Build an Investor Update Around What Changed, Not the Original Pitch
Build an Investor Update Around What Changed, Not the Original Pitch
A pitch deck introduces a company to someone who has never seen it. An investor update does nearly the opposite job for people who have already decided: it tells them what is different since the last time they looked. Confusing the two is easy, because the deck is sitting right there, most of it still true, and none of it mentions that February came in at half the pace you projected.
The update's subject is the delta — the distance between the picture your investors last received and the one you have now. That distance includes movement in your favor and movement against it. If you could paste your update into a stranger's inbox and have it work as an introduction, you didn't write an update. You rewrote the deck.
Four things belong in the delta, and the deck cannot supply any of them: what you said last time and whether it happened; what you observed, kept separate from why you think it happened; whether this month's numbers can be laid beside last month's at all; and what the change now calls for, including a request for help when a real one exists. That request is not automatically a fundraising ask, and often isn't one.
Recover the baseline your reader actually received
Start from the update you sent, not the one you wish you had sent.
That sounds obvious. It isn't, because your memory of last month's expectations has been quietly revised by everything you've learned since. The document your investors are actually holding is the sent version: its period, its numbers, its definitions, and the sentences where you said what you expected to happen next. That document is the only fair comparison point. Reconstructing a more favorable past — "we always knew the enterprise motion would take longer" — is not context. It's a rewrite of your own record, with the reader's copy still open on their screen.
So before you write anything, find last month's update and extract three things:
The period. Which days or weeks the numbers cover, and whether that period has closed. An update that compares January to "recent months" isn't comparing anything.
The expectations you set. Quote them in the terms you originally used. If you wrote that you expected to close four pilots by the end of August, this month's update says how many you closed. Not how many you're confident about now. Not how many are in late-stage conversations.
The questions you left open. If you flagged something as unresolved last time, this update resolves it, advances it, or says plainly that it's still unresolved. Letting an open question quietly disappear is a small dishonesty that accumulates.
The temptation to re-explain the business is strongest right here, because the baseline is a document about your company and you have new readers on the thread. Resist it. Every paragraph spent re-describing what you sell is a paragraph not spent on the six shops you lost, and your existing investors already paid for that description once.
Put the observation beside the expectation
State what happened. Then state what you think it means. Keep them in different sentences.
"We had six net additions in February, against the ten to fourteen we projected" is an observation. "The channel that brought in January's new shops has gotten more expensive" is an interpretation. Both belong in a good update. The order matters because a reader can accept the first and argue with the second, but only if you've kept them apart. Blend them into one confident sentence — "growth slowed as acquisition costs rose" — and your reader can't tell which half you know.
This is where most updates soften. A missed expectation gets wrapped in a longer explanation, and the explanation arrives before the reader has registered the miss. Say the number first. Then say what you know about it, and what you don't.
That includes saying "we don't know yet." A confident diagnosis you can't support costs more than an honest gap, because it's the thing you'll have to walk back next month, in front of the same readers, who will remember it. A delay can be entirely clear while its cause is still contested. Those are two different facts and your update can hold both.
Keep recurring measures comparable
Every recurring number carries three pieces of invisible machinery: a period, a definition, and a source. Two numbers form a trend only when all three match.
Change any of them and you no longer have a trend. You have two numbers that happen to sit in adjacent boxes, and drawing a line between them is a choice you should make in front of your reader rather than behind them.
Which brings us to the most common failure in recurring updates. A headline stalls, and the fix looks like a better headline: a new measure, a more meaningful lens, a segment where the line still goes up. Sometimes the new measure genuinely is better — you learned something real about the business and the old number was measuring the wrong thing. The test isn't whether the new number is defensible. It's whether you would have made the same switch in a month when the old number looked great. If the answer is no, you're not improving your reporting. You're improving your appearance.
Aaron Harris's post on investor updates (blog.aaronkharris.com/investor-updates) makes a narrow version of this point: report consistent measures, add the context a reader needs to interpret them, and make requests visible rather than buried in a closing line. He's also careful not to overclaim, writing that he doubts a causal link between the quality of updates and the quality of companies. That's worth keeping in mind for the whole exercise. A clearer update makes your company legible. It does not make your company better, and it should never be used to make it look better than it is. Harris's essay is one practitioner's advice, not a reporting standard, and it settles nothing about what your own disclosures legally have to contain.
A worked pair: two invented updates
The company, the numbers, and the update text below are invented, including the metric definitions. Nothing here comes from a real company's records, and a real measure would need to be checked against the system that actually produces it.
Windlass sells dispatch and route-planning software to small heating, plumbing, and electrical businesses, priced per technician. It sends a monthly update to its investors. Both updates below cover a calendar month closed on the last day, so the reporting basis — the period every measure is counted over — is the same in each.
Update 1 — January
Period: January, closed.
Active shops: 84, up from 71 in December. 19 new shops started, 6 lost. An "active shop" is an account where at least one technician completed a dispatched job during the month.
What we expect: At this pace we expect to pass 100 active shops by the end of March. That assumes 10 to 14 net additions per month, which is what we've seen for the last two months.
What we're noticing: Nine of the thirteen net additions came from shops with more than one technician. We don't know yet whether that's because multi-technician shops are a better fit or because that's simply where our sales conversations happened to land. We're going to look at it in February.
What we need: Nothing this month.
Update 2 — February
Period: February, closed.
