How did one novelist go from ninety queries to a two book deal in eighteen months?

Slowly, and then in a nine day rush. The writer we will call Marguerite sent ninety two queries across eighteen months, got two offers of representation in the same week, and signed a two book deal eleven months after that. Nothing about the path was lucky in the way people mean when they say lucky.

What follows is a composite, reconstructed from the kind of submission log any working writer keeps. The names are withheld and the details are typical rather than unique. The value is not in the outcome. It is in the six decision points where she changed something specific, and in the fact that she could see what to change because she had written it all down.

Round one: the manuscript that was not finished

She queried twenty two agents with a literary suspense novel at 118,000 words. The book had been through four passes and two beta readers, both friends. She thought it was done. It was not done, and the tell was in the first chapter: two pages of weather and a childhood memory before anyone spoke.

She queried in batches of five, which was the one thing she got right. Five out, wait, five more. Over eleven weeks she collected twenty form rejections, one no response, and one request for the first fifty pages that came back as a pass with a single line: "the voice did not carry me far enough."

A query request rate of one in twenty two is not catastrophic. It is also not a signal you can act on, because a single request tells you almost nothing. What told her something was the shape of the responses.

Keep reading: Is the small press market still worth submitting to now that distribution has consolidated?

Reading the pattern in twenty form rejections

Form rejections feel like noise. They are not entirely. Sorted by what the agent had actually seen, they separate into groups that mean different things.

Nineteen stops at the query stage pointed at the letter, not the book. She had pitched the novel as literary fiction with a mystery element, cited two comparison titles more than a decade old, and buried the actual hook in the third paragraph. The hook was that the narrator was the one who had reported her sister missing, and had lied when she did it.

That sentence went into paragraph one of every later query.

The revise and resubmit that reset the opening fifty pages

The one agent who read the partial sent an unsolicited paragraph of notes, which is rarer than writers assume and worth more than an offer from someone who does not read closely. The notes said the book started in the wrong place and that the first real scene was on page thirty one.

She cut the first thirty pages. Not trimmed. Cut. Then she rewrote forward from the new opening, which changed what the reader knew and when, and that unpicked about a third of the middle. The revision took five months and brought the book to 97,000 words.

This is the part people skip. Twenty rejections is not evidence that the book is wrong; it is evidence that the package is not working. Distinguishing between the two requires you to know exactly what each agent saw. If you cannot answer "did that person read pages, and which pages," you cannot diagnose anything.

Round two: a tighter list of thirty agents chosen by client list

Round one had been assembled from a public database, sorted by whether the agent was open. Round two was built differently. She went to the acknowledgments pages of eleven novels published in the previous three years that shared readers with hers, pulled the agent names, and checked each one's recent sales in the trade announcements.

Her criteria, in order:

1. Has sold something in this space in the last twenty four months.
2. Is at an agency whose name appears on deals with houses she would want.
3. Has an open list, or at least no explicit closed notice.
4. Wants materials she can actually supply in the requested format.

Thirty names, queried in batches of six. Response times ranged from four days to never. Eight full requests came in over ten weeks, which is a request rate of roughly one in four, against one in twenty two the first time. Same author, better book, far better targeting. You cannot separate those three variables, and she did not pretend to.

Keep reading: What should I check in a small press contract before I sign away my rights?

The small press offer that created a deadline

Two months into round two, an independent press she had submitted to directly the previous autumn, before the rewrite, came back wanting the book. She sent the revised version. They still wanted it, and offered a modest advance in the low four figures with a standard reversion clause and no option on the second book.

She did not accept, and she did not decline. She asked for two weeks to consider, which is normal and was granted. What the offer actually did was convert an open ended wait into a dated event, and a dated event is the only thing in this process that reliably makes other people move.

Nudging the outstanding fulls with an offer in hand

She had eight fulls out. The email went to all eight the same afternoon, four sentences long:

I wanted to let you know I have received an offer of publication from a press and have been asked to respond by the 24th. You have the full manuscript, which I sent on the date noted below. If you are still considering it, I would be glad to hear from you before then. Either way, thank you for the read.

No pressure language, no exclamation marks, and crucially the date each agent had received the manuscript, because agents genuinely do not remember. Of the eight, three withdrew immediately, two never replied, one asked for a week and passed, and two read fast and called.

The mechanics here are worth stating plainly. A nudge with an offer attached is legitimate and expected. A nudge without one is a status query, which you send once, after the agent's stated response time has passed, and not before.

See how DraftAndQuery handles this for independent authorship and publishing

Choosing between two offers of representation

Both calls were good. Deciding took four days and came down to five questions she asked both, in the same words.

• What would you want me to change before submission, and how long would that take?
• Which editors and houses would be on your first round, and why those?
• What happens if the first round does not sell the book?
• How do you handle subsidiary rights: film, audio, translation, and who does that work?
• What are your termination terms, and who owns the relationship with editors afterward?

Agent A wanted a light edit and had a nine editor first round. Agent B wanted a structural revision to the last third, had six editors and a clear reason for each, and answered the "if it does not sell" question with an actual plan rather than optimism. She signed with B, declined the press with thanks, and the revised book sold in a two book deal to a mid sized house the following spring.

What the submission log made visible that memory did not

Three things she could not have reconstructed from memory:

First, that nineteen of twenty rejections in round one stopped at the query. That diagnosis is only available if you record what each agent asked for and what you actually sent.

Second, which version each person had. Six of the round two agents had seen the 97,000 word cut. Two, from a stale portion of the list, had somehow been sent the older file. She caught it because the log recorded the file version against the recipient, and she was able to write to both and correct it rather than wonder for a year why they went quiet.

Third, the dates in that nudge email. Agents respond to specifics. "I sent this on March 3" is a different message from "a while ago."

The unglamorous part is the part that works

The story reads like a rewrite story. It is really a records story wearing a rewrite's coat. Every decision she made well was made from written evidence, and every wobble came from a gap in that evidence.

That is the job DraftAndQuery does: which version of the manuscript is with which agent or press, when it went, when their stated response window closes, and which follow ups are due this week. Keep that, and the next ninety queries will teach you something instead of just costing you time.