I want to be honest about something. Most social media audits are dressed-up opinions. An analyst pulls some engagement numbers, eyeballs the content mix, compares a few competitors, and delivers a slide deck that confirms what the client already suspected. The recommendations are safe. The data is thin. And six months later, nothing has changed.

That's not an indictment of analysts. It's a structural problem. Cross-referencing audience demographics, competitor benchmarks, content performance, and ideal customer profiles simultaneously is genuinely difficult. Even experienced practitioners end up sequencing it: look at the audience, then at the content, then at the competitors, and hope the connections among them become clear at the end. They rarely do.

I've been using DOJO AI as a collaborator for some time now. I do not hand things over to it and walk away. I use it the way I'd use a very good analyst who reads faster than any human and doesn't lose track of things across a hundred-page dataset. This project gave me a chance to test that model under real conditions.

The Assignment

A global B2B loyalty and rewards platform asked for a LinkedIn strategy audit. They had a clearly defined 2026 ICP (ideal customer profile) and a hunch that their LinkedIn presence wasn't reaching those people. They wanted evidence, not instinct. And they wanted it to be actionable.

The company operates in a space where the sales cycle is long, the buyer is senior, and the decision to switch platforms involves procurement, product, and the C-suite. Getting LinkedIn right isn't a nice-to-have. It's a channel that either builds a pipeline over time or quietly fails while everyone assumes it's working.

What I Fed Into DOJO

This is where the process gets interesting, and where most audits fall short before they start. I assembled nine source files:

• LinkedIn audience demographic exports (four CSVs covering company data, interests and traits, job experience, and sociodemographic breakdowns)

• A screenshot of the LinkedIn page followers showing the follower count and growth trend

• The client's 2026 ICP document (a detailed Word file outlining target personas, firmographics, and buyer roles)

• A competitor analytics file (Excel, covering eight direct competitors across follower counts, posting frequency, and engagement)

• A RivalIQ export (CSV, covering 12 months of LinkedIn post-level data, including post type, topic, and engagement per post)

• A content themes and sub-themes document (the client's own proposed content pillars for 2026)

Nine files. Different formats. Different levels of structure. This is not a clean dataset. It's the kind of thing that normally requires a project brief, a prep session, and at least a week of analyst time to normalize before anyone can start drawing conclusions.

I uploaded everything into DOJO and asked it to work through the data systematically: first the audience composition, then the competitive position, then the content audit, and finally the gap between all three and the ICP.

What DOJO Found

The audience doesn't match the buyer

The demographic exports showed a highly senior follower base: 71% had 12 or more years of professional experience, and 68% were between 35 and 54. That sounds like a good thing. The problem arose when DOJO cross-referenced the job function breakdown against the 2026 ICP.

The ICP targets platform operators: HR SaaS companies, loyalty SaaS builders, and channel partners who embed the client's platform into their own products. The buying committee for those deals includes Chief Product Officers, VP of Product, Solutions Architects, and Integration Leads. None of those functions appeared with any meaningful frequency in the follower data. What was present was a heavy concentration of HR professionals and corporate roles more consistent with end-enterprise buyers, the previous ICP the company had been targeting for years.

In other words, the follower base was a fossil record of an older go-to-market strategy.

40% of the content was talking to the wrong room

DOJO analyzed 112 posts across 12 months. The content mix breakdown was the finding that I expected the client to push back on the most.

Content Mix by Category

Nearly 30% of posts were recruitment content. Another 11% were sustainability and eco-action content. Together, that's 40% of the editorial calendar serving internal audiences and ESG stakeholders, neither of which appears anywhere in the 2026 ICP. Both content types might support HR goals or brand reputation in specific contexts. But on a LinkedIn page that's supposed to be building a pipeline with platform operators, they're noise.

The content doing the most strategic work was performing the worst

This was the finding that genuinely surprised me.

Engagement by Content Category

The highest-performing content by engagement was company milestones and people-centric news. That tracks. Human content tends to outperform on LinkedIn. But the category with the lowest engagement was channel and partner program content, which is precisely the content most relevant to the ICP. Platform operators deciding whether to embed a third-party loyalty engine into their product need to understand how the partnership model works. That content exists. It just isn't connecting.

DOJO flagged this as a messaging problem, not a volume problem. The content was there; the framing was wrong. It was being written for end-enterprise buyers rather than for the platform builders who evaluate it from an API and revenue-share perspective.

The buying committee was largely absent

DOJO mapped the follower base against the full buying committee for an ICP deal: economic buyers (CPO, CHRO, VP Product), technical evaluators (Solutions Architects, Integration Leads), and champions (Product Managers, HR Tech Consultants).

Buying Committee Presence Heatmap

The heatmap told a clear story. CEO presence was strong but ambiguous: CEOs could be founders of potential platform-operator companies or enterprise buyers, and the data couldn't distinguish between them. The influencer layer (Solutions Architects and Product Managers) was moderately represented. Operations and Product Management functions were severely under-indexed despite being central to the API-first value proposition the client leads with.

No economic buyer presence in a follower base is not a content quality problem. It's a distribution and targeting problem. Organic reach alone won't close that gap.

The recommended shift

Based on all of the above, DOJO produced a recommended content rebalancing: a clear before-and-after showing where the mix needed to move.

Current vs. Target Content Mix

Recruitment drops to under 10%. Sustainability content gets folded into a broader corporate values lane rather than standing as a dedicated pillar. ICP-relevant thought leadership (the kind that speaks to platform operators and their technical and commercial concerns) expands to 35% of the mix. Proprietary data and benchmark content that performed well commercially in the audit receive a dedicated 15% allocation. And a Financial Services vertical lane gets introduced to exploit the existing high-affinity concentration of FS followers already in the base.

What This Process Actually Looks Like

DOJO didn't produce this in one pass. We worked through the audience data first, then the competitive benchmark, then the content audit, then the ICP reconciliation, then the buying committee mapping. Each stage is built on the last. I was asking questions throughout, pushing back on interpretations, and redirecting when a finding needed more depth.

The final output was a full strategy intelligence report: eight embedded visualizations, written recommendations across five strategic priorities, a pillar-by-pillar content framework, and a three-horizon activation roadmap. It was converted to a Word document for client delivery and to HTML as a standalone presentation file.

That took a fraction of the time it would have taken to do it manually. More importantly, it surfaced connections between datasets that would have been easy to miss in a sequential analysis: the correlation between follower function gaps and content category underperformance, the mismatch between the ICP's 2026 target and the legacy audience composition, the specific framing shift needed for partner program content to land differently with a technical buyer.

What I'd Tell Other Practitioners

AI didn't do the thinking. It held the threads I couldn't hold simultaneously across nine files, four data formats, and six distinct analytical questions. The strategic judgment, the framing decisions, the challenge to findings that looked wrong: that was still my job.

What changed is that I could do that job at a level of data integration that wasn't practically possible before. The audit was more rigorous, the recommendations were more specific, and the report was more coherent than anything produced by a manual process from the same inputs.

If you're running social media audits or LinkedIn strategy reviews for clients, the bottleneck has never been analysis frameworks. It's been data volume and cross-referencing at speed. That bottleneck is gone.

I'm curious: for those of you doing this kind of work, which part of the audit process still feels like it shouldn't be as slow as it is?

Andrew Jenkins is the CEO of Volterra, a social media marketing agency based in Toronto, and the author of Social Media Marketing for Business (Kogan Page or Amazon). He uses DOJO AI as a collaborator in strategic client work.