No" to clients having agreed to AI use on their accounts.
You have 86.2 automatable hours a week sitting in manual workflows — 60.3 of them recoverable after the workflow-maturity adjustment. A one-page cost-of-inaction summary gives leadership a concrete financial frame for approving phase 2 and phase 3 spend.
↪ Based on your audit: Your Measurement & ROI score is 28/100.
HubSpot, Slack and Google Workspace all ship AI capability included in your current plans — the content assistant, Workflow Builder and Gemini in Docs. Nobody has enabled them. This is pure upside with no procurement and no new spend.
↪ Based on your audit: Your Tools & Stack score of 69/100 is your strongest dimension.
The automation opportunities identified across this report recover an estimated 60.3 hours per week — worth £4,523/week at your blended rate of £75/hour (from the £50–£100/hr band you reported). That is the direct-saving basis behind Section 04’s headline — the re-deployment scenario there is modelled on top of it, not in addition. The quick wins above are the first step toward it, and need no new tools or budget.
Sequenced by dependency, not by deadline. Move at your own pace — but follow the order. Each phase unlocks the next.
Timing is indicative. Phase 1 can start immediately with no new spend — its budget band above is your team’s existing hours priced for planning, not money out the door. Phase 2 timing depends on your approval and procurement cycles. Phases 3 and 4 expand as the early phases prove their return. Your personalised Action Plan — a separate Google Sheet, linked in your delivery email — contains week-by-week tasks for Phases 1 and 2.
Department scorecards, tool ecosystem analysis, the full scoring methodology, and the template pack.
Reference material — dip in as needed. Nothing here is required reading before you act.
AI readiness isn’t uniform across your agency. Each department has different strengths, gaps, and starting points.
Ratings come from what you told us about each department in the questionnaire, not from an external audit. Strong means this department can lead the rollout of what already works; Developing means it needs a named owner and support before scaling; Gap means start with the basics here first.
| Department | Opportunity | Impact | Effort |
|---|---|---|---|
| PR & Media Relations | AI media monitoring and sentiment analysis | High | Low |
| Content & Editorial | AI-assisted first-draft generation with a shared prompt library | High | Low |
| Client Services | Automated coverage reporting with AI-drafted commentary | High | Medium |
| Media Planning & Buying | Automated performance data consolidation across platforms | Medium | Medium |
| New Business | RAG-powered credentials and case-study retrieval for pitches | High | High |
Based on everything in this report, here are your three highest-priority actions. Do these first.
One page, leadership sign-off, shared with the whole team. Everything else depends on it.
The media monitoring digest is the highest-value, lowest-risk starting point at 20 hours a week.
Measurement & ROI is your weakest dimension at 28/100. Without evidence, phase 3 investment never gets approved.
This report is built to be handed over — Northbank Communications does not need Nexolve to deliver it.
A complete audit of every tool in your stack — where there’s overlap, where there’s waste, and where API and integration capabilities unlock automations.
The tools you told us your team uses, judged only against our verified tools database — where we cannot verify a tool, we say so rather than guess.
| Tool | Category | Capabilities | API | Webhooks | MCP |
|---|---|---|---|---|---|
| HubSpot | CRM & Marketing | 5 | ● | ● | ● |
| Cision | Media Intelligence | 4 | ● | ● | ○ |
| Meltwater | Media Intelligence | 4 | ● | ● | ○ |
| Asana | Project Management | 3 | ● | ● | ○ |
| Monday.com | Project Management | 4 | ● | ● | ○ |
| Canva | Design | 3 | ● | ○ | ○ |
| Slack | Communication | 3 | ● | ● | ● |
| Google Workspace | Productivity | 4 | ● | ● | ● |
| Google Analytics | Analytics & BI | 3 | ● | ● | ○ |
| SEMrush | SEO & Paid | 3 | ● | ● | ○ |
● Available ○ Not available
Capabilities — how many distinct capability areas (reporting, workflow automation, data collection…) the tool covers in our tools database; the overlap alerts below compare tools on these. API — the tool can send and receive data automatically, so nothing is copied across by hand. Webhooks — it can tell your other tools the moment something happens. MCP — an AI assistant can operate it directly.
Tools with 3+ shared capabilities — consolidation could reduce costs and complexity.
Pricing is indicative, based on published vendor tiers at time of analysis — confirm current pricing before acting.
Features you're paying for but likely not using.
Higher tiers that unlock API/automation for the tools you already use.
