Editor note
Every section this week has the same complaint under different language: the number everyone trusted stopped meaning what it used to. Scoring models, service demand forecasts, benchmark scores, and burnout surveys are all built on baselines from before something changed, and nobody rebuilt them. Worth an afternoon this week checking which of your own dashboards are still measuring last year's business. |
Sales
Your Scoring Model Only Knows One Kind of Buyer · Heinz Marketing
Heinz is right that clicks and webinar attendance aren't intent signals, but the piece stops short of saying what to score instead. If your model still weights engagement over fit and stage, strip those fields out this week and rebuild around buying-committee behavior, not web analytics.
Your Live Demo Is Showing Up Too Late · Heinz Marketing
Gated demos only make sense if the buyer can't get the answer elsewhere, and increasingly they can get it from a chatbot in seconds. Pull the gate off the questions your product page can already answer, and save the live demo for the judgment calls a bot can't make.
Marketing
Which Types of Mentions Correlate the Strongest With AI Visibility? (260M AI Prompts Analyzed) · Ahrefs Blog
260 million prompts is a real sample, but the excerpt withholds the actual ranking of mention types, so treat this as a preview rather than a playbook until the full breakdown is public. Don't rewrite your earned-media strategy off a teaser.
Every OpenAI Publisher Partnership, and How Often ChatGPT Cites Them · Ahrefs Blog
The number that actually matters here is the per-publisher citation frequency, not the list of deals. That's the data point any publisher should demand before signing a similar licensing agreement with OpenAI.
Service
AI Service Demand is Easy to Underestimate. Here’s Why · CustomerThink
The real planning error isn't misreading current contact volume, it's missing the demand that customers gave up on before ever reaching you. Once automated service tools remove that friction, expect volume to rise, not fall, and budget headcount for that, not for a shrinking queue.
Handling a Data Subject Request Without Derailing Your CX Team · CustomerThink
A data subject request landing in a shared inbox with no clear owner is a compliance gap wearing a CX costume. If your team can't say in one sentence who routes it and how many days they have to respond, write that down this week, before the test case becomes a real one.
People
The Risks of Decentralizing Your Organization · Harvard Business Review
Decentralizing sounds like delegation but HBR's framing is really about control: three specific ways business units mishandle accountability once headquarters lets go. Worth reading before your next reorg, not after.
Change debt, not change fatigue, is driving burnout, research shows · HR Dive
Burnout researchers now have a name for what operators kept missing: change debt, the unresolved backlog from past upheaval compounding into exhaustion, independent of how much change is happening right now. If engagement is falling, audit what from the last reorg never got closed out, not how many new initiatives are on the calendar.
Technology
Claude Code's "Don't Ask Again" exposes passwords in 1,122 public GitHub files · Hacker News
1,122 public GitHub files with exposed passwords because a coding agent's 'don't ask again' setting quietly waved through credential access, that's not a hypothetical risk, that's a count. Audit every repo where that setting got flipped on this week, not next sprint.
MaxxBench – "Nice DeepSWE score. But is it true?" · Hacker News
MaxxBench is a rival benchmark casting doubt on a competitor's DeepSWE score, so read the critique but also read the incentive: the publisher sells the alternative measure it's proposing. Useful as a reminder to verify any benchmark claim independently, not as proof MaxxBench is the fix.
Operations
3M closes 16 factories, plans changes at 10 more in network revamp · Supply Chain Dive
$1 billion in productivity gains by 2027 against 16 closed factories and 10 more in revamp is a footprint story first and an operations story second. Watch whether the gains come from genuine process redesign or just fewer buildings to pay for.
Tilray Brands targets $20M in cost savings after craft brewery closures · Supply Chain Dive
$20 million in savings from four facility closures and a smaller SKU count is a small number next to 3M's, but it's the same logic: demand shifted and Tilray cut supply to match it rather than waiting it out.
One idea, distilled
Four unrelated stories this week share one failure mode: a number that used to be reliable quietly stopped being reliable, and nobody rebuilt the model underneath it. Sales teams still score leads on webinar attendance, service teams still size headcount off today's ticket volume, and benchmark scores still get quoted as if gaming them were hard, even as the thing generating the underlying behavior, whether that's a buyer, a customer, or a coding agent, keeps changing shape. The optimistic reading is that this is just maintenance, go recalibrate the dashboard and move on. The harder truth is that recalibration only works if you know which assumption broke, and most teams find out from an incident report, like 1,122 exposed files, or a burnout survey, rather than from a scheduled review. What would it take for your organization to catch a stale assumption before it shows up in someone else's postmortem? |
Sources (in order of appearance): Heinz Marketing, Heinz Marketing, Ahrefs Blog, Ahrefs Blog, CustomerThink, CustomerThink, Harvard Business Review, HR Dive, Hacker News, Hacker News, Supply Chain Dive, Supply Chain Dive