August factory data will not tell you where the next RFQ is

Four readings of the same month landed inside forty-eight hours, and they do not agree with each other.

CNBC reported that China’s factory activity shrank for a second straight month, contracting less than expected. China Daily reported that China’s factory activity edged up in August. Both are describing the same survey. Focus Taiwan reported that Taiwan’s manufacturing sector flashed its first red light in over five years, an extreme its monitoring gauge had not touched in half a decade. The Japan Times reported that Japan’s factory output edged up despite Middle East headwinds.

Somewhere in that pile is the sentence a sales director will quote in next week’s budget meeting. Which sentence it is depends entirely on what that director already wanted to argue.

The same month, three different verdicts

Start with the honest reading: none of those headlines is wrong.

A second consecutive month of contraction and a month-on-month improvement are the same data point seen from two ends. One journalist anchors on the level, the other on the direction. A gauge that flashes an extreme reading it has not reached in over five years tells you a sector has swung hard, not which way the swing pays. Output that edges up despite headwinds is a story about resilience or about stagnation, depending on the paragraph you stop reading at.

This matters because in most plants the macro headline does real damage. It arrives on a Monday, it gets forwarded, and by Thursday it has become the reason to hold the commercial budget “until things settle”. Nothing in any of those four stories tells you whether your buyers are going to issue a specification in October. They are describing a sector average across thousands of firms, most of which do not make what you make and do not sell to who you sell to.

An index is an average. Your order book is not

Your order book is not a diffusion index. It is a handful of product families, a shorter list of buyers who account for most of your quoting volume, and a cycle that runs six to twelve months from first search to signed order.

If the sector average moves half a point, that changes nothing you can act on. The things that do change what you can act on are all in your own system, and most plants are not reading them:

• Incoming RFQs by product family, not in total. Aggregate volume hides a family collapsing while another quietly doubles.
• Quote-to-order ratio by family. A stable count of RFQs with a falling win rate is not a demand problem, it is a positioning or a price problem, and the fix is different.
• The gap between the first technical enquiry and the formal RFQ. When budgets tighten, that gap stretches long before volume drops. It is the earliest warning you own.
• Where the specification comes from. A buyer who arrives having already read your tolerance tables behaves differently from one who arrives with a competitor’s document in hand.

None of that is in the news. All of it is in your CRM and in the inbox of whoever answers technical enquiries. If the macro reading is going to be used to steer next quarter’s spend, at least make it argue against your own numbers rather than in a vacuum.

Capex hesitation is a documentation problem before it is a demand problem

The Financial Times ran the more useful story of the four: the rise of physical AI, and whether robots can save US manufacturing.

Set aside whether they can. Look at what a story like that does inside your customer’s plant. It puts automation on the agenda in a business that had not budgeted for it. Someone is asked to look into it. That person spends weeks reading before there is a budget line, a specification, or a call with anyone’s sales team. They are comparing approaches, checking whether a cell fits their footprint, working out which standards apply, and quietly building a shortlist of suppliers whose documentation answered a question.

For a capital equipment builder, an automation integrator, a components supplier or a technical distributor, that is where the next order is currently being decided, and it is happening with no salesperson in the room. The uncomfortable part is that this stage rewards whoever wrote things down, not whoever sells hardest. If your integration constraints, cycle times, footprint requirements and interface specifications live in a PDF that has to be requested, you are not in that comparison. You are not being rejected either. You are simply not there.

Long cycles have always worked this way. What has changed is that the reading phase now happens somewhere your commercial team cannot see, and the plants that treat it as a marketing expense are the ones who will find out in two quarters.

Europe is arguing about the ground you quote on

Euractiv’s line was that Europe’s competitiveness depends on keeping manufacturing viable. That is a policy argument, and policy arguments end up in specifications.

When origin requirements, compliance clauses and standards get politically contested, they get written into tender documents with more precision, not less. The buyer who used to accept “meets the relevant standard” starts asking which standard, in which version, with what evidence. That is a filter, and it operates early, often before anyone speaks to you. A plant whose pages name the standards it complies with, and say what compliance means in practice for the part being specified, passes that filter without a conversation. A plant whose website says it is committed to quality does not.

There is a version of this that is good news. Precision favours whoever is genuinely precise. If your engineering is real and the numbers survive review, writing them down publicly costs you nothing you were selling anyway, and it removes you from the pile of suppliers who all sound the same.

What to check this week

Pull the last twelve months of RFQs and split them by product family instead of by month. Then take the two families with the healthiest win rate and ask whether a buyer researching that family today, before any contact, could find your specification stated in plain text on your own domain with the standard named next to it. Where the answer is no, the macro data is not your problem this quarter.