Why these sit alongside “activate what you already pay for”: together they add £1,380/mo (£16,560/yr), already netted off the benefit figure in the money section. And they buy API access the integration work in the later phases depends on — not features you already own. Budget them against phase 3–4, not phase 1.
Pricing is indicative, based on published vendor tiers at time of analysis — confirm current pricing before acting.
Capabilities no tool in your current stack covers.
Top integration opportunities already supported by your tools.
Full transparency on how every number in this report is calculated. Every score and financial figure is produced by a deterministic formula from your questionnaire answers — the same answers always produce the same numbers.
Methodology informed by NIST AI RMF, ISO/IEC 42001, and GDPR/EU AI Act obligations, adapted for agency environments.
| Dimension | Weight | Why This Weight |
|---|---|---|
| Strategy & Vision | 20% | Without leadership buy-in and strategic direction, all other efforts fragment. |
| Governance & Risk | 10% | The safety net. A low score dramatically increases the risk of incidents. |
| Data & Privacy | 10% | Data discipline determines what AI can safely touch. |
| Tools & Stack | 20% | Capability lives in the stack the team actually uses every day. |
| Workflow Automation | 10% | You can't automate chaos. Process maturity predicts deployment speed. |
| People & Skills | 15% | AI is only as effective as the people using it. Skills and culture determine adoption. |
| Measurement & ROI | 10% | What isn't measured doesn't compound. ROI evidence sustains investment. |
| Innovation & Experimentation | 5% | Structured experimentation keeps the agency ahead of the curve. |
The dashed marker on your scorecard is computed, not aspirational — the uplift is larger for weaker dimensions, because early gains come fastest where least is in place.
Each action carries an hour allowance by its effort band — Low 4–8h, Medium 12–24h, High 24–60h of hands-on team time. A phase’s budget band is the sum of its actions’ allowances priced at your blended rate; the spread within a band is the allowance width, and a wide phase range simply means it holds high-effort actions. These are your team’s existing hours priced for planning — not new spend.
The best/likely/worst scenario table in the money section pairs the ends of that published range with a time-to-run-rate assumption: conservative reaches full run-rate at month 18, expected reaches full run-rate at month 12, upside reaches full run-rate at month 9. First-12-months figures assume value ramps linearly from zero to the run-rate over that period, which realises 33% / 50% / 63% of the run-rate respectively. The ramp lengths are Nexolve modelling assumptions anchored on the 12-month roadmap, not measured delivery data.
The recovery rates below are Nexolve modelling assumptions — the share of the hours you reported for each task area that we model as recoverable. They are not measured from your agency, and they are the same for every report built on your agency profile; the hours they are applied to are yours.
| Task area | Recovery rate |
|---|---|
| Reporting & analytics | 70% |
| Admin & operations | 50% |
| Client communications | 40% |
| Content creation | 45% |
| Data collection | 65% |
| Media monitoring | 75% |
| Competitor research | 60% |
| Meetings | 15% |
Peer positions ("ahead of peers", "in line with peers") compare your dimension scores against Nexolve benchmark averages for agencies of your size band. We deliberately avoid percentile claims — positions are stated only where our benchmark data supports them.
Agency-size benchmark averages: Nexolve benchmark dataset (April 2026), updated quarterly. This report quotes no industry survey statistics — every figure in it is either something you told us or something calculated from it, by the formulas above.
These templates turn the recommendations in this report into immediate action. Each one is designed to be customised for your agency in under an hour.
Your one-page AI policy. Customise the bracketed sections, get leadership sign-off, and share with the team. Review quarterly.
1. Purpose — This policy sets out how [Agency Name] uses artificial intelligence tools in our work.
2. Approved Tools — ✓ Approved: [e.g. ChatGPT Team, Claude Pro, Canva AI] ✗ Not Approved: [e.g. Free-tier tools with data training]
3. Data Rules — OK to input: publicly available info, your own drafts, anonymised data. Never input: client financial data, PII, unpublished strategies, credentials.
4. Client Disclosure — We are transparent with our clients about AI use. Position: [Choose approach].
5. Quality Review — All AI-generated work reviewed by qualified team member before delivery.
6. IP & Copyright — Our position on AI-generated creative, image generation and text: [state position]. Client industries with stricter requirements: [list].
7. Incident Response — Report to [AI Champion] immediately. Incidents logged and used to update policy.
Use before adopting any new AI tool. Score each criterion 1-5. Tools scoring below 25/40 should not be adopted without exceptional justification.
Criteria: Data policy (GDPR, training) · Integration (API, n8n/Zapier) · Replaces vs Adds · Team adoption curve · Cost vs value · Scalability · Output quality (test with 3 real tasks) · Vendor stability
Role-specific starter prompts for Account Management, Content & Copywriting, SEO & Performance, PR & Communications, and Operations.