Active shops: 90, up from 84. 18 new shops started, 12 lost. Last month we said we expected 10 to 14 net additions per month on the way to 100 by the end of March. We added six net. Reaching 100 in March would now require ten net additions that month, immediately after six in February, and we don't think that's the likely case.
The losses: Losses doubled, from 6 in January to 12 in February. Nine of the twelve accounts we lost had never had a second technician complete a job. We don't yet know whether that's an onboarding problem, a change in which shops are finding us, or something else. We haven't separated those yet, and we'd rather say so than pick the explanation we like best.
A change to the measure: Starting with this update, our headline number is active shops with two or more technicians completing at least one dispatched job during the month. February on that basis is 55. January on that basis is 46.
Read that against the right number. The previous series — 71, 84, 90 — counted any shop where at least one technician completed a dispatched job during the month. The new series counts something narrower and different. The 90 and the 55 are not the same measure, and the drop from one to the other is not a decline. The new series starts at 46 and 55. The old one ends at 90 and should be treated as closed.
We made this change because the retention difference between the two groups is large enough that the old headline was telling us less than we thought. We are aware that the new headline is smaller.
What we're doing about it: In March we'll run a test on the next 40 new accounts. Half will get a required step in onboarding: invite a second technician before the account's first job is dispatched. Half will onboard as they do now. In April we'll report the two-technician rate and the retention rate for each group. Forty accounts is small enough that we'll treat the result as a signal, not an answer.
What we need: One introduction. If you know someone who has run onboarding or customer success for software sold to owner-operators — where the buyer is also the person doing the work — we'd like thirty minutes with them. The specific question is whether requiring a second seat before the trial's first job helped or just delayed signups. We're not raising, and there's no funding ask in this update.
What the second update does that the first one couldn't
The second update never restates Windlass's market, product, or founding story, and it doesn't need to. Everything in it depends on reading the first one, which is exactly right.
Notice the shape of the comparison. The expectation sits next to the observation: "we said 10 to 14, we added six net." The reader doesn't have to go looking for the original projection, and they don't have to take your word that the miss was small.
Notice that the loss figure arrives before the explanation of the loss. The nine-of-twelve detail is an observation. The three candidate causes are labeled as candidates. The update says outright that they haven't been separated — which is a more useful sentence than a tidy theory, and a much cheaper one to live with next month.
Notice that the definition change is unflattering and is stated anyway. The new headline is 55 against a previous 90. A founder optimizing for how the update looks would have kept "active shops" and mentioned the two-technician number in a smaller paragraph, or introduced it as a complement rather than a replacement. The honest version says: this is a different measure, here is why we changed it, do not compare these two numbers, and yes, the new one is smaller.
And notice the request. It names a specific kind of person, a specific question, and a thirty-minute commitment. It's tied to the problem the update just identified — the second-technician gap — not to a general wish for help. It also isn't a fundraising ask, and it says so, which saves every reader the work of figuring out whether they're being warmed up. If you have nothing useful to request, say nothing. A manufactured ask is its own kind of noise, and your investors can tell.
The test before you send
Put last month's update beside this month's draft and read them in order, as a pair.
Every sentence in the new one should be doing something the old one made possible: reporting against an expectation it set, answering a question it left open, correcting an interpretation it offered, or replacing a measure it used. If a sentence could have been written without reading last month's update, it's probably deck material wearing an update's clothes, and it's spending attention your readers already paid for.
The end of the update is where the whole thing lands: last month's expectation, this month's observation, what you now think it means, and whatever you actually need. Windlass's version reads — 10 to 14 expected, six net; the losses are concentrated among single-technician accounts and we don't know why yet; the measure is now narrower, and the narrower number is 55; in March we're testing one fix on 40 accounts; we'd like an introduction to someone who has solved the second-seat problem. The definition change makes that update less tidy than the month before it, because the headline number in it is smaller than the headline number in the last one. Keep it visible anyway. That smaller number is the one you now believe, and your investors would rather read a 55 they can trust than a 90 they can't.
Frequently asked questions
What job does an investor update do that a pitch deck does not?
A pitch deck introduces a company to someone who has never seen it. An investor update tells people who have already decided what is different since the last time they looked. Its subject is the delta, including movement in your favor and against it. If the draft could work as a stranger's introduction, it is a rewritten deck, not an update.
What should you recover from the last sent update before writing?
Find the sent version, not the one you wish you had sent. Extract the period the numbers cover and whether it has closed, the expectations you set quoted in your original terms, and the open questions you left. The sent version is the only fair comparison point. Reconstructing a more favorable past rewrites your own record.
How should a missed expectation be presented?
State what happened first, then state what you think it means in a separate sentence. A reader can accept the observation and argue with the interpretation only if the two are kept apart. It is also acceptable to say you do not know yet. A confident diagnosis you cannot support costs more than an honest gap, because you may have to walk it back.
When is changing a recurring measure legitimate?
Two numbers form a trend only when the period, definition, and source all match. Change any of them and drawing a line between the numbers is a choice you should make in front of the reader. The test is whether you would have made the same switch in a month when the old number looked great. If not, you are improving appearance, not reporting. If the change is real, say it is a different measure, explain why, and tell readers not to compare. In the invented Windlass example, the new series starts at 46 and 55, the old series ends at 90 and should be treated as closed, and the drop from 90 to 55 is not a decline.
What is the test before sending an update?
Put last month's update beside this month's draft and read them in order as a pair. Every sentence should report against an expectation the old update set, answer a question it left open, correct an interpretation it offered, or replace a measure it used. If a sentence could have been written without reading last month's update, it is probably deck material wearing an update's clothes.