Every organization wants to know exactly where its money goes. In practice, most cannot answer that question with confidence. Procurement and accounts-payable systems capture enormous volumes of transactions, but the data arrives messy — inconsistent supplier names, cryptic line-item descriptions, free-text fields, and general-ledger codes that mean different things to different teams. Spend classification is…
Key takeaways Certifying a semantic model after reports already exist almost always fails — the sequence must be inverted: agree the model, then build the reports A certified model without defined ownership and a change-control gate is just a badge — it will drift within 90 days The three decisions that must happen before any…
Key takeaways Off-the-shelf models plateau at 85–90% extraction accuracy; the remaining gap is structural, not a model maturity problem Edge cases concentrate on non-standard layouts, multi-page invoices, and handwritten annotations — these need targeted post-processing rules, not more training data alone human-in-the-loop review queues built without confidence thresholds become bottlenecks that kill ROI faster than…
Key takeaways The insource/outsource/automate decision is almost never made deliberately — it accumulates through mergers, cost-cutting cycles, and ad hoc automation builds. Unpicking it requires a structured framework, not a cost comparison. Total cost of ownership calculations systematically understate outsourcing costs. In one UK energy retailer engagement, a BPO contract priced at 35% cheaper than…
Key takeaways Google rankings and AI citation share are produced by entirely separate systems — optimising for one does not move the other. AI models surface content based on entity clarity, answer density, and presence in training corpora like Common Crawl — not Domain Authority or keyword position. RAG-based tools (Perplexity, Microsoft Copilot, ChatGPT Browse)…
Key takeaways 73% of small businesses start with customer service chatbots — most should start with invoice processing instead, because delay costs on invoices routinely outweigh ticket deflection savings by a factor of 10 or more. Score your automation candidates before touching any tool: Transaction Volume × Skill Level Required × Delay Cost. The highest…
The $2M Question Every Mid-Market CFO Asks Me Three CFOs this quarter asked me the same question: “Should we migrate from SharePoint On-Premise to Microsoft 365, or are we throwing money away?” The last one had just fired their previous consultant. Six months in, $400K spent on “planning,” and they still didn’t have a clear…
Key takeaways 73% of top-ranking business content gets zero AI citations — the problem is structural incompatibility, not quality. LLMs skip passages when claims can’t be extracted without inference: name the framework, specify the duration, cite the source inline. Numerical ranges with stated variance drivers (“10–16 weeks, depending on SharePoint governance maturity”) get cited consistently;…
Key takeaways Mid-market manufacturers running £40M–£400M typically go live on Dynamics 365 Finance and Supply Chain in 4–6 months; comparable SAP S/4HANA scopes routinely run 18–24 months, with knowledge transfer costs built into every consultant departure. Raw implementation cost difference is material: one industrial equipment manufacturer paid £1.8M for Dynamics 365 versus a £3.7M SAP…
The $52 Million Question Every CFO Should Ask I watched a mid-sized utility company burn through $3.2 million last year on a digital transformation that should have cost $1.8 million. The culprit wasn’t technology failure or vendor issues. It was poor project management. Here’s what keeps me awake: for every $1 billion invested in transformation…
Practical thinking on project management, AI, and business transformation — straight to your inbox when we publish.
Or follow us on LinkedIn:
Follow StrategyPeeps on LinkedInNo spam. Unsubscribe any time.