Document any process before automating it. Covers: trigger, steps, tools, people, time, frequency, pain points, and automation potential.
Drop into proposals, contracts, or standalone communication. Covers: what you use AI for, what you never use AI for, how you protect data, your quality standard.
Track progress monthly: team members using AI, hours saved, automations live, incidents, roadmap progress, what's working / what's not, next month priorities.
Detailed guidance for each item in your roadmap — what tools to use, what to look out for, and the exact first step to take.
Establish the minimum governance layer covering approved tools, client-data rules and disclosure obligations. Expect this to take a low effort slot in phase 1, owned by the Governance & Risk lead.
Paste this into ChatGPT or Claude, answer its questions, and review the draft it gives you — you are editing, not starting from a blank page:
Block 90 minutes this week to scope "publish the ai usage policy" with the Governance & Risk lead and agree the definition of done.
Review DPAs for every AI vendor against GDPR obligations and log findings in a reusable checklist. Expect this to take a low effort slot in phase 1, owned by the Governance & Risk lead.
Paste this into ChatGPT or Claude, answer its questions, and review the draft it gives you — you are editing, not starting from a blank page:
Block 90 minutes this week to scope "complete vendor dpa and data-residency audit" with the Governance & Risk lead and agree the definition of done.
Designate one champion per major function and form a steering group that meets monthly. Expect this to take a low effort slot in phase 1, owned by the Strategy & Vision lead.
Block 90 minutes this week to scope "appoint an ai champion per function" with the Strategy & Vision lead and agree the definition of done.
Feed alert output into an AI call that produces a structured morning brief, eliminating manual feed-scanning. Expect this to take a medium effort slot in phase 2, owned by the Workflow Automation lead.
This is a technical build (API connections or a dashboard), not a document. Three realistic routes: someone in-house who is comfortable with APIs; a non-developer working with an AI coding assistant such as Claude or ChatGPT, which can produce and explain a build like this step by step (allow focused days, not hours, and keep client data out of the session); or a freelance integrator — budget roughly £2–5k and 1–2 weeks. “No new tooling cost” on this action refers to licences; the build time is real.
Block 90 minutes this week to scope "automate the daily media monitoring digest" with the Workflow Automation lead and agree the definition of done.
Connect monitoring and analytics data into a template, then use AI to draft commentary for human review. Expect this to take a medium effort slot in phase 2, owned by the Workflow Automation lead.
This is a technical build (API connections or a dashboard), not a document. Three realistic routes: someone in-house who is comfortable with APIs; a non-developer working with an AI coding assistant such as Claude or ChatGPT, which can produce and explain a build like this step by step (allow focused days, not hours, and keep client data out of the session); or a freelance integrator — budget roughly £2–5k and 1–2 weeks. “No new tooling cost” on this action refers to licences; the build time is real.
Block 90 minutes this week to scope "automate monthly client coverage reporting" with the Workflow Automation lead and agree the definition of done.
Activate the content assistant, predictive lead scoring and email optimisation you already pay for. Expect this to take a low effort slot in phase 2, owned by the Tools & Stack lead.
Block 90 minutes this week to scope "audit and enable hubspot built-in ai features" with the Tools & Stack lead and agree the definition of done.
A structured plan covering skill development, tool investment and quarterly milestones, presented at each board cycle. Expect this to take a medium effort slot in phase 3, owned by the Strategy & Vision lead.
Block 90 minutes this week to scope "develop a two-year ai capability roadmap" with the Strategy & Vision lead and agree the definition of done.
Track hours saved, output quality and cost per AI-assisted workflow. Your weakest dimension needs visible evidence. Expect this to take a medium effort slot in phase 3, owned by the Measurement & ROI lead.
This is a technical build (API connections or a dashboard), not a document. Three realistic routes: someone in-house who is comfortable with APIs; a non-developer working with an AI coding assistant such as Claude or ChatGPT, which can produce and explain a build like this step by step (allow focused days, not hours, and keep client data out of the session); or a freelance integrator — budget roughly £2–5k and 1–2 weeks. “No new tooling cost” on this action refers to licences; the build time is real.
Block 90 minutes this week to scope "build an ai roi tracking dashboard" with the Measurement & ROI lead and agree the definition of done.
The two share three core capabilities. Evaluate which better serves PR and crisis comms, then redirect the saved licence cost into AI tooling. Expect this to take a medium effort slot in phase 3, owned by the Tools & Stack lead.
Block 90 minutes this week to scope "consolidate cision and meltwater onto one platform" with the Tools & Stack lead and agree the definition of done.
Index credentials decks, case studies and brand guidelines so any team member can query institutional knowledge in seconds. Expect this to take a high effort slot in phase 4, owned by the Tools & Stack lead.
This is a technical build (API connections or a dashboard), not a document. Three realistic routes: someone in-house who is comfortable with APIs; a non-developer working with an AI coding assistant such as Claude or ChatGPT, which can produce and explain a build like this step by step (allow focused days, not hours, and keep client data out of the session); or a freelance integrator — budget roughly £2–5k and 1–2 weeks. “No new tooling cost” on this action refers to licences; the build time is real.
Block 90 minutes this week to scope "build a rag knowledge base for credentials and case studies" with the Tools & Stack lead and agree the definition of done.
Consolidate analytics, monitoring and CRM data into one queryable layer — the infrastructure that makes client-facing AI reporting commercially viable. Expect this to take a high effort slot in phase 4, owned by the Data & Privacy lead.
This is a technical build (API connections or a dashboard), not a document. Three realistic routes: someone in-house who is comfortable with APIs; a non-developer working with an AI coding assistant such as Claude or ChatGPT, which can produce and explain a build like this step by step (allow focused days, not hours, and keep client data out of the session); or a freelance integrator — budget roughly £2–5k and 1–2 weeks. “No new tooling cost” on this action refers to licences; the build time is real.
Block 90 minutes this week to scope "integrate client analytics into a unified data layer" with the Data & Privacy lead and agree the definition of done.
Convert proven internal workflows into a billable service line with its own credentials and pricing. Expect this to take a high effort slot in phase 4, owned by the Innovation & Experimentation lead.
Block 90 minutes this week to scope "package ai capability as a client offering" with the Innovation & Experimentation lead and agree the definition of done.
Every answer you gave, played back. Each score, hour and £ figure in this report is computed from these — so check them. Spot an error? Reply to your delivery email (or write to hello@nexolve.co.uk) within 7 days and we will regenerate your report on the corrected answers, free.
Agency-wide totals across the team, as you reported them. Every recovery calculation in section 04 is a fraction of these hours — if one is wrong, tell us and the numbers will be recomputed.
The plan in this report fails or succeeds on adoption, not tooling. This kit is how to roll it out: the sequence of communication, a champions model, and the handful of metrics that tell you whether it is actually sticking.
Keyed to the same phases as Section 11, Your sequence. The mechanics here are deliberately the same for every agency — rollout does not vary by score — but every R-number and target is this report’s own.
One person per function who is visibly enthusiastic and actually does the work — not the most senior person available. Their job, about 2–3 hours a week:
1. Hold a weekly 30-minute open session where anyone can bring a task and leave with a working prompt or workflow. 2. Collect the wins and the failures — both feed the quarterly review. 3. Be the named route for “can I use this tool for this?” questions, so the policy has a human front door.
Track these monthly from Phase 1. They answer “is this sticking?” — the financial model in Section 04 answers “was it worth it?”.
| Metric | How to measure it | Healthy looks like |
|---|---|---|
| Weekly active use of approved tools | Count of team members who used an approved AI tool for real work this week, from champions’ sessions or tool seat data | Over half the team by month 3, rising |
| Hours recovered on the priced task areas | Lightweight time tracking on the task areas this report prices, against the 60.3 hrs/week the model targets | A visible downward trend by month 3 — hitting the full figure is a month-12 test, not a month-3 one |
| Automations live | Workflows running unattended in production, against the 6 quick wins in Section 10 | 3+ live by month 3 |
| Incidents and near-misses reported | Entries in the incident log — wrong output caught in review, data pasted where it should not have been | A small, non-zero number. Zero after month 3 usually means under-reporting, not perfection |
| Readiness re-score | Retake the audit at month 12 and compare dimension by dimension | Tracking toward the 78/100 (Integrated) 12-month target on your scorecard |
Northbank Communications scored 51/100 — placing you at the Established level. That's not a judgement — it's a starting point. Every agency in our database began somewhere, and the ones pulling ahead aren't the biggest or best-funded. They're the ones that started.
This report has given you a clear map: where you are, where the gaps are, and exactly where to start. Your single highest-impact first move is Publish a one-page AI usage policy — it's low effort and can begin this week.
Questions about this report?
hello@nexolve.co.uk
Prepared by Nexolve · © 2026 Nexolve. All rights reserved.